diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,20124 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.09195402298850575,
+  "eval_steps": 200,
+  "global_step": 2000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 4.597701149425287e-05,
+      "grad_norm": 29.670974731445312,
+      "learning_rate": 0.0,
+      "loss": 0.9156931638717651,
+      "num_input_tokens_seen": 226,
+      "step": 1,
+      "train_runtime": 7.4708,
+      "train_tokens_per_second": 30.251
+    },
+    {
+      "epoch": 9.195402298850575e-05,
+      "grad_norm": 24.900712966918945,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.9386992454528809,
+      "num_input_tokens_seen": 540,
+      "step": 2,
+      "train_runtime": 9.438,
+      "train_tokens_per_second": 57.216
+    },
+    {
+      "epoch": 0.00013793103448275863,
+      "grad_norm": 35.45130157470703,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.9111277461051941,
+      "num_input_tokens_seen": 784,
+      "step": 3,
+      "train_runtime": 11.3773,
+      "train_tokens_per_second": 68.909
+    },
+    {
+      "epoch": 0.0001839080459770115,
+      "grad_norm": 20.591541290283203,
+      "learning_rate": 1.2e-05,
+      "loss": 0.7331118583679199,
+      "num_input_tokens_seen": 1050,
+      "step": 4,
+      "train_runtime": 13.2971,
+      "train_tokens_per_second": 78.964
+    },
+    {
+      "epoch": 0.00022988505747126436,
+      "grad_norm": 15.996316909790039,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.6205352544784546,
+      "num_input_tokens_seen": 1360,
+      "step": 5,
+      "train_runtime": 15.2314,
+      "train_tokens_per_second": 89.289
+    },
+    {
+      "epoch": 0.00027586206896551725,
+      "grad_norm": 15.900446891784668,
+      "learning_rate": 2e-05,
+      "loss": 0.5837218165397644,
+      "num_input_tokens_seen": 1646,
+      "step": 6,
+      "train_runtime": 17.1596,
+      "train_tokens_per_second": 95.923
+    },
+    {
+      "epoch": 0.0003218390804597701,
+      "grad_norm": 10.117097854614258,
+      "learning_rate": 1.998997493734336e-05,
+      "loss": 0.471426784992218,
+      "num_input_tokens_seen": 1994,
+      "step": 7,
+      "train_runtime": 19.0879,
+      "train_tokens_per_second": 104.464
+    },
+    {
+      "epoch": 0.000367816091954023,
+      "grad_norm": 13.1370210647583,
+      "learning_rate": 1.997994987468672e-05,
+      "loss": 0.5827452540397644,
+      "num_input_tokens_seen": 2340,
+      "step": 8,
+      "train_runtime": 21.0482,
+      "train_tokens_per_second": 111.173
+    },
+    {
+      "epoch": 0.00041379310344827585,
+      "grad_norm": 12.377201080322266,
+      "learning_rate": 1.9969924812030075e-05,
+      "loss": 0.6739121675491333,
+      "num_input_tokens_seen": 2694,
+      "step": 9,
+      "train_runtime": 22.9885,
+      "train_tokens_per_second": 117.189
+    },
+    {
+      "epoch": 0.0004597701149425287,
+      "grad_norm": 12.438722610473633,
+      "learning_rate": 1.9959899749373436e-05,
+      "loss": 0.5355566143989563,
+      "num_input_tokens_seen": 3032,
+      "step": 10,
+      "train_runtime": 24.9747,
+      "train_tokens_per_second": 121.403
+    },
+    {
+      "epoch": 0.0005057471264367816,
+      "grad_norm": 9.76511287689209,
+      "learning_rate": 1.9949874686716792e-05,
+      "loss": 0.6640444993972778,
+      "num_input_tokens_seen": 3464,
+      "step": 11,
+      "train_runtime": 27.0217,
+      "train_tokens_per_second": 128.193
+    },
+    {
+      "epoch": 0.0005517241379310345,
+      "grad_norm": 10.614198684692383,
+      "learning_rate": 1.9939849624060153e-05,
+      "loss": 0.5045055150985718,
+      "num_input_tokens_seen": 3700,
+      "step": 12,
+      "train_runtime": 29.0095,
+      "train_tokens_per_second": 127.545
+    },
+    {
+      "epoch": 0.0005977011494252873,
+      "grad_norm": 9.133809089660645,
+      "learning_rate": 1.992982456140351e-05,
+      "loss": 0.4403928518295288,
+      "num_input_tokens_seen": 4022,
+      "step": 13,
+      "train_runtime": 30.9759,
+      "train_tokens_per_second": 129.843
+    },
+    {
+      "epoch": 0.0006436781609195402,
+      "grad_norm": 11.173455238342285,
+      "learning_rate": 1.991979949874687e-05,
+      "loss": 0.6436553001403809,
+      "num_input_tokens_seen": 4304,
+      "step": 14,
+      "train_runtime": 32.9504,
+      "train_tokens_per_second": 130.621
+    },
+    {
+      "epoch": 0.000689655172413793,
+      "grad_norm": 10.636919975280762,
+      "learning_rate": 1.9909774436090226e-05,
+      "loss": 0.5784497261047363,
+      "num_input_tokens_seen": 4642,
+      "step": 15,
+      "train_runtime": 34.9078,
+      "train_tokens_per_second": 132.979
+    },
+    {
+      "epoch": 0.000735632183908046,
+      "grad_norm": 9.944845199584961,
+      "learning_rate": 1.9899749373433587e-05,
+      "loss": 0.6040048599243164,
+      "num_input_tokens_seen": 5336,
+      "step": 16,
+      "train_runtime": 36.9248,
+      "train_tokens_per_second": 144.51
+    },
+    {
+      "epoch": 0.0007816091954022989,
+      "grad_norm": 10.018655776977539,
+      "learning_rate": 1.9889724310776943e-05,
+      "loss": 0.43519648909568787,
+      "num_input_tokens_seen": 5602,
+      "step": 17,
+      "train_runtime": 38.8536,
+      "train_tokens_per_second": 144.182
+    },
+    {
+      "epoch": 0.0008275862068965517,
+      "grad_norm": 10.092058181762695,
+      "learning_rate": 1.9879699248120303e-05,
+      "loss": 0.6520774364471436,
+      "num_input_tokens_seen": 6002,
+      "step": 18,
+      "train_runtime": 40.8317,
+      "train_tokens_per_second": 146.994
+    },
+    {
+      "epoch": 0.0008735632183908046,
+      "grad_norm": 12.305459976196289,
+      "learning_rate": 1.986967418546366e-05,
+      "loss": 0.6159787178039551,
+      "num_input_tokens_seen": 6300,
+      "step": 19,
+      "train_runtime": 42.7875,
+      "train_tokens_per_second": 147.239
+    },
+    {
+      "epoch": 0.0009195402298850574,
+      "grad_norm": 8.482680320739746,
+      "learning_rate": 1.9859649122807017e-05,
+      "loss": 0.47896331548690796,
+      "num_input_tokens_seen": 6586,
+      "step": 20,
+      "train_runtime": 44.7248,
+      "train_tokens_per_second": 147.256
+    },
+    {
+      "epoch": 0.0009655172413793104,
+      "grad_norm": 9.402422904968262,
+      "learning_rate": 1.9849624060150377e-05,
+      "loss": 0.5928805470466614,
+      "num_input_tokens_seen": 6918,
+      "step": 21,
+      "train_runtime": 46.7373,
+      "train_tokens_per_second": 148.019
+    },
+    {
+      "epoch": 0.0010114942528735632,
+      "grad_norm": 12.021410942077637,
+      "learning_rate": 1.9839598997493737e-05,
+      "loss": 0.6990500092506409,
+      "num_input_tokens_seen": 7338,
+      "step": 22,
+      "train_runtime": 48.7861,
+      "train_tokens_per_second": 150.412
+    },
+    {
+      "epoch": 0.0010574712643678162,
+      "grad_norm": 9.270434379577637,
+      "learning_rate": 1.9829573934837094e-05,
+      "loss": 0.5009465217590332,
+      "num_input_tokens_seen": 7698,
+      "step": 23,
+      "train_runtime": 50.7911,
+      "train_tokens_per_second": 151.562
+    },
+    {
+      "epoch": 0.001103448275862069,
+      "grad_norm": 11.439963340759277,
+      "learning_rate": 1.981954887218045e-05,
+      "loss": 0.8264314532279968,
+      "num_input_tokens_seen": 8072,
+      "step": 24,
+      "train_runtime": 52.751,
+      "train_tokens_per_second": 153.021
+    },
+    {
+      "epoch": 0.0011494252873563218,
+      "grad_norm": 11.08813762664795,
+      "learning_rate": 1.980952380952381e-05,
+      "loss": 0.4600914716720581,
+      "num_input_tokens_seen": 8340,
+      "step": 25,
+      "train_runtime": 54.6931,
+      "train_tokens_per_second": 152.487
+    },
+    {
+      "epoch": 0.0011954022988505746,
+      "grad_norm": 9.795487403869629,
+      "learning_rate": 1.9799498746867168e-05,
+      "loss": 0.5551027059555054,
+      "num_input_tokens_seen": 8680,
+      "step": 26,
+      "train_runtime": 56.6889,
+      "train_tokens_per_second": 153.116
+    },
+    {
+      "epoch": 0.0012413793103448277,
+      "grad_norm": 10.223047256469727,
+      "learning_rate": 1.9789473684210528e-05,
+      "loss": 0.41297563910484314,
+      "num_input_tokens_seen": 8928,
+      "step": 27,
+      "train_runtime": 58.6221,
+      "train_tokens_per_second": 152.297
+    },
+    {
+      "epoch": 0.0012873563218390805,
+      "grad_norm": 9.74205493927002,
+      "learning_rate": 1.977944862155389e-05,
+      "loss": 0.5344108939170837,
+      "num_input_tokens_seen": 9184,
+      "step": 28,
+      "train_runtime": 60.5843,
+      "train_tokens_per_second": 151.59
+    },
+    {
+      "epoch": 0.0013333333333333333,
+      "grad_norm": 9.142971992492676,
+      "learning_rate": 1.9769423558897245e-05,
+      "loss": 0.45299530029296875,
+      "num_input_tokens_seen": 9520,
+      "step": 29,
+      "train_runtime": 62.5418,
+      "train_tokens_per_second": 152.218
+    },
+    {
+      "epoch": 0.001379310344827586,
+      "grad_norm": 11.530330657958984,
+      "learning_rate": 1.9759398496240602e-05,
+      "loss": 0.63418048620224,
+      "num_input_tokens_seen": 9828,
+      "step": 30,
+      "train_runtime": 64.4704,
+      "train_tokens_per_second": 152.442
+    },
+    {
+      "epoch": 0.0014252873563218391,
+      "grad_norm": 10.06902027130127,
+      "learning_rate": 1.974937343358396e-05,
+      "loss": 0.5002079010009766,
+      "num_input_tokens_seen": 10140,
+      "step": 31,
+      "train_runtime": 81.028,
+      "train_tokens_per_second": 125.142
+    },
+    {
+      "epoch": 0.001471264367816092,
+      "grad_norm": 10.205739974975586,
+      "learning_rate": 1.973934837092732e-05,
+      "loss": 0.6954036951065063,
+      "num_input_tokens_seen": 10526,
+      "step": 32,
+      "train_runtime": 82.9924,
+      "train_tokens_per_second": 126.831
+    },
+    {
+      "epoch": 0.0015172413793103448,
+      "grad_norm": 11.820531845092773,
+      "learning_rate": 1.972932330827068e-05,
+      "loss": 0.43554025888442993,
+      "num_input_tokens_seen": 10756,
+      "step": 33,
+      "train_runtime": 84.9196,
+      "train_tokens_per_second": 126.661
+    },
+    {
+      "epoch": 0.0015632183908045978,
+      "grad_norm": 8.295219421386719,
+      "learning_rate": 1.9719298245614036e-05,
+      "loss": 0.536799967288971,
+      "num_input_tokens_seen": 11066,
+      "step": 34,
+      "train_runtime": 86.8693,
+      "train_tokens_per_second": 127.387
+    },
+    {
+      "epoch": 0.0016091954022988506,
+      "grad_norm": 8.585680961608887,
+      "learning_rate": 1.9709273182957396e-05,
+      "loss": 0.3999759554862976,
+      "num_input_tokens_seen": 11334,
+      "step": 35,
+      "train_runtime": 88.8176,
+      "train_tokens_per_second": 127.61
+    },
+    {
+      "epoch": 0.0016551724137931034,
+      "grad_norm": 7.839630126953125,
+      "learning_rate": 1.9699248120300753e-05,
+      "loss": 0.6960058212280273,
+      "num_input_tokens_seen": 12256,
+      "step": 36,
+      "train_runtime": 91.1022,
+      "train_tokens_per_second": 134.53
+    },
+    {
+      "epoch": 0.0017011494252873562,
+      "grad_norm": 8.731598854064941,
+      "learning_rate": 1.968922305764411e-05,
+      "loss": 0.5457684397697449,
+      "num_input_tokens_seen": 12564,
+      "step": 37,
+      "train_runtime": 93.0479,
+      "train_tokens_per_second": 135.027
+    },
+    {
+      "epoch": 0.0017471264367816092,
+      "grad_norm": 9.757588386535645,
+      "learning_rate": 1.967919799498747e-05,
+      "loss": 0.5432316064834595,
+      "num_input_tokens_seen": 12910,
+      "step": 38,
+      "train_runtime": 95.066,
+      "train_tokens_per_second": 135.8
+    },
+    {
+      "epoch": 0.001793103448275862,
+      "grad_norm": 7.411613941192627,
+      "learning_rate": 1.966917293233083e-05,
+      "loss": 0.3205088675022125,
+      "num_input_tokens_seen": 13254,
+      "step": 39,
+      "train_runtime": 97.0558,
+      "train_tokens_per_second": 136.561
+    },
+    {
+      "epoch": 0.0018390804597701149,
+      "grad_norm": 9.789390563964844,
+      "learning_rate": 1.9659147869674187e-05,
+      "loss": 0.5437139272689819,
+      "num_input_tokens_seen": 13612,
+      "step": 40,
+      "train_runtime": 98.9998,
+      "train_tokens_per_second": 137.495
+    },
+    {
+      "epoch": 0.001885057471264368,
+      "grad_norm": 8.682579040527344,
+      "learning_rate": 1.9649122807017544e-05,
+      "loss": 0.3659517765045166,
+      "num_input_tokens_seen": 13854,
+      "step": 41,
+      "train_runtime": 100.9294,
+      "train_tokens_per_second": 137.264
+    },
+    {
+      "epoch": 0.0019310344827586207,
+      "grad_norm": 10.214683532714844,
+      "learning_rate": 1.9639097744360904e-05,
+      "loss": 0.5193214416503906,
+      "num_input_tokens_seen": 14106,
+      "step": 42,
+      "train_runtime": 103.0205,
+      "train_tokens_per_second": 136.924
+    },
+    {
+      "epoch": 0.0019770114942528737,
+      "grad_norm": 9.461170196533203,
+      "learning_rate": 1.962907268170426e-05,
+      "loss": 0.5038692951202393,
+      "num_input_tokens_seen": 14548,
+      "step": 43,
+      "train_runtime": 105.1014,
+      "train_tokens_per_second": 138.419
+    },
+    {
+      "epoch": 0.0020229885057471263,
+      "grad_norm": 10.129467010498047,
+      "learning_rate": 1.961904761904762e-05,
+      "loss": 0.5905882716178894,
+      "num_input_tokens_seen": 14882,
+      "step": 44,
+      "train_runtime": 107.1422,
+      "train_tokens_per_second": 138.9
+    },
+    {
+      "epoch": 0.0020689655172413794,
+      "grad_norm": 9.763384819030762,
+      "learning_rate": 1.960902255639098e-05,
+      "loss": 0.4151766002178192,
+      "num_input_tokens_seen": 15164,
+      "step": 45,
+      "train_runtime": 109.1638,
+      "train_tokens_per_second": 138.911
+    },
+    {
+      "epoch": 0.0021149425287356324,
+      "grad_norm": 10.556844711303711,
+      "learning_rate": 1.9598997493734338e-05,
+      "loss": 0.42272478342056274,
+      "num_input_tokens_seen": 15464,
+      "step": 46,
+      "train_runtime": 111.1925,
+      "train_tokens_per_second": 139.074
+    },
+    {
+      "epoch": 0.002160919540229885,
+      "grad_norm": 9.503744125366211,
+      "learning_rate": 1.9588972431077695e-05,
+      "loss": 0.522967517375946,
+      "num_input_tokens_seen": 15754,
+      "step": 47,
+      "train_runtime": 113.199,
+      "train_tokens_per_second": 139.171
+    },
+    {
+      "epoch": 0.002206896551724138,
+      "grad_norm": 8.815255165100098,
+      "learning_rate": 1.9578947368421055e-05,
+      "loss": 0.5505152344703674,
+      "num_input_tokens_seen": 16132,
+      "step": 48,
+      "train_runtime": 115.2014,
+      "train_tokens_per_second": 140.033
+    },
+    {
+      "epoch": 0.0022528735632183906,
+      "grad_norm": 9.219003677368164,
+      "learning_rate": 1.956892230576441e-05,
+      "loss": 0.7738328576087952,
+      "num_input_tokens_seen": 16782,
+      "step": 49,
+      "train_runtime": 117.2265,
+      "train_tokens_per_second": 143.159
+    },
+    {
+      "epoch": 0.0022988505747126436,
+      "grad_norm": 10.014391899108887,
+      "learning_rate": 1.9558897243107772e-05,
+      "loss": 0.5957150459289551,
+      "num_input_tokens_seen": 17050,
+      "step": 50,
+      "train_runtime": 119.1627,
+      "train_tokens_per_second": 143.082
+    },
+    {
+      "epoch": 0.0023448275862068967,
+      "grad_norm": 12.310578346252441,
+      "learning_rate": 1.954887218045113e-05,
+      "loss": 0.6443823575973511,
+      "num_input_tokens_seen": 17384,
+      "step": 51,
+      "train_runtime": 121.1125,
+      "train_tokens_per_second": 143.536
+    },
+    {
+      "epoch": 0.0023908045977011493,
+      "grad_norm": 9.436968803405762,
+      "learning_rate": 1.953884711779449e-05,
+      "loss": 0.5737659931182861,
+      "num_input_tokens_seen": 17660,
+      "step": 52,
+      "train_runtime": 123.0371,
+      "train_tokens_per_second": 143.534
+    },
+    {
+      "epoch": 0.0024367816091954023,
+      "grad_norm": 7.799676418304443,
+      "learning_rate": 1.9528822055137845e-05,
+      "loss": 0.4736555814743042,
+      "num_input_tokens_seen": 17926,
+      "step": 53,
+      "train_runtime": 124.969,
+      "train_tokens_per_second": 143.444
+    },
+    {
+      "epoch": 0.0024827586206896553,
+      "grad_norm": 10.809428215026855,
+      "learning_rate": 1.9518796992481202e-05,
+      "loss": 0.5015234351158142,
+      "num_input_tokens_seen": 18186,
+      "step": 54,
+      "train_runtime": 126.9191,
+      "train_tokens_per_second": 143.288
+    },
+    {
+      "epoch": 0.002528735632183908,
+      "grad_norm": 9.467635154724121,
+      "learning_rate": 1.9508771929824562e-05,
+      "loss": 0.6202517747879028,
+      "num_input_tokens_seen": 18542,
+      "step": 55,
+      "train_runtime": 128.8986,
+      "train_tokens_per_second": 143.85
+    },
+    {
+      "epoch": 0.002574712643678161,
+      "grad_norm": 9.191046714782715,
+      "learning_rate": 1.9498746867167923e-05,
+      "loss": 0.5081424117088318,
+      "num_input_tokens_seen": 18892,
+      "step": 56,
+      "train_runtime": 130.848,
+      "train_tokens_per_second": 144.381
+    },
+    {
+      "epoch": 0.002620689655172414,
+      "grad_norm": 8.688216209411621,
+      "learning_rate": 1.948872180451128e-05,
+      "loss": 0.44278833270072937,
+      "num_input_tokens_seen": 19232,
+      "step": 57,
+      "train_runtime": 132.8132,
+      "train_tokens_per_second": 144.805
+    },
+    {
+      "epoch": 0.0026666666666666666,
+      "grad_norm": 8.183144569396973,
+      "learning_rate": 1.947869674185464e-05,
+      "loss": 0.49157488346099854,
+      "num_input_tokens_seen": 19502,
+      "step": 58,
+      "train_runtime": 134.7752,
+      "train_tokens_per_second": 144.7
+    },
+    {
+      "epoch": 0.0027126436781609196,
+      "grad_norm": 9.218542098999023,
+      "learning_rate": 1.9468671679197996e-05,
+      "loss": 0.5725117921829224,
+      "num_input_tokens_seen": 19910,
+      "step": 59,
+      "train_runtime": 136.7257,
+      "train_tokens_per_second": 145.62
+    },
+    {
+      "epoch": 0.002758620689655172,
+      "grad_norm": 8.710495948791504,
+      "learning_rate": 1.9458646616541353e-05,
+      "loss": 0.4720567464828491,
+      "num_input_tokens_seen": 20186,
+      "step": 60,
+      "train_runtime": 138.6459,
+      "train_tokens_per_second": 145.594
+    },
+    {
+      "epoch": 0.0028045977011494252,
+      "grad_norm": 10.533473014831543,
+      "learning_rate": 1.9448621553884713e-05,
+      "loss": 0.438642680644989,
+      "num_input_tokens_seen": 20454,
+      "step": 61,
+      "train_runtime": 153.9405,
+      "train_tokens_per_second": 132.87
+    },
+    {
+      "epoch": 0.0028505747126436783,
+      "grad_norm": 8.03281021118164,
+      "learning_rate": 1.9438596491228074e-05,
+      "loss": 0.5351825952529907,
+      "num_input_tokens_seen": 20798,
+      "step": 62,
+      "train_runtime": 155.8779,
+      "train_tokens_per_second": 133.425
+    },
+    {
+      "epoch": 0.002896551724137931,
+      "grad_norm": 8.456392288208008,
+      "learning_rate": 1.942857142857143e-05,
+      "loss": 0.5822426676750183,
+      "num_input_tokens_seen": 21272,
+      "step": 63,
+      "train_runtime": 157.885,
+      "train_tokens_per_second": 134.731
+    },
+    {
+      "epoch": 0.002942528735632184,
+      "grad_norm": 11.524544715881348,
+      "learning_rate": 1.9418546365914787e-05,
+      "loss": 0.7060003280639648,
+      "num_input_tokens_seen": 21562,
+      "step": 64,
+      "train_runtime": 159.8005,
+      "train_tokens_per_second": 134.931
+    },
+    {
+      "epoch": 0.002988505747126437,
+      "grad_norm": 7.4113969802856445,
+      "learning_rate": 1.9408521303258147e-05,
+      "loss": 0.3441489338874817,
+      "num_input_tokens_seen": 21814,
+      "step": 65,
+      "train_runtime": 161.7111,
+      "train_tokens_per_second": 134.895
+    },
+    {
+      "epoch": 0.0030344827586206895,
+      "grad_norm": 8.198412895202637,
+      "learning_rate": 1.9398496240601504e-05,
+      "loss": 0.505254864692688,
+      "num_input_tokens_seen": 22102,
+      "step": 66,
+      "train_runtime": 163.6425,
+      "train_tokens_per_second": 135.063
+    },
+    {
+      "epoch": 0.0030804597701149425,
+      "grad_norm": 9.072737693786621,
+      "learning_rate": 1.9388471177944864e-05,
+      "loss": 0.5357642769813538,
+      "num_input_tokens_seen": 22426,
+      "step": 67,
+      "train_runtime": 165.5707,
+      "train_tokens_per_second": 135.447
+    },
+    {
+      "epoch": 0.0031264367816091956,
+      "grad_norm": 7.890419960021973,
+      "learning_rate": 1.9378446115288224e-05,
+      "loss": 0.5315457582473755,
+      "num_input_tokens_seen": 22722,
+      "step": 68,
+      "train_runtime": 167.5015,
+      "train_tokens_per_second": 135.653
+    },
+    {
+      "epoch": 0.003172413793103448,
+      "grad_norm": 9.176838874816895,
+      "learning_rate": 1.936842105263158e-05,
+      "loss": 0.6139141321182251,
+      "num_input_tokens_seen": 23144,
+      "step": 69,
+      "train_runtime": 169.4794,
+      "train_tokens_per_second": 136.559
+    },
+    {
+      "epoch": 0.003218390804597701,
+      "grad_norm": 9.316536903381348,
+      "learning_rate": 1.9358395989974938e-05,
+      "loss": 0.6673556566238403,
+      "num_input_tokens_seen": 23446,
+      "step": 70,
+      "train_runtime": 171.4083,
+      "train_tokens_per_second": 136.785
+    },
+    {
+      "epoch": 0.0032643678160919542,
+      "grad_norm": 10.201543807983398,
+      "learning_rate": 1.9348370927318295e-05,
+      "loss": 0.5292945504188538,
+      "num_input_tokens_seen": 23768,
+      "step": 71,
+      "train_runtime": 173.3509,
+      "train_tokens_per_second": 137.109
+    },
+    {
+      "epoch": 0.003310344827586207,
+      "grad_norm": 8.47014045715332,
+      "learning_rate": 1.9338345864661655e-05,
+      "loss": 0.41301482915878296,
+      "num_input_tokens_seen": 23986,
+      "step": 72,
+      "train_runtime": 175.333,
+      "train_tokens_per_second": 136.803
+    },
+    {
+      "epoch": 0.00335632183908046,
+      "grad_norm": 7.975886821746826,
+      "learning_rate": 1.9328320802005015e-05,
+      "loss": 0.46357280015945435,
+      "num_input_tokens_seen": 24270,
+      "step": 73,
+      "train_runtime": 177.5143,
+      "train_tokens_per_second": 136.721
+    },
+    {
+      "epoch": 0.0034022988505747124,
+      "grad_norm": 8.95450496673584,
+      "learning_rate": 1.9318295739348372e-05,
+      "loss": 0.5103389620780945,
+      "num_input_tokens_seen": 24520,
+      "step": 74,
+      "train_runtime": 179.4746,
+      "train_tokens_per_second": 136.621
+    },
+    {
+      "epoch": 0.0034482758620689655,
+      "grad_norm": 9.743664741516113,
+      "learning_rate": 1.9308270676691732e-05,
+      "loss": 0.6401261687278748,
+      "num_input_tokens_seen": 24800,
+      "step": 75,
+      "train_runtime": 181.3884,
+      "train_tokens_per_second": 136.723
+    },
+    {
+      "epoch": 0.0034942528735632185,
+      "grad_norm": 9.316505432128906,
+      "learning_rate": 1.929824561403509e-05,
+      "loss": 0.4908076524734497,
+      "num_input_tokens_seen": 25068,
+      "step": 76,
+      "train_runtime": 183.292,
+      "train_tokens_per_second": 136.765
+    },
+    {
+      "epoch": 0.003540229885057471,
+      "grad_norm": 7.853257179260254,
+      "learning_rate": 1.9288220551378446e-05,
+      "loss": 0.4684579372406006,
+      "num_input_tokens_seen": 25304,
+      "step": 77,
+      "train_runtime": 185.2082,
+      "train_tokens_per_second": 136.625
+    },
+    {
+      "epoch": 0.003586206896551724,
+      "grad_norm": 9.16893196105957,
+      "learning_rate": 1.9278195488721806e-05,
+      "loss": 0.5689425468444824,
+      "num_input_tokens_seen": 25570,
+      "step": 78,
+      "train_runtime": 187.1387,
+      "train_tokens_per_second": 136.637
+    },
+    {
+      "epoch": 0.003632183908045977,
+      "grad_norm": 9.800451278686523,
+      "learning_rate": 1.9268170426065166e-05,
+      "loss": 0.5715272426605225,
+      "num_input_tokens_seen": 25864,
+      "step": 79,
+      "train_runtime": 189.0995,
+      "train_tokens_per_second": 136.775
+    },
+    {
+      "epoch": 0.0036781609195402297,
+      "grad_norm": 9.618143081665039,
+      "learning_rate": 1.9258145363408523e-05,
+      "loss": 0.5264904499053955,
+      "num_input_tokens_seen": 26140,
+      "step": 80,
+      "train_runtime": 191.0752,
+      "train_tokens_per_second": 136.805
+    },
+    {
+      "epoch": 0.0037241379310344828,
+      "grad_norm": 8.853523254394531,
+      "learning_rate": 1.924812030075188e-05,
+      "loss": 0.41909778118133545,
+      "num_input_tokens_seen": 26420,
+      "step": 81,
+      "train_runtime": 192.9847,
+      "train_tokens_per_second": 136.902
+    },
+    {
+      "epoch": 0.003770114942528736,
+      "grad_norm": 10.117783546447754,
+      "learning_rate": 1.923809523809524e-05,
+      "loss": 0.5639157891273499,
+      "num_input_tokens_seen": 26702,
+      "step": 82,
+      "train_runtime": 194.9084,
+      "train_tokens_per_second": 136.998
+    },
+    {
+      "epoch": 0.0038160919540229884,
+      "grad_norm": 8.727978706359863,
+      "learning_rate": 1.9228070175438597e-05,
+      "loss": 0.5927528142929077,
+      "num_input_tokens_seen": 27088,
+      "step": 83,
+      "train_runtime": 196.8634,
+      "train_tokens_per_second": 137.598
+    },
+    {
+      "epoch": 0.0038620689655172414,
+      "grad_norm": 8.757131576538086,
+      "learning_rate": 1.9218045112781957e-05,
+      "loss": 0.5949188470840454,
+      "num_input_tokens_seen": 27510,
+      "step": 84,
+      "train_runtime": 198.8064,
+      "train_tokens_per_second": 138.376
+    },
+    {
+      "epoch": 0.003908045977011494,
+      "grad_norm": 7.067785739898682,
+      "learning_rate": 1.9208020050125317e-05,
+      "loss": 0.4642144739627838,
+      "num_input_tokens_seen": 27988,
+      "step": 85,
+      "train_runtime": 200.7676,
+      "train_tokens_per_second": 139.405
+    },
+    {
+      "epoch": 0.0039540229885057475,
+      "grad_norm": 8.996391296386719,
+      "learning_rate": 1.9197994987468674e-05,
+      "loss": 0.421725869178772,
+      "num_input_tokens_seen": 28210,
+      "step": 86,
+      "train_runtime": 202.6595,
+      "train_tokens_per_second": 139.199
+    },
+    {
+      "epoch": 0.004,
+      "grad_norm": 8.663325309753418,
+      "learning_rate": 1.918796992481203e-05,
+      "loss": 0.560427188873291,
+      "num_input_tokens_seen": 28532,
+      "step": 87,
+      "train_runtime": 204.5835,
+      "train_tokens_per_second": 139.464
+    },
+    {
+      "epoch": 0.004045977011494253,
+      "grad_norm": 10.188467979431152,
+      "learning_rate": 1.917794486215539e-05,
+      "loss": 0.5534644722938538,
+      "num_input_tokens_seen": 28828,
+      "step": 88,
+      "train_runtime": 206.5008,
+      "train_tokens_per_second": 139.602
+    },
+    {
+      "epoch": 0.004091954022988506,
+      "grad_norm": 9.026497840881348,
+      "learning_rate": 1.9167919799498748e-05,
+      "loss": 0.6749687194824219,
+      "num_input_tokens_seen": 29260,
+      "step": 89,
+      "train_runtime": 208.4575,
+      "train_tokens_per_second": 140.364
+    },
+    {
+      "epoch": 0.004137931034482759,
+      "grad_norm": 8.981388092041016,
+      "learning_rate": 1.9157894736842108e-05,
+      "loss": 0.48772114515304565,
+      "num_input_tokens_seen": 29562,
+      "step": 90,
+      "train_runtime": 210.3843,
+      "train_tokens_per_second": 140.514
+    },
+    {
+      "epoch": 0.004183908045977011,
+      "grad_norm": 9.265682220458984,
+      "learning_rate": 1.9147869674185465e-05,
+      "loss": 0.5756199359893799,
+      "num_input_tokens_seen": 29832,
+      "step": 91,
+      "train_runtime": 224.6014,
+      "train_tokens_per_second": 132.822
+    },
+    {
+      "epoch": 0.004229885057471265,
+      "grad_norm": 7.933191299438477,
+      "learning_rate": 1.9137844611528825e-05,
+      "loss": 0.6041742563247681,
+      "num_input_tokens_seen": 30484,
+      "step": 92,
+      "train_runtime": 226.6451,
+      "train_tokens_per_second": 134.501
+    },
+    {
+      "epoch": 0.004275862068965517,
+      "grad_norm": 9.629735946655273,
+      "learning_rate": 1.912781954887218e-05,
+      "loss": 0.6612107157707214,
+      "num_input_tokens_seen": 30942,
+      "step": 93,
+      "train_runtime": 228.6489,
+      "train_tokens_per_second": 135.325
+    },
+    {
+      "epoch": 0.00432183908045977,
+      "grad_norm": 8.997493743896484,
+      "learning_rate": 1.911779448621554e-05,
+      "loss": 0.5092484951019287,
+      "num_input_tokens_seen": 31340,
+      "step": 94,
+      "train_runtime": 230.7289,
+      "train_tokens_per_second": 135.83
+    },
+    {
+      "epoch": 0.004367816091954023,
+      "grad_norm": 8.316047668457031,
+      "learning_rate": 1.91077694235589e-05,
+      "loss": 0.46212780475616455,
+      "num_input_tokens_seen": 31596,
+      "step": 95,
+      "train_runtime": 232.8291,
+      "train_tokens_per_second": 135.705
+    },
+    {
+      "epoch": 0.004413793103448276,
+      "grad_norm": 9.808420181274414,
+      "learning_rate": 1.909774436090226e-05,
+      "loss": 0.541647732257843,
+      "num_input_tokens_seen": 32008,
+      "step": 96,
+      "train_runtime": 234.8559,
+      "train_tokens_per_second": 136.288
+    },
+    {
+      "epoch": 0.004459770114942529,
+      "grad_norm": 8.325944900512695,
+      "learning_rate": 1.9087719298245616e-05,
+      "loss": 0.6134788990020752,
+      "num_input_tokens_seen": 32284,
+      "step": 97,
+      "train_runtime": 236.785,
+      "train_tokens_per_second": 136.343
+    },
+    {
+      "epoch": 0.004505747126436781,
+      "grad_norm": 9.541683197021484,
+      "learning_rate": 1.9077694235588976e-05,
+      "loss": 0.5184969305992126,
+      "num_input_tokens_seen": 32520,
+      "step": 98,
+      "train_runtime": 238.7158,
+      "train_tokens_per_second": 136.229
+    },
+    {
+      "epoch": 0.004551724137931035,
+      "grad_norm": 8.778630256652832,
+      "learning_rate": 1.9067669172932333e-05,
+      "loss": 0.5224044322967529,
+      "num_input_tokens_seen": 32756,
+      "step": 99,
+      "train_runtime": 240.6471,
+      "train_tokens_per_second": 136.116
+    },
+    {
+      "epoch": 0.004597701149425287,
+      "grad_norm": 8.830920219421387,
+      "learning_rate": 1.905764411027569e-05,
+      "loss": 0.6011303067207336,
+      "num_input_tokens_seen": 33066,
+      "step": 100,
+      "train_runtime": 242.5824,
+      "train_tokens_per_second": 136.308
+    },
+    {
+      "epoch": 0.00464367816091954,
+      "grad_norm": 8.93142032623291,
+      "learning_rate": 1.904761904761905e-05,
+      "loss": 0.4904657006263733,
+      "num_input_tokens_seen": 33342,
+      "step": 101,
+      "train_runtime": 244.5372,
+      "train_tokens_per_second": 136.347
+    },
+    {
+      "epoch": 0.004689655172413793,
+      "grad_norm": 11.11080265045166,
+      "learning_rate": 1.903759398496241e-05,
+      "loss": 0.5706338286399841,
+      "num_input_tokens_seen": 33664,
+      "step": 102,
+      "train_runtime": 246.4553,
+      "train_tokens_per_second": 136.593
+    },
+    {
+      "epoch": 0.004735632183908046,
+      "grad_norm": 7.9319353103637695,
+      "learning_rate": 1.9027568922305766e-05,
+      "loss": 0.4006994664669037,
+      "num_input_tokens_seen": 33900,
+      "step": 103,
+      "train_runtime": 248.3577,
+      "train_tokens_per_second": 136.497
+    },
+    {
+      "epoch": 0.0047816091954022985,
+      "grad_norm": 11.371949195861816,
+      "learning_rate": 1.9017543859649123e-05,
+      "loss": 0.5322912335395813,
+      "num_input_tokens_seen": 34194,
+      "step": 104,
+      "train_runtime": 250.2764,
+      "train_tokens_per_second": 136.625
+    },
+    {
+      "epoch": 0.004827586206896552,
+      "grad_norm": 8.555707931518555,
+      "learning_rate": 1.9007518796992483e-05,
+      "loss": 0.4682633876800537,
+      "num_input_tokens_seen": 34756,
+      "step": 105,
+      "train_runtime": 252.2596,
+      "train_tokens_per_second": 137.779
+    },
+    {
+      "epoch": 0.004873563218390805,
+      "grad_norm": 12.503046989440918,
+      "learning_rate": 1.899749373433584e-05,
+      "loss": 0.43531227111816406,
+      "num_input_tokens_seen": 35054,
+      "step": 106,
+      "train_runtime": 254.1754,
+      "train_tokens_per_second": 137.913
+    },
+    {
+      "epoch": 0.004919540229885057,
+      "grad_norm": 8.476236343383789,
+      "learning_rate": 1.89874686716792e-05,
+      "loss": 0.3929433822631836,
+      "num_input_tokens_seen": 35300,
+      "step": 107,
+      "train_runtime": 256.0932,
+      "train_tokens_per_second": 137.84
+    },
+    {
+      "epoch": 0.004965517241379311,
+      "grad_norm": 9.59035873413086,
+      "learning_rate": 1.8977443609022557e-05,
+      "loss": 0.6225996017456055,
+      "num_input_tokens_seen": 35600,
+      "step": 108,
+      "train_runtime": 258.0261,
+      "train_tokens_per_second": 137.971
+    },
+    {
+      "epoch": 0.005011494252873563,
+      "grad_norm": 8.283434867858887,
+      "learning_rate": 1.8967418546365917e-05,
+      "loss": 0.5337879657745361,
+      "num_input_tokens_seen": 35966,
+      "step": 109,
+      "train_runtime": 259.9956,
+      "train_tokens_per_second": 138.333
+    },
+    {
+      "epoch": 0.005057471264367816,
+      "grad_norm": 7.715997219085693,
+      "learning_rate": 1.8957393483709274e-05,
+      "loss": 0.5463548898696899,
+      "num_input_tokens_seen": 36406,
+      "step": 110,
+      "train_runtime": 261.9563,
+      "train_tokens_per_second": 138.977
+    },
+    {
+      "epoch": 0.005103448275862069,
+      "grad_norm": 8.812264442443848,
+      "learning_rate": 1.894736842105263e-05,
+      "loss": 0.6220695972442627,
+      "num_input_tokens_seen": 37082,
+      "step": 111,
+      "train_runtime": 264.0358,
+      "train_tokens_per_second": 140.443
+    },
+    {
+      "epoch": 0.005149425287356322,
+      "grad_norm": 8.751511573791504,
+      "learning_rate": 1.893734335839599e-05,
+      "loss": 0.4591900706291199,
+      "num_input_tokens_seen": 37384,
+      "step": 112,
+      "train_runtime": 265.9926,
+      "train_tokens_per_second": 140.545
+    },
+    {
+      "epoch": 0.0051954022988505745,
+      "grad_norm": 8.574929237365723,
+      "learning_rate": 1.892731829573935e-05,
+      "loss": 0.5448266267776489,
+      "num_input_tokens_seen": 37634,
+      "step": 113,
+      "train_runtime": 267.9425,
+      "train_tokens_per_second": 140.456
+    },
+    {
+      "epoch": 0.005241379310344828,
+      "grad_norm": 9.245474815368652,
+      "learning_rate": 1.8917293233082708e-05,
+      "loss": 0.643197774887085,
+      "num_input_tokens_seen": 37920,
+      "step": 114,
+      "train_runtime": 269.9313,
+      "train_tokens_per_second": 140.48
+    },
+    {
+      "epoch": 0.0052873563218390806,
+      "grad_norm": 8.642900466918945,
+      "learning_rate": 1.890726817042607e-05,
+      "loss": 0.49965792894363403,
+      "num_input_tokens_seen": 38218,
+      "step": 115,
+      "train_runtime": 271.993,
+      "train_tokens_per_second": 140.511
+    },
+    {
+      "epoch": 0.005333333333333333,
+      "grad_norm": 10.202279090881348,
+      "learning_rate": 1.8897243107769425e-05,
+      "loss": 0.5130624771118164,
+      "num_input_tokens_seen": 38542,
+      "step": 116,
+      "train_runtime": 273.9988,
+      "train_tokens_per_second": 140.665
+    },
+    {
+      "epoch": 0.005379310344827587,
+      "grad_norm": 6.849987030029297,
+      "learning_rate": 1.8887218045112782e-05,
+      "loss": 0.35125046968460083,
+      "num_input_tokens_seen": 38800,
+      "step": 117,
+      "train_runtime": 275.9216,
+      "train_tokens_per_second": 140.62
+    },
+    {
+      "epoch": 0.005425287356321839,
+      "grad_norm": 9.680569648742676,
+      "learning_rate": 1.8877192982456142e-05,
+      "loss": 0.5921655297279358,
+      "num_input_tokens_seen": 39094,
+      "step": 118,
+      "train_runtime": 277.894,
+      "train_tokens_per_second": 140.68
+    },
+    {
+      "epoch": 0.005471264367816092,
+      "grad_norm": 9.317355155944824,
+      "learning_rate": 1.88671679197995e-05,
+      "loss": 0.5994471311569214,
+      "num_input_tokens_seen": 39378,
+      "step": 119,
+      "train_runtime": 279.809,
+      "train_tokens_per_second": 140.732
+    },
+    {
+      "epoch": 0.005517241379310344,
+      "grad_norm": 9.860715866088867,
+      "learning_rate": 1.885714285714286e-05,
+      "loss": 0.40653514862060547,
+      "num_input_tokens_seen": 39632,
+      "step": 120,
+      "train_runtime": 281.7164,
+      "train_tokens_per_second": 140.681
+    },
+    {
+      "epoch": 0.005563218390804598,
+      "grad_norm": 8.424166679382324,
+      "learning_rate": 1.8847117794486216e-05,
+      "loss": 0.4481711983680725,
+      "num_input_tokens_seen": 40052,
+      "step": 121,
+      "train_runtime": 296.5872,
+      "train_tokens_per_second": 135.043
+    },
+    {
+      "epoch": 0.0056091954022988505,
+      "grad_norm": 7.983243942260742,
+      "learning_rate": 1.8837092731829576e-05,
+      "loss": 0.5721695423126221,
+      "num_input_tokens_seen": 40526,
+      "step": 122,
+      "train_runtime": 298.5699,
+      "train_tokens_per_second": 135.734
+    },
+    {
+      "epoch": 0.005655172413793103,
+      "grad_norm": 9.069170951843262,
+      "learning_rate": 1.8827067669172933e-05,
+      "loss": 0.4698539972305298,
+      "num_input_tokens_seen": 40772,
+      "step": 123,
+      "train_runtime": 300.4958,
+      "train_tokens_per_second": 135.682
+    },
+    {
+      "epoch": 0.0057011494252873565,
+      "grad_norm": 8.430562973022461,
+      "learning_rate": 1.8817042606516293e-05,
+      "loss": 0.49493882060050964,
+      "num_input_tokens_seen": 41018,
+      "step": 124,
+      "train_runtime": 302.5117,
+      "train_tokens_per_second": 135.591
+    },
+    {
+      "epoch": 0.005747126436781609,
+      "grad_norm": 9.015573501586914,
+      "learning_rate": 1.880701754385965e-05,
+      "loss": 0.5125147700309753,
+      "num_input_tokens_seen": 41288,
+      "step": 125,
+      "train_runtime": 304.4739,
+      "train_tokens_per_second": 135.604
+    },
+    {
+      "epoch": 0.005793103448275862,
+      "grad_norm": 9.86005973815918,
+      "learning_rate": 1.879699248120301e-05,
+      "loss": 0.4860873818397522,
+      "num_input_tokens_seen": 41610,
+      "step": 126,
+      "train_runtime": 306.5398,
+      "train_tokens_per_second": 135.741
+    },
+    {
+      "epoch": 0.005839080459770115,
+      "grad_norm": 9.298399925231934,
+      "learning_rate": 1.8786967418546367e-05,
+      "loss": 0.4224809408187866,
+      "num_input_tokens_seen": 41880,
+      "step": 127,
+      "train_runtime": 308.5124,
+      "train_tokens_per_second": 135.748
+    },
+    {
+      "epoch": 0.005885057471264368,
+      "grad_norm": 9.295427322387695,
+      "learning_rate": 1.8776942355889727e-05,
+      "loss": 0.5424748659133911,
+      "num_input_tokens_seen": 42238,
+      "step": 128,
+      "train_runtime": 310.5302,
+      "train_tokens_per_second": 136.019
+    },
+    {
+      "epoch": 0.00593103448275862,
+      "grad_norm": 9.500751495361328,
+      "learning_rate": 1.8766917293233084e-05,
+      "loss": 0.4178183078765869,
+      "num_input_tokens_seen": 42520,
+      "step": 129,
+      "train_runtime": 312.4986,
+      "train_tokens_per_second": 136.065
+    },
+    {
+      "epoch": 0.005977011494252874,
+      "grad_norm": 7.819894313812256,
+      "learning_rate": 1.8756892230576444e-05,
+      "loss": 0.38070225715637207,
+      "num_input_tokens_seen": 42782,
+      "step": 130,
+      "train_runtime": 314.4247,
+      "train_tokens_per_second": 136.064
+    },
+    {
+      "epoch": 0.006022988505747126,
+      "grad_norm": 7.9321513175964355,
+      "learning_rate": 1.87468671679198e-05,
+      "loss": 0.39679110050201416,
+      "num_input_tokens_seen": 43024,
+      "step": 131,
+      "train_runtime": 316.3573,
+      "train_tokens_per_second": 135.998
+    },
+    {
+      "epoch": 0.006068965517241379,
+      "grad_norm": 8.515424728393555,
+      "learning_rate": 1.873684210526316e-05,
+      "loss": 0.49936944246292114,
+      "num_input_tokens_seen": 43340,
+      "step": 132,
+      "train_runtime": 318.2971,
+      "train_tokens_per_second": 136.162
+    },
+    {
+      "epoch": 0.0061149425287356325,
+      "grad_norm": 10.381828308105469,
+      "learning_rate": 1.8726817042606518e-05,
+      "loss": 0.6803163290023804,
+      "num_input_tokens_seen": 43844,
+      "step": 133,
+      "train_runtime": 320.4357,
+      "train_tokens_per_second": 136.826
+    },
+    {
+      "epoch": 0.006160919540229885,
+      "grad_norm": 8.881560325622559,
+      "learning_rate": 1.8716791979949875e-05,
+      "loss": 0.4254298806190491,
+      "num_input_tokens_seen": 44104,
+      "step": 134,
+      "train_runtime": 322.6003,
+      "train_tokens_per_second": 136.714
+    },
+    {
+      "epoch": 0.006206896551724138,
+      "grad_norm": 8.319070816040039,
+      "learning_rate": 1.8706766917293235e-05,
+      "loss": 0.30514371395111084,
+      "num_input_tokens_seen": 44342,
+      "step": 135,
+      "train_runtime": 324.7113,
+      "train_tokens_per_second": 136.558
+    },
+    {
+      "epoch": 0.006252873563218391,
+      "grad_norm": 8.256769180297852,
+      "learning_rate": 1.869674185463659e-05,
+      "loss": 0.40613967180252075,
+      "num_input_tokens_seen": 44628,
+      "step": 136,
+      "train_runtime": 326.8512,
+      "train_tokens_per_second": 136.539
+    },
+    {
+      "epoch": 0.006298850574712644,
+      "grad_norm": 9.814496994018555,
+      "learning_rate": 1.868671679197995e-05,
+      "loss": 0.40587806701660156,
+      "num_input_tokens_seen": 44894,
+      "step": 137,
+      "train_runtime": 328.814,
+      "train_tokens_per_second": 136.533
+    },
+    {
+      "epoch": 0.006344827586206896,
+      "grad_norm": 9.767206192016602,
+      "learning_rate": 1.8676691729323312e-05,
+      "loss": 0.39910808205604553,
+      "num_input_tokens_seen": 45126,
+      "step": 138,
+      "train_runtime": 330.7651,
+      "train_tokens_per_second": 136.429
+    },
+    {
+      "epoch": 0.00639080459770115,
+      "grad_norm": 8.380473136901855,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 0.5884116291999817,
+      "num_input_tokens_seen": 45448,
+      "step": 139,
+      "train_runtime": 332.7571,
+      "train_tokens_per_second": 136.58
+    },
+    {
+      "epoch": 0.006436781609195402,
+      "grad_norm": 7.716179847717285,
+      "learning_rate": 1.8656641604010025e-05,
+      "loss": 0.49266794323921204,
+      "num_input_tokens_seen": 45766,
+      "step": 140,
+      "train_runtime": 334.7245,
+      "train_tokens_per_second": 136.727
+    },
+    {
+      "epoch": 0.006482758620689655,
+      "grad_norm": 9.636268615722656,
+      "learning_rate": 1.8646616541353386e-05,
+      "loss": 0.5481572151184082,
+      "num_input_tokens_seen": 46080,
+      "step": 141,
+      "train_runtime": 336.6869,
+      "train_tokens_per_second": 136.863
+    },
+    {
+      "epoch": 0.0065287356321839084,
+      "grad_norm": 8.68143081665039,
+      "learning_rate": 1.8636591478696742e-05,
+      "loss": 0.629855751991272,
+      "num_input_tokens_seen": 46476,
+      "step": 142,
+      "train_runtime": 338.6892,
+      "train_tokens_per_second": 137.223
+    },
+    {
+      "epoch": 0.006574712643678161,
+      "grad_norm": 9.662667274475098,
+      "learning_rate": 1.8626566416040103e-05,
+      "loss": 0.529405951499939,
+      "num_input_tokens_seen": 46708,
+      "step": 143,
+      "train_runtime": 340.6349,
+      "train_tokens_per_second": 137.12
+    },
+    {
+      "epoch": 0.006620689655172414,
+      "grad_norm": 10.860309600830078,
+      "learning_rate": 1.861654135338346e-05,
+      "loss": 0.41112494468688965,
+      "num_input_tokens_seen": 47002,
+      "step": 144,
+      "train_runtime": 342.6129,
+      "train_tokens_per_second": 137.187
+    },
+    {
+      "epoch": 0.006666666666666667,
+      "grad_norm": 8.208048820495605,
+      "learning_rate": 1.860651629072682e-05,
+      "loss": 0.4882567524909973,
+      "num_input_tokens_seen": 47324,
+      "step": 145,
+      "train_runtime": 344.617,
+      "train_tokens_per_second": 137.323
+    },
+    {
+      "epoch": 0.00671264367816092,
+      "grad_norm": 8.881811141967773,
+      "learning_rate": 1.8596491228070176e-05,
+      "loss": 0.48573899269104004,
+      "num_input_tokens_seen": 47620,
+      "step": 146,
+      "train_runtime": 346.5379,
+      "train_tokens_per_second": 137.416
+    },
+    {
+      "epoch": 0.006758620689655172,
+      "grad_norm": 9.245759963989258,
+      "learning_rate": 1.8586466165413533e-05,
+      "loss": 0.44191601872444153,
+      "num_input_tokens_seen": 48142,
+      "step": 147,
+      "train_runtime": 348.5843,
+      "train_tokens_per_second": 138.107
+    },
+    {
+      "epoch": 0.006804597701149425,
+      "grad_norm": 8.131628036499023,
+      "learning_rate": 1.8576441102756893e-05,
+      "loss": 0.4785696268081665,
+      "num_input_tokens_seen": 48756,
+      "step": 148,
+      "train_runtime": 350.6028,
+      "train_tokens_per_second": 139.063
+    },
+    {
+      "epoch": 0.006850574712643678,
+      "grad_norm": 10.151355743408203,
+      "learning_rate": 1.8566416040100254e-05,
+      "loss": 0.6337138414382935,
+      "num_input_tokens_seen": 49072,
+      "step": 149,
+      "train_runtime": 352.5517,
+      "train_tokens_per_second": 139.191
+    },
+    {
+      "epoch": 0.006896551724137931,
+      "grad_norm": 10.633296012878418,
+      "learning_rate": 1.855639097744361e-05,
+      "loss": 0.5732593536376953,
+      "num_input_tokens_seen": 49412,
+      "step": 150,
+      "train_runtime": 354.4895,
+      "train_tokens_per_second": 139.389
+    },
+    {
+      "epoch": 0.0069425287356321835,
+      "grad_norm": 9.780455589294434,
+      "learning_rate": 1.8546365914786967e-05,
+      "loss": 0.43314051628112793,
+      "num_input_tokens_seen": 49680,
+      "step": 151,
+      "train_runtime": 368.6791,
+      "train_tokens_per_second": 134.751
+    },
+    {
+      "epoch": 0.006988505747126437,
+      "grad_norm": 7.924914836883545,
+      "learning_rate": 1.8536340852130327e-05,
+      "loss": 0.45942002534866333,
+      "num_input_tokens_seen": 49990,
+      "step": 152,
+      "train_runtime": 370.611,
+      "train_tokens_per_second": 134.885
+    },
+    {
+      "epoch": 0.00703448275862069,
+      "grad_norm": 7.610330104827881,
+      "learning_rate": 1.8526315789473684e-05,
+      "loss": 0.5856990218162537,
+      "num_input_tokens_seen": 50302,
+      "step": 153,
+      "train_runtime": 372.5391,
+      "train_tokens_per_second": 135.025
+    },
+    {
+      "epoch": 0.007080459770114942,
+      "grad_norm": 8.95278549194336,
+      "learning_rate": 1.8516290726817044e-05,
+      "loss": 0.5225367546081543,
+      "num_input_tokens_seen": 50570,
+      "step": 154,
+      "train_runtime": 374.5819,
+      "train_tokens_per_second": 135.004
+    },
+    {
+      "epoch": 0.007126436781609196,
+      "grad_norm": 8.637958526611328,
+      "learning_rate": 1.8506265664160404e-05,
+      "loss": 0.5034955739974976,
+      "num_input_tokens_seen": 50962,
+      "step": 155,
+      "train_runtime": 376.7162,
+      "train_tokens_per_second": 135.28
+    },
+    {
+      "epoch": 0.007172413793103448,
+      "grad_norm": 8.947890281677246,
+      "learning_rate": 1.849624060150376e-05,
+      "loss": 0.4520954489707947,
+      "num_input_tokens_seen": 51234,
+      "step": 156,
+      "train_runtime": 378.6622,
+      "train_tokens_per_second": 135.303
+    },
+    {
+      "epoch": 0.007218390804597701,
+      "grad_norm": 8.568826675415039,
+      "learning_rate": 1.8486215538847118e-05,
+      "loss": 0.4783710241317749,
+      "num_input_tokens_seen": 51550,
+      "step": 157,
+      "train_runtime": 380.5796,
+      "train_tokens_per_second": 135.451
+    },
+    {
+      "epoch": 0.007264367816091954,
+      "grad_norm": 7.674930572509766,
+      "learning_rate": 1.8476190476190478e-05,
+      "loss": 0.5886533260345459,
+      "num_input_tokens_seen": 52198,
+      "step": 158,
+      "train_runtime": 382.5782,
+      "train_tokens_per_second": 136.437
+    },
+    {
+      "epoch": 0.007310344827586207,
+      "grad_norm": 8.045730590820312,
+      "learning_rate": 1.8466165413533835e-05,
+      "loss": 0.5183379054069519,
+      "num_input_tokens_seen": 52594,
+      "step": 159,
+      "train_runtime": 384.5283,
+      "train_tokens_per_second": 136.775
+    },
+    {
+      "epoch": 0.0073563218390804595,
+      "grad_norm": 11.347859382629395,
+      "learning_rate": 1.8456140350877195e-05,
+      "loss": 0.4976785182952881,
+      "num_input_tokens_seen": 52876,
+      "step": 160,
+      "train_runtime": 386.4424,
+      "train_tokens_per_second": 136.828
+    },
+    {
+      "epoch": 0.007402298850574713,
+      "grad_norm": 7.577949047088623,
+      "learning_rate": 1.8446115288220552e-05,
+      "loss": 0.4495604336261749,
+      "num_input_tokens_seen": 53224,
+      "step": 161,
+      "train_runtime": 388.3505,
+      "train_tokens_per_second": 137.051
+    },
+    {
+      "epoch": 0.0074482758620689656,
+      "grad_norm": 11.509404182434082,
+      "learning_rate": 1.8436090225563912e-05,
+      "loss": 0.37604063749313354,
+      "num_input_tokens_seen": 53454,
+      "step": 162,
+      "train_runtime": 390.2515,
+      "train_tokens_per_second": 136.973
+    },
+    {
+      "epoch": 0.007494252873563218,
+      "grad_norm": 8.486335754394531,
+      "learning_rate": 1.842606516290727e-05,
+      "loss": 0.5968654155731201,
+      "num_input_tokens_seen": 53916,
+      "step": 163,
+      "train_runtime": 392.2164,
+      "train_tokens_per_second": 137.465
+    },
+    {
+      "epoch": 0.007540229885057472,
+      "grad_norm": 9.746296882629395,
+      "learning_rate": 1.8416040100250626e-05,
+      "loss": 0.5125172734260559,
+      "num_input_tokens_seen": 54222,
+      "step": 164,
+      "train_runtime": 394.1474,
+      "train_tokens_per_second": 137.568
+    },
+    {
+      "epoch": 0.007586206896551724,
+      "grad_norm": 8.613889694213867,
+      "learning_rate": 1.8406015037593986e-05,
+      "loss": 0.5511026382446289,
+      "num_input_tokens_seen": 54706,
+      "step": 165,
+      "train_runtime": 396.1306,
+      "train_tokens_per_second": 138.101
+    },
+    {
+      "epoch": 0.007632183908045977,
+      "grad_norm": 8.980238914489746,
+      "learning_rate": 1.8395989974937346e-05,
+      "loss": 0.511992335319519,
+      "num_input_tokens_seen": 54992,
+      "step": 166,
+      "train_runtime": 398.0508,
+      "train_tokens_per_second": 138.153
+    },
+    {
+      "epoch": 0.00767816091954023,
+      "grad_norm": 7.981934070587158,
+      "learning_rate": 1.8385964912280703e-05,
+      "loss": 0.4814477264881134,
+      "num_input_tokens_seen": 55798,
+      "step": 167,
+      "train_runtime": 400.1758,
+      "train_tokens_per_second": 139.434
+    },
+    {
+      "epoch": 0.007724137931034483,
+      "grad_norm": 9.586648941040039,
+      "learning_rate": 1.8375939849624063e-05,
+      "loss": 0.4873974323272705,
+      "num_input_tokens_seen": 56136,
+      "step": 168,
+      "train_runtime": 402.1064,
+      "train_tokens_per_second": 139.605
+    },
+    {
+      "epoch": 0.0077701149425287355,
+      "grad_norm": 8.589614868164062,
+      "learning_rate": 1.836591478696742e-05,
+      "loss": 0.5130251049995422,
+      "num_input_tokens_seen": 56516,
+      "step": 169,
+      "train_runtime": 404.0908,
+      "train_tokens_per_second": 139.86
+    },
+    {
+      "epoch": 0.007816091954022988,
+      "grad_norm": 10.078131675720215,
+      "learning_rate": 1.8355889724310777e-05,
+      "loss": 0.5906585454940796,
+      "num_input_tokens_seen": 56804,
+      "step": 170,
+      "train_runtime": 406.0306,
+      "train_tokens_per_second": 139.901
+    },
+    {
+      "epoch": 0.00786206896551724,
+      "grad_norm": 9.921672821044922,
+      "learning_rate": 1.8345864661654137e-05,
+      "loss": 0.4799719452857971,
+      "num_input_tokens_seen": 57094,
+      "step": 171,
+      "train_runtime": 407.9674,
+      "train_tokens_per_second": 139.947
+    },
+    {
+      "epoch": 0.007908045977011495,
+      "grad_norm": 7.844490051269531,
+      "learning_rate": 1.8335839598997497e-05,
+      "loss": 0.597183346748352,
+      "num_input_tokens_seen": 57430,
+      "step": 172,
+      "train_runtime": 409.9136,
+      "train_tokens_per_second": 140.103
+    },
+    {
+      "epoch": 0.007954022988505748,
+      "grad_norm": 9.238640785217285,
+      "learning_rate": 1.8325814536340854e-05,
+      "loss": 0.4416566491127014,
+      "num_input_tokens_seen": 57660,
+      "step": 173,
+      "train_runtime": 411.8262,
+      "train_tokens_per_second": 140.011
+    },
+    {
+      "epoch": 0.008,
+      "grad_norm": 8.418259620666504,
+      "learning_rate": 1.831578947368421e-05,
+      "loss": 0.4849053621292114,
+      "num_input_tokens_seen": 57982,
+      "step": 174,
+      "train_runtime": 413.7912,
+      "train_tokens_per_second": 140.124
+    },
+    {
+      "epoch": 0.008045977011494253,
+      "grad_norm": 7.610781669616699,
+      "learning_rate": 1.830576441102757e-05,
+      "loss": 0.41063031554222107,
+      "num_input_tokens_seen": 58264,
+      "step": 175,
+      "train_runtime": 415.766,
+      "train_tokens_per_second": 140.137
+    },
+    {
+      "epoch": 0.008091954022988505,
+      "grad_norm": 7.702761650085449,
+      "learning_rate": 1.8295739348370928e-05,
+      "loss": 0.3769613802433014,
+      "num_input_tokens_seen": 58524,
+      "step": 176,
+      "train_runtime": 417.77,
+      "train_tokens_per_second": 140.087
+    },
+    {
+      "epoch": 0.008137931034482758,
+      "grad_norm": 9.193915367126465,
+      "learning_rate": 1.8285714285714288e-05,
+      "loss": 0.5968765616416931,
+      "num_input_tokens_seen": 58908,
+      "step": 177,
+      "train_runtime": 419.7369,
+      "train_tokens_per_second": 140.345
+    },
+    {
+      "epoch": 0.008183908045977012,
+      "grad_norm": 7.960973739624023,
+      "learning_rate": 1.8275689223057648e-05,
+      "loss": 0.6170963644981384,
+      "num_input_tokens_seen": 59560,
+      "step": 178,
+      "train_runtime": 421.7529,
+      "train_tokens_per_second": 141.22
+    },
+    {
+      "epoch": 0.008229885057471265,
+      "grad_norm": 7.9756035804748535,
+      "learning_rate": 1.8265664160401005e-05,
+      "loss": 0.47514936327934265,
+      "num_input_tokens_seen": 59860,
+      "step": 179,
+      "train_runtime": 423.7014,
+      "train_tokens_per_second": 141.279
+    },
+    {
+      "epoch": 0.008275862068965517,
+      "grad_norm": 7.550826549530029,
+      "learning_rate": 1.825563909774436e-05,
+      "loss": 0.4637778699398041,
+      "num_input_tokens_seen": 60178,
+      "step": 180,
+      "train_runtime": 425.6629,
+      "train_tokens_per_second": 141.375
+    },
+    {
+      "epoch": 0.00832183908045977,
+      "grad_norm": 8.703726768493652,
+      "learning_rate": 1.824561403508772e-05,
+      "loss": 0.41379302740097046,
+      "num_input_tokens_seen": 60442,
+      "step": 181,
+      "train_runtime": 440.0628,
+      "train_tokens_per_second": 137.349
+    },
+    {
+      "epoch": 0.008367816091954023,
+      "grad_norm": 8.947075843811035,
+      "learning_rate": 1.823558897243108e-05,
+      "loss": 0.46399661898612976,
+      "num_input_tokens_seen": 60822,
+      "step": 182,
+      "train_runtime": 442.0168,
+      "train_tokens_per_second": 137.601
+    },
+    {
+      "epoch": 0.008413793103448275,
+      "grad_norm": 9.358521461486816,
+      "learning_rate": 1.822556390977444e-05,
+      "loss": 0.6975663900375366,
+      "num_input_tokens_seen": 61202,
+      "step": 183,
+      "train_runtime": 443.9751,
+      "train_tokens_per_second": 137.85
+    },
+    {
+      "epoch": 0.00845977011494253,
+      "grad_norm": 6.9549689292907715,
+      "learning_rate": 1.8215538847117796e-05,
+      "loss": 0.32578420639038086,
+      "num_input_tokens_seen": 61468,
+      "step": 184,
+      "train_runtime": 445.9064,
+      "train_tokens_per_second": 137.85
+    },
+    {
+      "epoch": 0.008505747126436782,
+      "grad_norm": 8.928411483764648,
+      "learning_rate": 1.8205513784461156e-05,
+      "loss": 0.4039687514305115,
+      "num_input_tokens_seen": 61766,
+      "step": 185,
+      "train_runtime": 447.8816,
+      "train_tokens_per_second": 137.907
+    },
+    {
+      "epoch": 0.008551724137931035,
+      "grad_norm": 8.656428337097168,
+      "learning_rate": 1.8195488721804512e-05,
+      "loss": 0.5474002361297607,
+      "num_input_tokens_seen": 62068,
+      "step": 186,
+      "train_runtime": 449.8232,
+      "train_tokens_per_second": 137.983
+    },
+    {
+      "epoch": 0.008597701149425287,
+      "grad_norm": 7.867074012756348,
+      "learning_rate": 1.818546365914787e-05,
+      "loss": 0.2982150912284851,
+      "num_input_tokens_seen": 62318,
+      "step": 187,
+      "train_runtime": 451.7609,
+      "train_tokens_per_second": 137.945
+    },
+    {
+      "epoch": 0.00864367816091954,
+      "grad_norm": 7.8986287117004395,
+      "learning_rate": 1.817543859649123e-05,
+      "loss": 0.3921367824077606,
+      "num_input_tokens_seen": 62572,
+      "step": 188,
+      "train_runtime": 453.6977,
+      "train_tokens_per_second": 137.916
+    },
+    {
+      "epoch": 0.008689655172413793,
+      "grad_norm": 7.843879699707031,
+      "learning_rate": 1.816541353383459e-05,
+      "loss": 0.36684906482696533,
+      "num_input_tokens_seen": 62842,
+      "step": 189,
+      "train_runtime": 455.6408,
+      "train_tokens_per_second": 137.92
+    },
+    {
+      "epoch": 0.008735632183908045,
+      "grad_norm": 10.188214302062988,
+      "learning_rate": 1.8155388471177946e-05,
+      "loss": 0.6560608744621277,
+      "num_input_tokens_seen": 63174,
+      "step": 190,
+      "train_runtime": 457.5784,
+      "train_tokens_per_second": 138.062
+    },
+    {
+      "epoch": 0.0087816091954023,
+      "grad_norm": 7.0697712898254395,
+      "learning_rate": 1.8145363408521303e-05,
+      "loss": 0.2936539053916931,
+      "num_input_tokens_seen": 63402,
+      "step": 191,
+      "train_runtime": 459.4785,
+      "train_tokens_per_second": 137.987
+    },
+    {
+      "epoch": 0.008827586206896552,
+      "grad_norm": 9.822193145751953,
+      "learning_rate": 1.8135338345864663e-05,
+      "loss": 0.5231074690818787,
+      "num_input_tokens_seen": 63744,
+      "step": 192,
+      "train_runtime": 461.4404,
+      "train_tokens_per_second": 138.141
+    },
+    {
+      "epoch": 0.008873563218390805,
+      "grad_norm": 8.857881546020508,
+      "learning_rate": 1.812531328320802e-05,
+      "loss": 0.5226064920425415,
+      "num_input_tokens_seen": 64048,
+      "step": 193,
+      "train_runtime": 463.4527,
+      "train_tokens_per_second": 138.197
+    },
+    {
+      "epoch": 0.008919540229885057,
+      "grad_norm": 8.514666557312012,
+      "learning_rate": 1.811528822055138e-05,
+      "loss": 0.370896577835083,
+      "num_input_tokens_seen": 64300,
+      "step": 194,
+      "train_runtime": 465.41,
+      "train_tokens_per_second": 138.158
+    },
+    {
+      "epoch": 0.00896551724137931,
+      "grad_norm": 9.272273063659668,
+      "learning_rate": 1.810526315789474e-05,
+      "loss": 0.5460460186004639,
+      "num_input_tokens_seen": 64596,
+      "step": 195,
+      "train_runtime": 467.3302,
+      "train_tokens_per_second": 138.223
+    },
+    {
+      "epoch": 0.009011494252873562,
+      "grad_norm": 8.173891067504883,
+      "learning_rate": 1.8095238095238097e-05,
+      "loss": 0.43280303478240967,
+      "num_input_tokens_seen": 64906,
+      "step": 196,
+      "train_runtime": 469.2914,
+      "train_tokens_per_second": 138.306
+    },
+    {
+      "epoch": 0.009057471264367817,
+      "grad_norm": 9.298201560974121,
+      "learning_rate": 1.8085213032581454e-05,
+      "loss": 0.43618613481521606,
+      "num_input_tokens_seen": 65220,
+      "step": 197,
+      "train_runtime": 471.2687,
+      "train_tokens_per_second": 138.392
+    },
+    {
+      "epoch": 0.00910344827586207,
+      "grad_norm": 7.345490455627441,
+      "learning_rate": 1.8075187969924814e-05,
+      "loss": 0.5126155614852905,
+      "num_input_tokens_seen": 65574,
+      "step": 198,
+      "train_runtime": 473.2274,
+      "train_tokens_per_second": 138.568
+    },
+    {
+      "epoch": 0.009149425287356322,
+      "grad_norm": 7.111669063568115,
+      "learning_rate": 1.806516290726817e-05,
+      "loss": 0.2610500752925873,
+      "num_input_tokens_seen": 65814,
+      "step": 199,
+      "train_runtime": 475.1321,
+      "train_tokens_per_second": 138.517
+    },
+    {
+      "epoch": 0.009195402298850575,
+      "grad_norm": 8.334724426269531,
+      "learning_rate": 1.805513784461153e-05,
+      "loss": 0.4848451316356659,
+      "num_input_tokens_seen": 66088,
+      "step": 200,
+      "train_runtime": 477.042,
+      "train_tokens_per_second": 138.537
+    },
+    {
+      "epoch": 0.009195402298850575,
+      "eval_loss": 1.8940600156784058,
+      "eval_runtime": 54.1829,
+      "eval_samples_per_second": 18.456,
+      "eval_steps_per_second": 9.228,
+      "num_input_tokens_seen": 66088,
+      "step": 200
+    },
+    {
+      "epoch": 0.009241379310344827,
+      "grad_norm": 9.731633186340332,
+      "learning_rate": 1.8045112781954888e-05,
+      "loss": 0.5664269924163818,
+      "num_input_tokens_seen": 66364,
+      "step": 201,
+      "train_runtime": 533.1769,
+      "train_tokens_per_second": 124.469
+    },
+    {
+      "epoch": 0.00928735632183908,
+      "grad_norm": 10.086454391479492,
+      "learning_rate": 1.8035087719298248e-05,
+      "loss": 0.5216547250747681,
+      "num_input_tokens_seen": 66678,
+      "step": 202,
+      "train_runtime": 535.0951,
+      "train_tokens_per_second": 124.61
+    },
+    {
+      "epoch": 0.009333333333333334,
+      "grad_norm": 9.126751899719238,
+      "learning_rate": 1.8025062656641605e-05,
+      "loss": 0.6206774115562439,
+      "num_input_tokens_seen": 66944,
+      "step": 203,
+      "train_runtime": 537.0143,
+      "train_tokens_per_second": 124.66
+    },
+    {
+      "epoch": 0.009379310344827587,
+      "grad_norm": 7.979254722595215,
+      "learning_rate": 1.8015037593984962e-05,
+      "loss": 0.39576491713523865,
+      "num_input_tokens_seen": 67204,
+      "step": 204,
+      "train_runtime": 538.9277,
+      "train_tokens_per_second": 124.699
+    },
+    {
+      "epoch": 0.00942528735632184,
+      "grad_norm": 9.559494972229004,
+      "learning_rate": 1.8005012531328322e-05,
+      "loss": 0.5241124629974365,
+      "num_input_tokens_seen": 67520,
+      "step": 205,
+      "train_runtime": 540.8486,
+      "train_tokens_per_second": 124.841
+    },
+    {
+      "epoch": 0.009471264367816092,
+      "grad_norm": 9.92201042175293,
+      "learning_rate": 1.7994987468671682e-05,
+      "loss": 0.5017831921577454,
+      "num_input_tokens_seen": 67792,
+      "step": 206,
+      "train_runtime": 542.751,
+      "train_tokens_per_second": 124.904
+    },
+    {
+      "epoch": 0.009517241379310344,
+      "grad_norm": 8.381772994995117,
+      "learning_rate": 1.798496240601504e-05,
+      "loss": 0.4413483440876007,
+      "num_input_tokens_seen": 68060,
+      "step": 207,
+      "train_runtime": 544.6632,
+      "train_tokens_per_second": 124.958
+    },
+    {
+      "epoch": 0.009563218390804597,
+      "grad_norm": 9.174842834472656,
+      "learning_rate": 1.79749373433584e-05,
+      "loss": 0.4749239683151245,
+      "num_input_tokens_seen": 68334,
+      "step": 208,
+      "train_runtime": 546.5773,
+      "train_tokens_per_second": 125.022
+    },
+    {
+      "epoch": 0.009609195402298851,
+      "grad_norm": 9.39853286743164,
+      "learning_rate": 1.7964912280701756e-05,
+      "loss": 0.45542603731155396,
+      "num_input_tokens_seen": 68588,
+      "step": 209,
+      "train_runtime": 548.4819,
+      "train_tokens_per_second": 125.051
+    },
+    {
+      "epoch": 0.009655172413793104,
+      "grad_norm": 9.451595306396484,
+      "learning_rate": 1.7954887218045113e-05,
+      "loss": 0.6160299777984619,
+      "num_input_tokens_seen": 68944,
+      "step": 210,
+      "train_runtime": 550.3957,
+      "train_tokens_per_second": 125.263
+    },
+    {
+      "epoch": 0.009701149425287357,
+      "grad_norm": 11.262086868286133,
+      "learning_rate": 1.7944862155388473e-05,
+      "loss": 0.5597932934761047,
+      "num_input_tokens_seen": 69172,
+      "step": 211,
+      "train_runtime": 565.0919,
+      "train_tokens_per_second": 122.408
+    },
+    {
+      "epoch": 0.00974712643678161,
+      "grad_norm": 11.364030838012695,
+      "learning_rate": 1.7934837092731833e-05,
+      "loss": 0.6930623650550842,
+      "num_input_tokens_seen": 69440,
+      "step": 212,
+      "train_runtime": 567.0488,
+      "train_tokens_per_second": 122.459
+    },
+    {
+      "epoch": 0.009793103448275862,
+      "grad_norm": 12.371257781982422,
+      "learning_rate": 1.792481203007519e-05,
+      "loss": 0.5638558268547058,
+      "num_input_tokens_seen": 69754,
+      "step": 213,
+      "train_runtime": 568.9985,
+      "train_tokens_per_second": 122.591
+    },
+    {
+      "epoch": 0.009839080459770114,
+      "grad_norm": 10.8510160446167,
+      "learning_rate": 1.7914786967418547e-05,
+      "loss": 0.567751407623291,
+      "num_input_tokens_seen": 70054,
+      "step": 214,
+      "train_runtime": 570.9281,
+      "train_tokens_per_second": 122.702
+    },
+    {
+      "epoch": 0.009885057471264367,
+      "grad_norm": 8.684098243713379,
+      "learning_rate": 1.7904761904761907e-05,
+      "loss": 0.5753757357597351,
+      "num_input_tokens_seen": 70450,
+      "step": 215,
+      "train_runtime": 572.8701,
+      "train_tokens_per_second": 122.977
+    },
+    {
+      "epoch": 0.009931034482758621,
+      "grad_norm": 9.463613510131836,
+      "learning_rate": 1.7894736842105264e-05,
+      "loss": 0.5026508569717407,
+      "num_input_tokens_seen": 70746,
+      "step": 216,
+      "train_runtime": 574.8099,
+      "train_tokens_per_second": 123.077
+    },
+    {
+      "epoch": 0.009977011494252874,
+      "grad_norm": 9.900351524353027,
+      "learning_rate": 1.7884711779448624e-05,
+      "loss": 0.648265540599823,
+      "num_input_tokens_seen": 71298,
+      "step": 217,
+      "train_runtime": 576.7914,
+      "train_tokens_per_second": 123.611
+    },
+    {
+      "epoch": 0.010022988505747127,
+      "grad_norm": 8.408405303955078,
+      "learning_rate": 1.7874686716791984e-05,
+      "loss": 0.5788682103157043,
+      "num_input_tokens_seen": 71618,
+      "step": 218,
+      "train_runtime": 578.7088,
+      "train_tokens_per_second": 123.755
+    },
+    {
+      "epoch": 0.010068965517241379,
+      "grad_norm": 8.421172142028809,
+      "learning_rate": 1.786466165413534e-05,
+      "loss": 0.5744531154632568,
+      "num_input_tokens_seen": 71920,
+      "step": 219,
+      "train_runtime": 580.6368,
+      "train_tokens_per_second": 123.864
+    },
+    {
+      "epoch": 0.010114942528735632,
+      "grad_norm": 7.934989929199219,
+      "learning_rate": 1.7854636591478698e-05,
+      "loss": 0.6625332832336426,
+      "num_input_tokens_seen": 72472,
+      "step": 220,
+      "train_runtime": 582.6318,
+      "train_tokens_per_second": 124.387
+    },
+    {
+      "epoch": 0.010160919540229884,
+      "grad_norm": 8.542694091796875,
+      "learning_rate": 1.7844611528822054e-05,
+      "loss": 0.580016553401947,
+      "num_input_tokens_seen": 72848,
+      "step": 221,
+      "train_runtime": 584.5837,
+      "train_tokens_per_second": 124.615
+    },
+    {
+      "epoch": 0.010206896551724139,
+      "grad_norm": 8.42661190032959,
+      "learning_rate": 1.7834586466165415e-05,
+      "loss": 0.5382593274116516,
+      "num_input_tokens_seen": 73170,
+      "step": 222,
+      "train_runtime": 586.531,
+      "train_tokens_per_second": 124.75
+    },
+    {
+      "epoch": 0.010252873563218391,
+      "grad_norm": 8.81701374053955,
+      "learning_rate": 1.7824561403508775e-05,
+      "loss": 0.41666749119758606,
+      "num_input_tokens_seen": 73432,
+      "step": 223,
+      "train_runtime": 588.6068,
+      "train_tokens_per_second": 124.756
+    },
+    {
+      "epoch": 0.010298850574712644,
+      "grad_norm": 9.079622268676758,
+      "learning_rate": 1.781453634085213e-05,
+      "loss": 0.6324032545089722,
+      "num_input_tokens_seen": 73754,
+      "step": 224,
+      "train_runtime": 590.5476,
+      "train_tokens_per_second": 124.891
+    },
+    {
+      "epoch": 0.010344827586206896,
+      "grad_norm": 9.136917114257812,
+      "learning_rate": 1.7804511278195492e-05,
+      "loss": 0.6062464118003845,
+      "num_input_tokens_seen": 74080,
+      "step": 225,
+      "train_runtime": 592.4978,
+      "train_tokens_per_second": 125.03
+    },
+    {
+      "epoch": 0.010390804597701149,
+      "grad_norm": 7.6709980964660645,
+      "learning_rate": 1.779448621553885e-05,
+      "loss": 0.44233664870262146,
+      "num_input_tokens_seen": 74330,
+      "step": 226,
+      "train_runtime": 594.437,
+      "train_tokens_per_second": 125.043
+    },
+    {
+      "epoch": 0.010436781609195402,
+      "grad_norm": 8.36764907836914,
+      "learning_rate": 1.7784461152882205e-05,
+      "loss": 0.5167356133460999,
+      "num_input_tokens_seen": 74582,
+      "step": 227,
+      "train_runtime": 596.4444,
+      "train_tokens_per_second": 125.044
+    },
+    {
+      "epoch": 0.010482758620689656,
+      "grad_norm": 6.792186260223389,
+      "learning_rate": 1.7774436090225566e-05,
+      "loss": 0.4979870915412903,
+      "num_input_tokens_seen": 75022,
+      "step": 228,
+      "train_runtime": 598.4481,
+      "train_tokens_per_second": 125.361
+    },
+    {
+      "epoch": 0.010528735632183909,
+      "grad_norm": 9.433555603027344,
+      "learning_rate": 1.7764411027568926e-05,
+      "loss": 0.5935546159744263,
+      "num_input_tokens_seen": 75376,
+      "step": 229,
+      "train_runtime": 600.5684,
+      "train_tokens_per_second": 125.508
+    },
+    {
+      "epoch": 0.010574712643678161,
+      "grad_norm": 8.143898010253906,
+      "learning_rate": 1.7754385964912283e-05,
+      "loss": 0.4781757593154907,
+      "num_input_tokens_seen": 75634,
+      "step": 230,
+      "train_runtime": 602.5668,
+      "train_tokens_per_second": 125.52
+    },
+    {
+      "epoch": 0.010620689655172414,
+      "grad_norm": 7.581504821777344,
+      "learning_rate": 1.774436090225564e-05,
+      "loss": 0.39986687898635864,
+      "num_input_tokens_seen": 75914,
+      "step": 231,
+      "train_runtime": 604.489,
+      "train_tokens_per_second": 125.584
+    },
+    {
+      "epoch": 0.010666666666666666,
+      "grad_norm": 8.80359935760498,
+      "learning_rate": 1.7734335839599e-05,
+      "loss": 0.4914165139198303,
+      "num_input_tokens_seen": 76178,
+      "step": 232,
+      "train_runtime": 606.4381,
+      "train_tokens_per_second": 125.615
+    },
+    {
+      "epoch": 0.010712643678160919,
+      "grad_norm": 8.052352905273438,
+      "learning_rate": 1.7724310776942356e-05,
+      "loss": 0.5605274438858032,
+      "num_input_tokens_seen": 76612,
+      "step": 233,
+      "train_runtime": 608.4309,
+      "train_tokens_per_second": 125.917
+    },
+    {
+      "epoch": 0.010758620689655173,
+      "grad_norm": 7.698636531829834,
+      "learning_rate": 1.7714285714285717e-05,
+      "loss": 0.31394222378730774,
+      "num_input_tokens_seen": 76950,
+      "step": 234,
+      "train_runtime": 610.4578,
+      "train_tokens_per_second": 126.053
+    },
+    {
+      "epoch": 0.010804597701149426,
+      "grad_norm": 8.41732120513916,
+      "learning_rate": 1.7704260651629073e-05,
+      "loss": 0.5563385486602783,
+      "num_input_tokens_seen": 77250,
+      "step": 235,
+      "train_runtime": 612.4796,
+      "train_tokens_per_second": 126.127
+    },
+    {
+      "epoch": 0.010850574712643678,
+      "grad_norm": 9.47237777709961,
+      "learning_rate": 1.7694235588972433e-05,
+      "loss": 0.6081884503364563,
+      "num_input_tokens_seen": 77526,
+      "step": 236,
+      "train_runtime": 614.459,
+      "train_tokens_per_second": 126.17
+    },
+    {
+      "epoch": 0.010896551724137931,
+      "grad_norm": 8.935263633728027,
+      "learning_rate": 1.768421052631579e-05,
+      "loss": 0.6447509527206421,
+      "num_input_tokens_seen": 77850,
+      "step": 237,
+      "train_runtime": 616.4103,
+      "train_tokens_per_second": 126.296
+    },
+    {
+      "epoch": 0.010942528735632184,
+      "grad_norm": 10.19480037689209,
+      "learning_rate": 1.767418546365915e-05,
+      "loss": 0.47084376215934753,
+      "num_input_tokens_seen": 78160,
+      "step": 238,
+      "train_runtime": 618.3501,
+      "train_tokens_per_second": 126.401
+    },
+    {
+      "epoch": 0.010988505747126436,
+      "grad_norm": 7.387281894683838,
+      "learning_rate": 1.7664160401002507e-05,
+      "loss": 0.41510701179504395,
+      "num_input_tokens_seen": 78438,
+      "step": 239,
+      "train_runtime": 620.2698,
+      "train_tokens_per_second": 126.458
+    },
+    {
+      "epoch": 0.011034482758620689,
+      "grad_norm": 8.466607093811035,
+      "learning_rate": 1.7654135338345867e-05,
+      "loss": 0.4125990867614746,
+      "num_input_tokens_seen": 78710,
+      "step": 240,
+      "train_runtime": 622.196,
+      "train_tokens_per_second": 126.504
+    },
+    {
+      "epoch": 0.011080459770114943,
+      "grad_norm": 9.628037452697754,
+      "learning_rate": 1.7644110275689224e-05,
+      "loss": 0.5155237317085266,
+      "num_input_tokens_seen": 79176,
+      "step": 241,
+      "train_runtime": 637.8479,
+      "train_tokens_per_second": 124.13
+    },
+    {
+      "epoch": 0.011126436781609196,
+      "grad_norm": 7.532957077026367,
+      "learning_rate": 1.7634085213032584e-05,
+      "loss": 0.5341253280639648,
+      "num_input_tokens_seen": 79764,
+      "step": 242,
+      "train_runtime": 639.8913,
+      "train_tokens_per_second": 124.652
+    },
+    {
+      "epoch": 0.011172413793103448,
+      "grad_norm": 7.834677696228027,
+      "learning_rate": 1.762406015037594e-05,
+      "loss": 0.5943319797515869,
+      "num_input_tokens_seen": 80424,
+      "step": 243,
+      "train_runtime": 642.1863,
+      "train_tokens_per_second": 125.235
+    },
+    {
+      "epoch": 0.011218390804597701,
+      "grad_norm": 8.330830574035645,
+      "learning_rate": 1.7614035087719298e-05,
+      "loss": 0.6359134912490845,
+      "num_input_tokens_seen": 80886,
+      "step": 244,
+      "train_runtime": 644.3059,
+      "train_tokens_per_second": 125.54
+    },
+    {
+      "epoch": 0.011264367816091954,
+      "grad_norm": 7.60586404800415,
+      "learning_rate": 1.7604010025062658e-05,
+      "loss": 0.39419102668762207,
+      "num_input_tokens_seen": 81144,
+      "step": 245,
+      "train_runtime": 646.2392,
+      "train_tokens_per_second": 125.563
+    },
+    {
+      "epoch": 0.011310344827586206,
+      "grad_norm": 8.247746467590332,
+      "learning_rate": 1.7593984962406015e-05,
+      "loss": 0.38935500383377075,
+      "num_input_tokens_seen": 81372,
+      "step": 246,
+      "train_runtime": 648.145,
+      "train_tokens_per_second": 125.546
+    },
+    {
+      "epoch": 0.01135632183908046,
+      "grad_norm": 10.329381942749023,
+      "learning_rate": 1.7583959899749375e-05,
+      "loss": 0.47521886229515076,
+      "num_input_tokens_seen": 81656,
+      "step": 247,
+      "train_runtime": 650.0725,
+      "train_tokens_per_second": 125.611
+    },
+    {
+      "epoch": 0.011402298850574713,
+      "grad_norm": 8.573349952697754,
+      "learning_rate": 1.7573934837092735e-05,
+      "loss": 0.5382678508758545,
+      "num_input_tokens_seen": 81930,
+      "step": 248,
+      "train_runtime": 652.0055,
+      "train_tokens_per_second": 125.658
+    },
+    {
+      "epoch": 0.011448275862068966,
+      "grad_norm": 9.41952133178711,
+      "learning_rate": 1.7563909774436092e-05,
+      "loss": 0.4593239426612854,
+      "num_input_tokens_seen": 82202,
+      "step": 249,
+      "train_runtime": 653.9525,
+      "train_tokens_per_second": 125.7
+    },
+    {
+      "epoch": 0.011494252873563218,
+      "grad_norm": 7.771377086639404,
+      "learning_rate": 1.755388471177945e-05,
+      "loss": 0.4767267107963562,
+      "num_input_tokens_seen": 82492,
+      "step": 250,
+      "train_runtime": 655.9119,
+      "train_tokens_per_second": 125.767
+    },
+    {
+      "epoch": 0.01154022988505747,
+      "grad_norm": 8.981367111206055,
+      "learning_rate": 1.754385964912281e-05,
+      "loss": 0.5896744132041931,
+      "num_input_tokens_seen": 82810,
+      "step": 251,
+      "train_runtime": 657.9051,
+      "train_tokens_per_second": 125.869
+    },
+    {
+      "epoch": 0.011586206896551723,
+      "grad_norm": 8.728583335876465,
+      "learning_rate": 1.7533834586466166e-05,
+      "loss": 0.5037809014320374,
+      "num_input_tokens_seen": 83116,
+      "step": 252,
+      "train_runtime": 659.8608,
+      "train_tokens_per_second": 125.96
+    },
+    {
+      "epoch": 0.011632183908045978,
+      "grad_norm": 7.11443567276001,
+      "learning_rate": 1.7523809523809526e-05,
+      "loss": 0.4327583909034729,
+      "num_input_tokens_seen": 83444,
+      "step": 253,
+      "train_runtime": 661.7819,
+      "train_tokens_per_second": 126.09
+    },
+    {
+      "epoch": 0.01167816091954023,
+      "grad_norm": 8.682472229003906,
+      "learning_rate": 1.7513784461152883e-05,
+      "loss": 0.44737380743026733,
+      "num_input_tokens_seen": 83728,
+      "step": 254,
+      "train_runtime": 663.7561,
+      "train_tokens_per_second": 126.143
+    },
+    {
+      "epoch": 0.011724137931034483,
+      "grad_norm": 8.399706840515137,
+      "learning_rate": 1.7503759398496243e-05,
+      "loss": 0.5722436904907227,
+      "num_input_tokens_seen": 84356,
+      "step": 255,
+      "train_runtime": 665.8361,
+      "train_tokens_per_second": 126.692
+    },
+    {
+      "epoch": 0.011770114942528736,
+      "grad_norm": 7.891780853271484,
+      "learning_rate": 1.74937343358396e-05,
+      "loss": 0.42720746994018555,
+      "num_input_tokens_seen": 84672,
+      "step": 256,
+      "train_runtime": 667.7623,
+      "train_tokens_per_second": 126.8
+    },
+    {
+      "epoch": 0.011816091954022988,
+      "grad_norm": 9.670726776123047,
+      "learning_rate": 1.7483709273182957e-05,
+      "loss": 0.4248063862323761,
+      "num_input_tokens_seen": 84942,
+      "step": 257,
+      "train_runtime": 669.7404,
+      "train_tokens_per_second": 126.828
+    },
+    {
+      "epoch": 0.01186206896551724,
+      "grad_norm": 7.518779277801514,
+      "learning_rate": 1.7473684210526317e-05,
+      "loss": 0.44924628734588623,
+      "num_input_tokens_seen": 85242,
+      "step": 258,
+      "train_runtime": 671.6717,
+      "train_tokens_per_second": 126.91
+    },
+    {
+      "epoch": 0.011908045977011495,
+      "grad_norm": 7.856861591339111,
+      "learning_rate": 1.7463659147869677e-05,
+      "loss": 0.43991750478744507,
+      "num_input_tokens_seen": 85640,
+      "step": 259,
+      "train_runtime": 673.6511,
+      "train_tokens_per_second": 127.128
+    },
+    {
+      "epoch": 0.011954022988505748,
+      "grad_norm": 8.80428409576416,
+      "learning_rate": 1.7453634085213034e-05,
+      "loss": 0.5700347423553467,
+      "num_input_tokens_seen": 86002,
+      "step": 260,
+      "train_runtime": 675.5867,
+      "train_tokens_per_second": 127.3
+    },
+    {
+      "epoch": 0.012,
+      "grad_norm": 7.360157012939453,
+      "learning_rate": 1.744360902255639e-05,
+      "loss": 0.37258443236351013,
+      "num_input_tokens_seen": 86264,
+      "step": 261,
+      "train_runtime": 677.6028,
+      "train_tokens_per_second": 127.308
+    },
+    {
+      "epoch": 0.012045977011494253,
+      "grad_norm": 8.775689125061035,
+      "learning_rate": 1.743358395989975e-05,
+      "loss": 0.5349915623664856,
+      "num_input_tokens_seen": 86546,
+      "step": 262,
+      "train_runtime": 679.5822,
+      "train_tokens_per_second": 127.352
+    },
+    {
+      "epoch": 0.012091954022988505,
+      "grad_norm": 7.555972576141357,
+      "learning_rate": 1.7423558897243108e-05,
+      "loss": 0.438320517539978,
+      "num_input_tokens_seen": 86822,
+      "step": 263,
+      "train_runtime": 681.6542,
+      "train_tokens_per_second": 127.37
+    },
+    {
+      "epoch": 0.012137931034482758,
+      "grad_norm": 7.634513854980469,
+      "learning_rate": 1.7413533834586468e-05,
+      "loss": 0.45846620202064514,
+      "num_input_tokens_seen": 87164,
+      "step": 264,
+      "train_runtime": 683.6555,
+      "train_tokens_per_second": 127.497
+    },
+    {
+      "epoch": 0.01218390804597701,
+      "grad_norm": 7.4254350662231445,
+      "learning_rate": 1.7403508771929828e-05,
+      "loss": 0.4867999255657196,
+      "num_input_tokens_seen": 87676,
+      "step": 265,
+      "train_runtime": 685.6908,
+      "train_tokens_per_second": 127.865
+    },
+    {
+      "epoch": 0.012229885057471265,
+      "grad_norm": 7.971969127655029,
+      "learning_rate": 1.7393483709273185e-05,
+      "loss": 0.45510679483413696,
+      "num_input_tokens_seen": 87938,
+      "step": 266,
+      "train_runtime": 687.6036,
+      "train_tokens_per_second": 127.891
+    },
+    {
+      "epoch": 0.012275862068965518,
+      "grad_norm": 9.114470481872559,
+      "learning_rate": 1.738345864661654e-05,
+      "loss": 0.3985937237739563,
+      "num_input_tokens_seen": 88246,
+      "step": 267,
+      "train_runtime": 689.5433,
+      "train_tokens_per_second": 127.977
+    },
+    {
+      "epoch": 0.01232183908045977,
+      "grad_norm": 7.794116020202637,
+      "learning_rate": 1.7373433583959902e-05,
+      "loss": 0.5491940379142761,
+      "num_input_tokens_seen": 88640,
+      "step": 268,
+      "train_runtime": 691.516,
+      "train_tokens_per_second": 128.182
+    },
+    {
+      "epoch": 0.012367816091954023,
+      "grad_norm": 6.931571960449219,
+      "learning_rate": 1.736340852130326e-05,
+      "loss": 0.476431667804718,
+      "num_input_tokens_seen": 88986,
+      "step": 269,
+      "train_runtime": 693.4528,
+      "train_tokens_per_second": 128.323
+    },
+    {
+      "epoch": 0.012413793103448275,
+      "grad_norm": 9.618606567382812,
+      "learning_rate": 1.735338345864662e-05,
+      "loss": 0.5247324705123901,
+      "num_input_tokens_seen": 89268,
+      "step": 270,
+      "train_runtime": 695.4365,
+      "train_tokens_per_second": 128.363
+    },
+    {
+      "epoch": 0.012459770114942528,
+      "grad_norm": 5.826930999755859,
+      "learning_rate": 1.7343358395989975e-05,
+      "loss": 0.317282497882843,
+      "num_input_tokens_seen": 89584,
+      "step": 271,
+      "train_runtime": 711.4524,
+      "train_tokens_per_second": 125.917
+    },
+    {
+      "epoch": 0.012505747126436782,
+      "grad_norm": 8.20704460144043,
+      "learning_rate": 1.7333333333333336e-05,
+      "loss": 0.6090211272239685,
+      "num_input_tokens_seen": 89970,
+      "step": 272,
+      "train_runtime": 713.6346,
+      "train_tokens_per_second": 126.073
+    },
+    {
+      "epoch": 0.012551724137931035,
+      "grad_norm": 9.9144868850708,
+      "learning_rate": 1.7323308270676692e-05,
+      "loss": 0.47073376178741455,
+      "num_input_tokens_seen": 90290,
+      "step": 273,
+      "train_runtime": 715.5598,
+      "train_tokens_per_second": 126.181
+    },
+    {
+      "epoch": 0.012597701149425287,
+      "grad_norm": 8.592867851257324,
+      "learning_rate": 1.731328320802005e-05,
+      "loss": 0.4184023439884186,
+      "num_input_tokens_seen": 90638,
+      "step": 274,
+      "train_runtime": 717.6489,
+      "train_tokens_per_second": 126.299
+    },
+    {
+      "epoch": 0.01264367816091954,
+      "grad_norm": 8.979439735412598,
+      "learning_rate": 1.730325814536341e-05,
+      "loss": 0.6420491337776184,
+      "num_input_tokens_seen": 91084,
+      "step": 275,
+      "train_runtime": 719.6293,
+      "train_tokens_per_second": 126.571
+    },
+    {
+      "epoch": 0.012689655172413793,
+      "grad_norm": 9.092283248901367,
+      "learning_rate": 1.729323308270677e-05,
+      "loss": 0.5877478122711182,
+      "num_input_tokens_seen": 91372,
+      "step": 276,
+      "train_runtime": 721.5768,
+      "train_tokens_per_second": 126.628
+    },
+    {
+      "epoch": 0.012735632183908045,
+      "grad_norm": 8.461853981018066,
+      "learning_rate": 1.7283208020050126e-05,
+      "loss": 0.4582377076148987,
+      "num_input_tokens_seen": 91686,
+      "step": 277,
+      "train_runtime": 723.5234,
+      "train_tokens_per_second": 126.722
+    },
+    {
+      "epoch": 0.0127816091954023,
+      "grad_norm": 8.02602767944336,
+      "learning_rate": 1.7273182957393487e-05,
+      "loss": 0.37842997908592224,
+      "num_input_tokens_seen": 91922,
+      "step": 278,
+      "train_runtime": 725.4315,
+      "train_tokens_per_second": 126.714
+    },
+    {
+      "epoch": 0.012827586206896552,
+      "grad_norm": 9.255718231201172,
+      "learning_rate": 1.7263157894736843e-05,
+      "loss": 0.4739605784416199,
+      "num_input_tokens_seen": 92214,
+      "step": 279,
+      "train_runtime": 727.3498,
+      "train_tokens_per_second": 126.781
+    },
+    {
+      "epoch": 0.012873563218390805,
+      "grad_norm": 7.671210765838623,
+      "learning_rate": 1.72531328320802e-05,
+      "loss": 0.49289393424987793,
+      "num_input_tokens_seen": 92466,
+      "step": 280,
+      "train_runtime": 729.2645,
+      "train_tokens_per_second": 126.794
+    },
+    {
+      "epoch": 0.012919540229885057,
+      "grad_norm": 9.932632446289062,
+      "learning_rate": 1.724310776942356e-05,
+      "loss": 0.6571778059005737,
+      "num_input_tokens_seen": 92786,
+      "step": 281,
+      "train_runtime": 731.2261,
+      "train_tokens_per_second": 126.891
+    },
+    {
+      "epoch": 0.01296551724137931,
+      "grad_norm": 8.414435386657715,
+      "learning_rate": 1.723308270676692e-05,
+      "loss": 0.44216346740722656,
+      "num_input_tokens_seen": 93148,
+      "step": 282,
+      "train_runtime": 733.235,
+      "train_tokens_per_second": 127.037
+    },
+    {
+      "epoch": 0.013011494252873563,
+      "grad_norm": 7.832511901855469,
+      "learning_rate": 1.7223057644110277e-05,
+      "loss": 0.46839600801467896,
+      "num_input_tokens_seen": 93472,
+      "step": 283,
+      "train_runtime": 735.225,
+      "train_tokens_per_second": 127.134
+    },
+    {
+      "epoch": 0.013057471264367817,
+      "grad_norm": 7.626879692077637,
+      "learning_rate": 1.7213032581453634e-05,
+      "loss": 0.5550291538238525,
+      "num_input_tokens_seen": 93986,
+      "step": 284,
+      "train_runtime": 737.2216,
+      "train_tokens_per_second": 127.487
+    },
+    {
+      "epoch": 0.01310344827586207,
+      "grad_norm": 7.270344257354736,
+      "learning_rate": 1.7203007518796994e-05,
+      "loss": 0.43810874223709106,
+      "num_input_tokens_seen": 94284,
+      "step": 285,
+      "train_runtime": 739.1505,
+      "train_tokens_per_second": 127.557
+    },
+    {
+      "epoch": 0.013149425287356322,
+      "grad_norm": 6.669817924499512,
+      "learning_rate": 1.719298245614035e-05,
+      "loss": 0.41238388419151306,
+      "num_input_tokens_seen": 94694,
+      "step": 286,
+      "train_runtime": 741.1069,
+      "train_tokens_per_second": 127.774
+    },
+    {
+      "epoch": 0.013195402298850575,
+      "grad_norm": 9.237834930419922,
+      "learning_rate": 1.718295739348371e-05,
+      "loss": 0.551144540309906,
+      "num_input_tokens_seen": 94988,
+      "step": 287,
+      "train_runtime": 743.0385,
+      "train_tokens_per_second": 127.837
+    },
+    {
+      "epoch": 0.013241379310344827,
+      "grad_norm": 7.017147064208984,
+      "learning_rate": 1.717293233082707e-05,
+      "loss": 0.5644881129264832,
+      "num_input_tokens_seen": 95292,
+      "step": 288,
+      "train_runtime": 744.9691,
+      "train_tokens_per_second": 127.914
+    },
+    {
+      "epoch": 0.01328735632183908,
+      "grad_norm": 9.167874336242676,
+      "learning_rate": 1.7162907268170428e-05,
+      "loss": 0.47601842880249023,
+      "num_input_tokens_seen": 95566,
+      "step": 289,
+      "train_runtime": 746.8893,
+      "train_tokens_per_second": 127.952
+    },
+    {
+      "epoch": 0.013333333333333334,
+      "grad_norm": 7.763336658477783,
+      "learning_rate": 1.7152882205513785e-05,
+      "loss": 0.6520231366157532,
+      "num_input_tokens_seen": 96018,
+      "step": 290,
+      "train_runtime": 748.8527,
+      "train_tokens_per_second": 128.22
+    },
+    {
+      "epoch": 0.013379310344827587,
+      "grad_norm": 13.905603408813477,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 0.4207511246204376,
+      "num_input_tokens_seen": 96266,
+      "step": 291,
+      "train_runtime": 750.7788,
+      "train_tokens_per_second": 128.222
+    },
+    {
+      "epoch": 0.01342528735632184,
+      "grad_norm": 7.5040507316589355,
+      "learning_rate": 1.7132832080200502e-05,
+      "loss": 0.4394516944885254,
+      "num_input_tokens_seen": 96552,
+      "step": 292,
+      "train_runtime": 752.6971,
+      "train_tokens_per_second": 128.275
+    },
+    {
+      "epoch": 0.013471264367816092,
+      "grad_norm": 7.481922626495361,
+      "learning_rate": 1.7122807017543862e-05,
+      "loss": 0.33811718225479126,
+      "num_input_tokens_seen": 96764,
+      "step": 293,
+      "train_runtime": 754.6065,
+      "train_tokens_per_second": 128.231
+    },
+    {
+      "epoch": 0.013517241379310345,
+      "grad_norm": 7.259603977203369,
+      "learning_rate": 1.711278195488722e-05,
+      "loss": 0.42554327845573425,
+      "num_input_tokens_seen": 97044,
+      "step": 294,
+      "train_runtime": 756.5344,
+      "train_tokens_per_second": 128.274
+    },
+    {
+      "epoch": 0.013563218390804597,
+      "grad_norm": 10.953673362731934,
+      "learning_rate": 1.710275689223058e-05,
+      "loss": 0.4624016284942627,
+      "num_input_tokens_seen": 97312,
+      "step": 295,
+      "train_runtime": 758.457,
+      "train_tokens_per_second": 128.303
+    },
+    {
+      "epoch": 0.01360919540229885,
+      "grad_norm": 8.060202598571777,
+      "learning_rate": 1.7092731829573936e-05,
+      "loss": 0.5388157367706299,
+      "num_input_tokens_seen": 97582,
+      "step": 296,
+      "train_runtime": 760.3841,
+      "train_tokens_per_second": 128.333
+    },
+    {
+      "epoch": 0.013655172413793104,
+      "grad_norm": 8.586666107177734,
+      "learning_rate": 1.7082706766917293e-05,
+      "loss": 0.49518680572509766,
+      "num_input_tokens_seen": 97874,
+      "step": 297,
+      "train_runtime": 762.3113,
+      "train_tokens_per_second": 128.391
+    },
+    {
+      "epoch": 0.013701149425287357,
+      "grad_norm": 8.171697616577148,
+      "learning_rate": 1.7072681704260653e-05,
+      "loss": 0.5041380524635315,
+      "num_input_tokens_seen": 98134,
+      "step": 298,
+      "train_runtime": 764.2276,
+      "train_tokens_per_second": 128.409
+    },
+    {
+      "epoch": 0.01374712643678161,
+      "grad_norm": 8.097844123840332,
+      "learning_rate": 1.7062656641604013e-05,
+      "loss": 0.4328901767730713,
+      "num_input_tokens_seen": 98426,
+      "step": 299,
+      "train_runtime": 766.1514,
+      "train_tokens_per_second": 128.468
+    },
+    {
+      "epoch": 0.013793103448275862,
+      "grad_norm": 10.759598731994629,
+      "learning_rate": 1.705263157894737e-05,
+      "loss": 0.378260999917984,
+      "num_input_tokens_seen": 98770,
+      "step": 300,
+      "train_runtime": 768.0928,
+      "train_tokens_per_second": 128.591
+    },
+    {
+      "epoch": 0.013839080459770114,
+      "grad_norm": 8.365592956542969,
+      "learning_rate": 1.7042606516290727e-05,
+      "loss": 0.6392844319343567,
+      "num_input_tokens_seen": 99046,
+      "step": 301,
+      "train_runtime": 783.4775,
+      "train_tokens_per_second": 126.418
+    },
+    {
+      "epoch": 0.013885057471264367,
+      "grad_norm": 8.994288444519043,
+      "learning_rate": 1.7032581453634087e-05,
+      "loss": 0.5737530589103699,
+      "num_input_tokens_seen": 99326,
+      "step": 302,
+      "train_runtime": 785.3924,
+      "train_tokens_per_second": 126.467
+    },
+    {
+      "epoch": 0.013931034482758621,
+      "grad_norm": 7.944206237792969,
+      "learning_rate": 1.7022556390977444e-05,
+      "loss": 0.4733693301677704,
+      "num_input_tokens_seen": 99582,
+      "step": 303,
+      "train_runtime": 787.2968,
+      "train_tokens_per_second": 126.486
+    },
+    {
+      "epoch": 0.013977011494252874,
+      "grad_norm": 10.713095664978027,
+      "learning_rate": 1.7012531328320804e-05,
+      "loss": 0.41429686546325684,
+      "num_input_tokens_seen": 99890,
+      "step": 304,
+      "train_runtime": 789.2387,
+      "train_tokens_per_second": 126.565
+    },
+    {
+      "epoch": 0.014022988505747127,
+      "grad_norm": 8.32962703704834,
+      "learning_rate": 1.7002506265664164e-05,
+      "loss": 0.4429081678390503,
+      "num_input_tokens_seen": 100146,
+      "step": 305,
+      "train_runtime": 791.1459,
+      "train_tokens_per_second": 126.583
+    },
+    {
+      "epoch": 0.01406896551724138,
+      "grad_norm": 8.988188743591309,
+      "learning_rate": 1.699248120300752e-05,
+      "loss": 0.5631526708602905,
+      "num_input_tokens_seen": 100492,
+      "step": 306,
+      "train_runtime": 793.0648,
+      "train_tokens_per_second": 126.713
+    },
+    {
+      "epoch": 0.014114942528735632,
+      "grad_norm": 7.367366313934326,
+      "learning_rate": 1.6982456140350878e-05,
+      "loss": 0.5590333342552185,
+      "num_input_tokens_seen": 100898,
+      "step": 307,
+      "train_runtime": 795.0135,
+      "train_tokens_per_second": 126.914
+    },
+    {
+      "epoch": 0.014160919540229884,
+      "grad_norm": 9.300395965576172,
+      "learning_rate": 1.6972431077694238e-05,
+      "loss": 0.4675888419151306,
+      "num_input_tokens_seen": 101212,
+      "step": 308,
+      "train_runtime": 796.9417,
+      "train_tokens_per_second": 127.001
+    },
+    {
+      "epoch": 0.014206896551724139,
+      "grad_norm": 9.40662956237793,
+      "learning_rate": 1.6962406015037595e-05,
+      "loss": 0.48680806159973145,
+      "num_input_tokens_seen": 101478,
+      "step": 309,
+      "train_runtime": 798.8809,
+      "train_tokens_per_second": 127.025
+    },
+    {
+      "epoch": 0.014252873563218391,
+      "grad_norm": 8.786764144897461,
+      "learning_rate": 1.6952380952380955e-05,
+      "loss": 0.4684376120567322,
+      "num_input_tokens_seen": 101804,
+      "step": 310,
+      "train_runtime": 800.8359,
+      "train_tokens_per_second": 127.122
+    },
+    {
+      "epoch": 0.014298850574712644,
+      "grad_norm": 8.8251314163208,
+      "learning_rate": 1.694235588972431e-05,
+      "loss": 0.5090711712837219,
+      "num_input_tokens_seen": 102078,
+      "step": 311,
+      "train_runtime": 802.8359,
+      "train_tokens_per_second": 127.147
+    },
+    {
+      "epoch": 0.014344827586206896,
+      "grad_norm": 9.90039348602295,
+      "learning_rate": 1.6932330827067672e-05,
+      "loss": 0.4405490458011627,
+      "num_input_tokens_seen": 102402,
+      "step": 312,
+      "train_runtime": 804.7714,
+      "train_tokens_per_second": 127.244
+    },
+    {
+      "epoch": 0.014390804597701149,
+      "grad_norm": 9.395299911499023,
+      "learning_rate": 1.692230576441103e-05,
+      "loss": 0.520408570766449,
+      "num_input_tokens_seen": 102676,
+      "step": 313,
+      "train_runtime": 806.7197,
+      "train_tokens_per_second": 127.276
+    },
+    {
+      "epoch": 0.014436781609195402,
+      "grad_norm": 7.501882553100586,
+      "learning_rate": 1.6912280701754385e-05,
+      "loss": 0.4870721101760864,
+      "num_input_tokens_seen": 103136,
+      "step": 314,
+      "train_runtime": 808.8103,
+      "train_tokens_per_second": 127.516
+    },
+    {
+      "epoch": 0.014482758620689656,
+      "grad_norm": 9.068699836730957,
+      "learning_rate": 1.6902255639097746e-05,
+      "loss": 0.5614039301872253,
+      "num_input_tokens_seen": 103436,
+      "step": 315,
+      "train_runtime": 810.7864,
+      "train_tokens_per_second": 127.575
+    },
+    {
+      "epoch": 0.014528735632183909,
+      "grad_norm": 9.348381996154785,
+      "learning_rate": 1.6892230576441106e-05,
+      "loss": 0.6185885667800903,
+      "num_input_tokens_seen": 103808,
+      "step": 316,
+      "train_runtime": 812.7838,
+      "train_tokens_per_second": 127.719
+    },
+    {
+      "epoch": 0.014574712643678161,
+      "grad_norm": 8.946013450622559,
+      "learning_rate": 1.6882205513784463e-05,
+      "loss": 0.643990159034729,
+      "num_input_tokens_seen": 104166,
+      "step": 317,
+      "train_runtime": 814.7717,
+      "train_tokens_per_second": 127.847
+    },
+    {
+      "epoch": 0.014620689655172414,
+      "grad_norm": 7.907708168029785,
+      "learning_rate": 1.6872180451127823e-05,
+      "loss": 0.37308964133262634,
+      "num_input_tokens_seen": 104450,
+      "step": 318,
+      "train_runtime": 816.7446,
+      "train_tokens_per_second": 127.886
+    },
+    {
+      "epoch": 0.014666666666666666,
+      "grad_norm": 9.479870796203613,
+      "learning_rate": 1.686215538847118e-05,
+      "loss": 0.5713379383087158,
+      "num_input_tokens_seen": 104776,
+      "step": 319,
+      "train_runtime": 818.7124,
+      "train_tokens_per_second": 127.977
+    },
+    {
+      "epoch": 0.014712643678160919,
+      "grad_norm": 6.283615589141846,
+      "learning_rate": 1.6852130325814536e-05,
+      "loss": 0.3166767954826355,
+      "num_input_tokens_seen": 105032,
+      "step": 320,
+      "train_runtime": 820.6333,
+      "train_tokens_per_second": 127.989
+    },
+    {
+      "epoch": 0.014758620689655172,
+      "grad_norm": 7.925952434539795,
+      "learning_rate": 1.6842105263157896e-05,
+      "loss": 0.44178274273872375,
+      "num_input_tokens_seen": 105318,
+      "step": 321,
+      "train_runtime": 822.5648,
+      "train_tokens_per_second": 128.036
+    },
+    {
+      "epoch": 0.014804597701149426,
+      "grad_norm": 8.26136589050293,
+      "learning_rate": 1.6832080200501257e-05,
+      "loss": 0.544378936290741,
+      "num_input_tokens_seen": 105596,
+      "step": 322,
+      "train_runtime": 824.4826,
+      "train_tokens_per_second": 128.075
+    },
+    {
+      "epoch": 0.014850574712643679,
+      "grad_norm": 9.990446090698242,
+      "learning_rate": 1.6822055137844613e-05,
+      "loss": 0.610353410243988,
+      "num_input_tokens_seen": 105872,
+      "step": 323,
+      "train_runtime": 826.437,
+      "train_tokens_per_second": 128.107
+    },
+    {
+      "epoch": 0.014896551724137931,
+      "grad_norm": 7.083834171295166,
+      "learning_rate": 1.681203007518797e-05,
+      "loss": 0.3682815730571747,
+      "num_input_tokens_seen": 106104,
+      "step": 324,
+      "train_runtime": 828.4051,
+      "train_tokens_per_second": 128.082
+    },
+    {
+      "epoch": 0.014942528735632184,
+      "grad_norm": 7.654078483581543,
+      "learning_rate": 1.680200501253133e-05,
+      "loss": 0.35451170802116394,
+      "num_input_tokens_seen": 106348,
+      "step": 325,
+      "train_runtime": 830.3205,
+      "train_tokens_per_second": 128.081
+    },
+    {
+      "epoch": 0.014988505747126436,
+      "grad_norm": 7.1275634765625,
+      "learning_rate": 1.6791979949874687e-05,
+      "loss": 0.44117680191993713,
+      "num_input_tokens_seen": 106656,
+      "step": 326,
+      "train_runtime": 832.2674,
+      "train_tokens_per_second": 128.151
+    },
+    {
+      "epoch": 0.015034482758620689,
+      "grad_norm": 11.717792510986328,
+      "learning_rate": 1.6781954887218047e-05,
+      "loss": 0.5222866535186768,
+      "num_input_tokens_seen": 106952,
+      "step": 327,
+      "train_runtime": 834.2345,
+      "train_tokens_per_second": 128.204
+    },
+    {
+      "epoch": 0.015080459770114943,
+      "grad_norm": 8.47635555267334,
+      "learning_rate": 1.6771929824561408e-05,
+      "loss": 0.556428849697113,
+      "num_input_tokens_seen": 107250,
+      "step": 328,
+      "train_runtime": 836.2166,
+      "train_tokens_per_second": 128.256
+    },
+    {
+      "epoch": 0.015126436781609196,
+      "grad_norm": 9.061455726623535,
+      "learning_rate": 1.6761904761904764e-05,
+      "loss": 0.6142518520355225,
+      "num_input_tokens_seen": 107688,
+      "step": 329,
+      "train_runtime": 838.2193,
+      "train_tokens_per_second": 128.472
+    },
+    {
+      "epoch": 0.015172413793103448,
+      "grad_norm": 7.577719211578369,
+      "learning_rate": 1.675187969924812e-05,
+      "loss": 0.44597262144088745,
+      "num_input_tokens_seen": 108310,
+      "step": 330,
+      "train_runtime": 840.2549,
+      "train_tokens_per_second": 128.901
+    },
+    {
+      "epoch": 0.015218390804597701,
+      "grad_norm": 8.179983139038086,
+      "learning_rate": 1.6741854636591478e-05,
+      "loss": 0.45987093448638916,
+      "num_input_tokens_seen": 108608,
+      "step": 331,
+      "train_runtime": 855.6631,
+      "train_tokens_per_second": 126.928
+    },
+    {
+      "epoch": 0.015264367816091954,
+      "grad_norm": 8.76684856414795,
+      "learning_rate": 1.6731829573934838e-05,
+      "loss": 0.47345876693725586,
+      "num_input_tokens_seen": 108880,
+      "step": 332,
+      "train_runtime": 857.7949,
+      "train_tokens_per_second": 126.93
+    },
+    {
+      "epoch": 0.015310344827586206,
+      "grad_norm": 9.098061561584473,
+      "learning_rate": 1.67218045112782e-05,
+      "loss": 0.4772406816482544,
+      "num_input_tokens_seen": 109184,
+      "step": 333,
+      "train_runtime": 859.9185,
+      "train_tokens_per_second": 126.97
+    },
+    {
+      "epoch": 0.01535632183908046,
+      "grad_norm": 10.331489562988281,
+      "learning_rate": 1.6711779448621555e-05,
+      "loss": 0.6067453026771545,
+      "num_input_tokens_seen": 109548,
+      "step": 334,
+      "train_runtime": 861.9056,
+      "train_tokens_per_second": 127.1
+    },
+    {
+      "epoch": 0.015402298850574713,
+      "grad_norm": 8.581093788146973,
+      "learning_rate": 1.6701754385964915e-05,
+      "loss": 0.36510738730430603,
+      "num_input_tokens_seen": 109778,
+      "step": 335,
+      "train_runtime": 863.8217,
+      "train_tokens_per_second": 127.084
+    },
+    {
+      "epoch": 0.015448275862068966,
+      "grad_norm": 8.387250900268555,
+      "learning_rate": 1.6691729323308272e-05,
+      "loss": 0.5398239493370056,
+      "num_input_tokens_seen": 110126,
+      "step": 336,
+      "train_runtime": 865.7672,
+      "train_tokens_per_second": 127.2
+    },
+    {
+      "epoch": 0.015494252873563218,
+      "grad_norm": 8.926971435546875,
+      "learning_rate": 1.668170426065163e-05,
+      "loss": 0.4762512147426605,
+      "num_input_tokens_seen": 110434,
+      "step": 337,
+      "train_runtime": 867.7732,
+      "train_tokens_per_second": 127.261
+    },
+    {
+      "epoch": 0.015540229885057471,
+      "grad_norm": 7.646859169006348,
+      "learning_rate": 1.667167919799499e-05,
+      "loss": 0.4378516674041748,
+      "num_input_tokens_seen": 110776,
+      "step": 338,
+      "train_runtime": 869.7589,
+      "train_tokens_per_second": 127.364
+    },
+    {
+      "epoch": 0.015586206896551724,
+      "grad_norm": 8.107491493225098,
+      "learning_rate": 1.666165413533835e-05,
+      "loss": 0.5538474917411804,
+      "num_input_tokens_seen": 111070,
+      "step": 339,
+      "train_runtime": 871.7858,
+      "train_tokens_per_second": 127.405
+    },
+    {
+      "epoch": 0.015632183908045976,
+      "grad_norm": 11.101648330688477,
+      "learning_rate": 1.6651629072681706e-05,
+      "loss": 0.6063014268875122,
+      "num_input_tokens_seen": 111402,
+      "step": 340,
+      "train_runtime": 873.7629,
+      "train_tokens_per_second": 127.497
+    },
+    {
+      "epoch": 0.01567816091954023,
+      "grad_norm": 9.16617488861084,
+      "learning_rate": 1.6641604010025063e-05,
+      "loss": 0.5937843918800354,
+      "num_input_tokens_seen": 111790,
+      "step": 341,
+      "train_runtime": 875.6943,
+      "train_tokens_per_second": 127.659
+    },
+    {
+      "epoch": 0.01572413793103448,
+      "grad_norm": 7.617735862731934,
+      "learning_rate": 1.6631578947368423e-05,
+      "loss": 0.39564117789268494,
+      "num_input_tokens_seen": 112048,
+      "step": 342,
+      "train_runtime": 877.6145,
+      "train_tokens_per_second": 127.673
+    },
+    {
+      "epoch": 0.015770114942528737,
+      "grad_norm": 8.599608421325684,
+      "learning_rate": 1.662155388471178e-05,
+      "loss": 0.5022115707397461,
+      "num_input_tokens_seen": 112352,
+      "step": 343,
+      "train_runtime": 879.5625,
+      "train_tokens_per_second": 127.736
+    },
+    {
+      "epoch": 0.01581609195402299,
+      "grad_norm": 8.8609037399292,
+      "learning_rate": 1.661152882205514e-05,
+      "loss": 0.39528289437294006,
+      "num_input_tokens_seen": 112638,
+      "step": 344,
+      "train_runtime": 881.4772,
+      "train_tokens_per_second": 127.783
+    },
+    {
+      "epoch": 0.015862068965517243,
+      "grad_norm": 8.134845733642578,
+      "learning_rate": 1.6601503759398497e-05,
+      "loss": 0.5726451277732849,
+      "num_input_tokens_seen": 112888,
+      "step": 345,
+      "train_runtime": 883.3984,
+      "train_tokens_per_second": 127.788
+    },
+    {
+      "epoch": 0.015908045977011495,
+      "grad_norm": 6.8687286376953125,
+      "learning_rate": 1.6591478696741857e-05,
+      "loss": 0.3758698105812073,
+      "num_input_tokens_seen": 113156,
+      "step": 346,
+      "train_runtime": 885.4027,
+      "train_tokens_per_second": 127.802
+    },
+    {
+      "epoch": 0.015954022988505748,
+      "grad_norm": 7.754480361938477,
+      "learning_rate": 1.6581453634085214e-05,
+      "loss": 0.4603171944618225,
+      "num_input_tokens_seen": 113532,
+      "step": 347,
+      "train_runtime": 887.3867,
+      "train_tokens_per_second": 127.94
+    },
+    {
+      "epoch": 0.016,
+      "grad_norm": 8.390192985534668,
+      "learning_rate": 1.6571428571428574e-05,
+      "loss": 0.505501925945282,
+      "num_input_tokens_seen": 113836,
+      "step": 348,
+      "train_runtime": 889.3595,
+      "train_tokens_per_second": 127.998
+    },
+    {
+      "epoch": 0.016045977011494253,
+      "grad_norm": 7.221319198608398,
+      "learning_rate": 1.656140350877193e-05,
+      "loss": 0.475628137588501,
+      "num_input_tokens_seen": 114130,
+      "step": 349,
+      "train_runtime": 891.4636,
+      "train_tokens_per_second": 128.025
+    },
+    {
+      "epoch": 0.016091954022988506,
+      "grad_norm": 8.97022819519043,
+      "learning_rate": 1.655137844611529e-05,
+      "loss": 0.44890573620796204,
+      "num_input_tokens_seen": 114394,
+      "step": 350,
+      "train_runtime": 893.4071,
+      "train_tokens_per_second": 128.042
+    },
+    {
+      "epoch": 0.016137931034482758,
+      "grad_norm": 10.288277626037598,
+      "learning_rate": 1.6541353383458648e-05,
+      "loss": 0.6184815168380737,
+      "num_input_tokens_seen": 114748,
+      "step": 351,
+      "train_runtime": 895.3702,
+      "train_tokens_per_second": 128.157
+    },
+    {
+      "epoch": 0.01618390804597701,
+      "grad_norm": 8.24072551727295,
+      "learning_rate": 1.6531328320802008e-05,
+      "loss": 0.5375922322273254,
+      "num_input_tokens_seen": 115160,
+      "step": 352,
+      "train_runtime": 897.3642,
+      "train_tokens_per_second": 128.331
+    },
+    {
+      "epoch": 0.016229885057471263,
+      "grad_norm": 8.401235580444336,
+      "learning_rate": 1.6521303258145365e-05,
+      "loss": 0.5189110636711121,
+      "num_input_tokens_seen": 115512,
+      "step": 353,
+      "train_runtime": 899.3421,
+      "train_tokens_per_second": 128.441
+    },
+    {
+      "epoch": 0.016275862068965516,
+      "grad_norm": 8.033496856689453,
+      "learning_rate": 1.651127819548872e-05,
+      "loss": 0.42611074447631836,
+      "num_input_tokens_seen": 115792,
+      "step": 354,
+      "train_runtime": 901.273,
+      "train_tokens_per_second": 128.476
+    },
+    {
+      "epoch": 0.01632183908045977,
+      "grad_norm": 8.075447082519531,
+      "learning_rate": 1.650125313283208e-05,
+      "loss": 0.5030648708343506,
+      "num_input_tokens_seen": 116042,
+      "step": 355,
+      "train_runtime": 903.1993,
+      "train_tokens_per_second": 128.479
+    },
+    {
+      "epoch": 0.016367816091954025,
+      "grad_norm": 8.209312438964844,
+      "learning_rate": 1.649122807017544e-05,
+      "loss": 0.4906337857246399,
+      "num_input_tokens_seen": 116326,
+      "step": 356,
+      "train_runtime": 905.1097,
+      "train_tokens_per_second": 128.521
+    },
+    {
+      "epoch": 0.016413793103448277,
+      "grad_norm": 8.38083553314209,
+      "learning_rate": 1.64812030075188e-05,
+      "loss": 0.39616087079048157,
+      "num_input_tokens_seen": 116580,
+      "step": 357,
+      "train_runtime": 907.0758,
+      "train_tokens_per_second": 128.523
+    },
+    {
+      "epoch": 0.01645977011494253,
+      "grad_norm": 7.717851638793945,
+      "learning_rate": 1.647117794486216e-05,
+      "loss": 0.5361239314079285,
+      "num_input_tokens_seen": 117006,
+      "step": 358,
+      "train_runtime": 909.1606,
+      "train_tokens_per_second": 128.697
+    },
+    {
+      "epoch": 0.016505747126436782,
+      "grad_norm": 7.047603130340576,
+      "learning_rate": 1.6461152882205516e-05,
+      "loss": 0.46394774317741394,
+      "num_input_tokens_seen": 117400,
+      "step": 359,
+      "train_runtime": 911.1297,
+      "train_tokens_per_second": 128.851
+    },
+    {
+      "epoch": 0.016551724137931035,
+      "grad_norm": 8.65599250793457,
+      "learning_rate": 1.6451127819548872e-05,
+      "loss": 0.5242359638214111,
+      "num_input_tokens_seen": 117698,
+      "step": 360,
+      "train_runtime": 913.0502,
+      "train_tokens_per_second": 128.906
+    },
+    {
+      "epoch": 0.016597701149425288,
+      "grad_norm": 8.625332832336426,
+      "learning_rate": 1.6441102756892233e-05,
+      "loss": 0.5002393126487732,
+      "num_input_tokens_seen": 117944,
+      "step": 361,
+      "train_runtime": 928.4831,
+      "train_tokens_per_second": 127.029
+    },
+    {
+      "epoch": 0.01664367816091954,
+      "grad_norm": 8.352180480957031,
+      "learning_rate": 1.643107769423559e-05,
+      "loss": 0.5270761847496033,
+      "num_input_tokens_seen": 118240,
+      "step": 362,
+      "train_runtime": 930.4629,
+      "train_tokens_per_second": 127.077
+    },
+    {
+      "epoch": 0.016689655172413793,
+      "grad_norm": 9.43403148651123,
+      "learning_rate": 1.642105263157895e-05,
+      "loss": 0.5236190557479858,
+      "num_input_tokens_seen": 118514,
+      "step": 363,
+      "train_runtime": 932.4238,
+      "train_tokens_per_second": 127.103
+    },
+    {
+      "epoch": 0.016735632183908045,
+      "grad_norm": 8.648951530456543,
+      "learning_rate": 1.6411027568922306e-05,
+      "loss": 0.6613885164260864,
+      "num_input_tokens_seen": 118822,
+      "step": 364,
+      "train_runtime": 934.3426,
+      "train_tokens_per_second": 127.172
+    },
+    {
+      "epoch": 0.016781609195402298,
+      "grad_norm": 7.909249782562256,
+      "learning_rate": 1.6401002506265667e-05,
+      "loss": 0.5001586675643921,
+      "num_input_tokens_seen": 119082,
+      "step": 365,
+      "train_runtime": 936.2579,
+      "train_tokens_per_second": 127.189
+    },
+    {
+      "epoch": 0.01682758620689655,
+      "grad_norm": 7.648731708526611,
+      "learning_rate": 1.6390977443609023e-05,
+      "loss": 0.5139956474304199,
+      "num_input_tokens_seen": 119444,
+      "step": 366,
+      "train_runtime": 938.2199,
+      "train_tokens_per_second": 127.309
+    },
+    {
+      "epoch": 0.016873563218390803,
+      "grad_norm": 6.9863972663879395,
+      "learning_rate": 1.6380952380952384e-05,
+      "loss": 0.452639102935791,
+      "num_input_tokens_seen": 119800,
+      "step": 367,
+      "train_runtime": 940.173,
+      "train_tokens_per_second": 127.423
+    },
+    {
+      "epoch": 0.01691954022988506,
+      "grad_norm": 8.20292854309082,
+      "learning_rate": 1.637092731829574e-05,
+      "loss": 0.5598886013031006,
+      "num_input_tokens_seen": 120176,
+      "step": 368,
+      "train_runtime": 942.1221,
+      "train_tokens_per_second": 127.559
+    },
+    {
+      "epoch": 0.016965517241379312,
+      "grad_norm": 9.673002243041992,
+      "learning_rate": 1.63609022556391e-05,
+      "loss": 0.5143294334411621,
+      "num_input_tokens_seen": 120440,
+      "step": 369,
+      "train_runtime": 944.0387,
+      "train_tokens_per_second": 127.58
+    },
+    {
+      "epoch": 0.017011494252873564,
+      "grad_norm": 9.281442642211914,
+      "learning_rate": 1.6350877192982457e-05,
+      "loss": 0.4952996075153351,
+      "num_input_tokens_seen": 120712,
+      "step": 370,
+      "train_runtime": 945.9497,
+      "train_tokens_per_second": 127.609
+    },
+    {
+      "epoch": 0.017057471264367817,
+      "grad_norm": 7.510295867919922,
+      "learning_rate": 1.6340852130325814e-05,
+      "loss": 0.44082415103912354,
+      "num_input_tokens_seen": 121034,
+      "step": 371,
+      "train_runtime": 947.9804,
+      "train_tokens_per_second": 127.676
+    },
+    {
+      "epoch": 0.01710344827586207,
+      "grad_norm": 9.4686279296875,
+      "learning_rate": 1.6330827067669174e-05,
+      "loss": 0.6736571788787842,
+      "num_input_tokens_seen": 121364,
+      "step": 372,
+      "train_runtime": 949.925,
+      "train_tokens_per_second": 127.762
+    },
+    {
+      "epoch": 0.017149425287356322,
+      "grad_norm": 8.580607414245605,
+      "learning_rate": 1.632080200501253e-05,
+      "loss": 0.43911564350128174,
+      "num_input_tokens_seen": 121612,
+      "step": 373,
+      "train_runtime": 951.8563,
+      "train_tokens_per_second": 127.763
+    },
+    {
+      "epoch": 0.017195402298850575,
+      "grad_norm": 8.830676078796387,
+      "learning_rate": 1.631077694235589e-05,
+      "loss": 0.5348225235939026,
+      "num_input_tokens_seen": 121974,
+      "step": 374,
+      "train_runtime": 953.975,
+      "train_tokens_per_second": 127.859
+    },
+    {
+      "epoch": 0.017241379310344827,
+      "grad_norm": 7.836789131164551,
+      "learning_rate": 1.630075187969925e-05,
+      "loss": 0.5751879215240479,
+      "num_input_tokens_seen": 122574,
+      "step": 375,
+      "train_runtime": 956.0011,
+      "train_tokens_per_second": 128.215
+    },
+    {
+      "epoch": 0.01728735632183908,
+      "grad_norm": 7.419215679168701,
+      "learning_rate": 1.6290726817042608e-05,
+      "loss": 0.5154892802238464,
+      "num_input_tokens_seen": 122868,
+      "step": 376,
+      "train_runtime": 957.9278,
+      "train_tokens_per_second": 128.264
+    },
+    {
+      "epoch": 0.017333333333333333,
+      "grad_norm": 5.741374492645264,
+      "learning_rate": 1.6280701754385965e-05,
+      "loss": 0.39768892526626587,
+      "num_input_tokens_seen": 123364,
+      "step": 377,
+      "train_runtime": 959.8945,
+      "train_tokens_per_second": 128.518
+    },
+    {
+      "epoch": 0.017379310344827585,
+      "grad_norm": 8.026482582092285,
+      "learning_rate": 1.6270676691729325e-05,
+      "loss": 0.4915647506713867,
+      "num_input_tokens_seen": 123676,
+      "step": 378,
+      "train_runtime": 961.815,
+      "train_tokens_per_second": 128.586
+    },
+    {
+      "epoch": 0.017425287356321838,
+      "grad_norm": 7.1091790199279785,
+      "learning_rate": 1.6260651629072682e-05,
+      "loss": 0.36034339666366577,
+      "num_input_tokens_seen": 123956,
+      "step": 379,
+      "train_runtime": 963.735,
+      "train_tokens_per_second": 128.62
+    },
+    {
+      "epoch": 0.01747126436781609,
+      "grad_norm": 7.859342575073242,
+      "learning_rate": 1.6250626566416042e-05,
+      "loss": 0.5458072423934937,
+      "num_input_tokens_seen": 124324,
+      "step": 380,
+      "train_runtime": 965.6798,
+      "train_tokens_per_second": 128.742
+    },
+    {
+      "epoch": 0.017517241379310346,
+      "grad_norm": 7.662074565887451,
+      "learning_rate": 1.62406015037594e-05,
+      "loss": 0.40609803795814514,
+      "num_input_tokens_seen": 124660,
+      "step": 381,
+      "train_runtime": 967.5932,
+      "train_tokens_per_second": 128.835
+    },
+    {
+      "epoch": 0.0175632183908046,
+      "grad_norm": 9.377715110778809,
+      "learning_rate": 1.623057644110276e-05,
+      "loss": 0.48562780022621155,
+      "num_input_tokens_seen": 124928,
+      "step": 382,
+      "train_runtime": 969.5044,
+      "train_tokens_per_second": 128.858
+    },
+    {
+      "epoch": 0.01760919540229885,
+      "grad_norm": 6.903769016265869,
+      "learning_rate": 1.6220551378446116e-05,
+      "loss": 0.4790268540382385,
+      "num_input_tokens_seen": 125410,
+      "step": 383,
+      "train_runtime": 971.4774,
+      "train_tokens_per_second": 129.092
+    },
+    {
+      "epoch": 0.017655172413793104,
+      "grad_norm": 9.729009628295898,
+      "learning_rate": 1.6210526315789473e-05,
+      "loss": 0.45791855454444885,
+      "num_input_tokens_seen": 125714,
+      "step": 384,
+      "train_runtime": 973.4006,
+      "train_tokens_per_second": 129.149
+    },
+    {
+      "epoch": 0.017701149425287357,
+      "grad_norm": 8.090932846069336,
+      "learning_rate": 1.6200501253132833e-05,
+      "loss": 0.5926223993301392,
+      "num_input_tokens_seen": 125990,
+      "step": 385,
+      "train_runtime": 975.3097,
+      "train_tokens_per_second": 129.179
+    },
+    {
+      "epoch": 0.01774712643678161,
+      "grad_norm": 8.990884780883789,
+      "learning_rate": 1.6190476190476193e-05,
+      "loss": 0.502015233039856,
+      "num_input_tokens_seen": 126296,
+      "step": 386,
+      "train_runtime": 977.2341,
+      "train_tokens_per_second": 129.238
+    },
+    {
+      "epoch": 0.017793103448275862,
+      "grad_norm": 11.029932022094727,
+      "learning_rate": 1.618045112781955e-05,
+      "loss": 0.49521729350090027,
+      "num_input_tokens_seen": 126564,
+      "step": 387,
+      "train_runtime": 979.1455,
+      "train_tokens_per_second": 129.26
+    },
+    {
+      "epoch": 0.017839080459770115,
+      "grad_norm": 7.295618057250977,
+      "learning_rate": 1.617042606516291e-05,
+      "loss": 0.4368196725845337,
+      "num_input_tokens_seen": 126802,
+      "step": 388,
+      "train_runtime": 981.0589,
+      "train_tokens_per_second": 129.25
+    },
+    {
+      "epoch": 0.017885057471264367,
+      "grad_norm": 7.750117301940918,
+      "learning_rate": 1.6160401002506267e-05,
+      "loss": 0.43421801924705505,
+      "num_input_tokens_seen": 127106,
+      "step": 389,
+      "train_runtime": 983.0081,
+      "train_tokens_per_second": 129.303
+    },
+    {
+      "epoch": 0.01793103448275862,
+      "grad_norm": 8.780847549438477,
+      "learning_rate": 1.6150375939849624e-05,
+      "loss": 0.5297210216522217,
+      "num_input_tokens_seen": 127348,
+      "step": 390,
+      "train_runtime": 984.9338,
+      "train_tokens_per_second": 129.296
+    },
+    {
+      "epoch": 0.017977011494252872,
+      "grad_norm": 8.037254333496094,
+      "learning_rate": 1.6140350877192984e-05,
+      "loss": 0.460166871547699,
+      "num_input_tokens_seen": 127628,
+      "step": 391,
+      "train_runtime": 1000.0858,
+      "train_tokens_per_second": 127.617
+    },
+    {
+      "epoch": 0.018022988505747125,
+      "grad_norm": 8.30826473236084,
+      "learning_rate": 1.6130325814536344e-05,
+      "loss": 0.5625733137130737,
+      "num_input_tokens_seen": 127920,
+      "step": 392,
+      "train_runtime": 1002.0399,
+      "train_tokens_per_second": 127.66
+    },
+    {
+      "epoch": 0.01806896551724138,
+      "grad_norm": 8.45525074005127,
+      "learning_rate": 1.61203007518797e-05,
+      "loss": 0.4925714135169983,
+      "num_input_tokens_seen": 128208,
+      "step": 393,
+      "train_runtime": 1003.9933,
+      "train_tokens_per_second": 127.698
+    },
+    {
+      "epoch": 0.018114942528735634,
+      "grad_norm": 7.612026691436768,
+      "learning_rate": 1.6110275689223058e-05,
+      "loss": 0.2628011703491211,
+      "num_input_tokens_seen": 128440,
+      "step": 394,
+      "train_runtime": 1005.9157,
+      "train_tokens_per_second": 127.685
+    },
+    {
+      "epoch": 0.018160919540229886,
+      "grad_norm": 7.3546833992004395,
+      "learning_rate": 1.6100250626566418e-05,
+      "loss": 0.4682520925998688,
+      "num_input_tokens_seen": 128752,
+      "step": 395,
+      "train_runtime": 1007.9693,
+      "train_tokens_per_second": 127.734
+    },
+    {
+      "epoch": 0.01820689655172414,
+      "grad_norm": 8.364006996154785,
+      "learning_rate": 1.6090225563909775e-05,
+      "loss": 0.43775445222854614,
+      "num_input_tokens_seen": 129068,
+      "step": 396,
+      "train_runtime": 1009.9523,
+      "train_tokens_per_second": 127.796
+    },
+    {
+      "epoch": 0.01825287356321839,
+      "grad_norm": 9.414189338684082,
+      "learning_rate": 1.6080200501253135e-05,
+      "loss": 0.4796205163002014,
+      "num_input_tokens_seen": 129402,
+      "step": 397,
+      "train_runtime": 1011.8954,
+      "train_tokens_per_second": 127.881
+    },
+    {
+      "epoch": 0.018298850574712644,
+      "grad_norm": 7.125086307525635,
+      "learning_rate": 1.6070175438596495e-05,
+      "loss": 0.46837103366851807,
+      "num_input_tokens_seen": 129590,
+      "step": 398,
+      "train_runtime": 1013.8308,
+      "train_tokens_per_second": 127.822
+    },
+    {
+      "epoch": 0.018344827586206897,
+      "grad_norm": 10.200335502624512,
+      "learning_rate": 1.6060150375939852e-05,
+      "loss": 0.6392890810966492,
+      "num_input_tokens_seen": 129860,
+      "step": 399,
+      "train_runtime": 1015.7951,
+      "train_tokens_per_second": 127.841
+    },
+    {
+      "epoch": 0.01839080459770115,
+      "grad_norm": 8.488134384155273,
+      "learning_rate": 1.605012531328321e-05,
+      "loss": 0.47278720140457153,
+      "num_input_tokens_seen": 130184,
+      "step": 400,
+      "train_runtime": 1017.7361,
+      "train_tokens_per_second": 127.915
+    },
+    {
+      "epoch": 0.01839080459770115,
+      "eval_loss": 1.83921217918396,
+      "eval_runtime": 54.5857,
+      "eval_samples_per_second": 18.32,
+      "eval_steps_per_second": 9.16,
+      "num_input_tokens_seen": 130184,
+      "step": 400
+    },
+    {
+      "epoch": 0.018436781609195402,
+      "grad_norm": 10.583003044128418,
+      "learning_rate": 1.6040100250626565e-05,
+      "loss": 0.664654016494751,
+      "num_input_tokens_seen": 130504,
+      "step": 401,
+      "train_runtime": 1074.343,
+      "train_tokens_per_second": 121.473
+    },
+    {
+      "epoch": 0.018482758620689654,
+      "grad_norm": 9.928531646728516,
+      "learning_rate": 1.6030075187969926e-05,
+      "loss": 0.4345932602882385,
+      "num_input_tokens_seen": 130760,
+      "step": 402,
+      "train_runtime": 1076.3619,
+      "train_tokens_per_second": 121.483
+    },
+    {
+      "epoch": 0.018528735632183907,
+      "grad_norm": 8.73652458190918,
+      "learning_rate": 1.6020050125313286e-05,
+      "loss": 0.5674711465835571,
+      "num_input_tokens_seen": 131072,
+      "step": 403,
+      "train_runtime": 1078.2913,
+      "train_tokens_per_second": 121.555
+    },
+    {
+      "epoch": 0.01857471264367816,
+      "grad_norm": 8.819191932678223,
+      "learning_rate": 1.6010025062656642e-05,
+      "loss": 0.5101428627967834,
+      "num_input_tokens_seen": 131360,
+      "step": 404,
+      "train_runtime": 1080.2613,
+      "train_tokens_per_second": 121.6
+    },
+    {
+      "epoch": 0.018620689655172412,
+      "grad_norm": 11.008004188537598,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.664053201675415,
+      "num_input_tokens_seen": 131736,
+      "step": 405,
+      "train_runtime": 1082.2012,
+      "train_tokens_per_second": 121.73
+    },
+    {
+      "epoch": 0.018666666666666668,
+      "grad_norm": 8.193777084350586,
+      "learning_rate": 1.598997493734336e-05,
+      "loss": 0.524330198764801,
+      "num_input_tokens_seen": 132174,
+      "step": 406,
+      "train_runtime": 1084.1663,
+      "train_tokens_per_second": 121.913
+    },
+    {
+      "epoch": 0.01871264367816092,
+      "grad_norm": 8.529227256774902,
+      "learning_rate": 1.5979949874686716e-05,
+      "loss": 0.31128376722335815,
+      "num_input_tokens_seen": 132404,
+      "step": 407,
+      "train_runtime": 1086.1,
+      "train_tokens_per_second": 121.908
+    },
+    {
+      "epoch": 0.018758620689655173,
+      "grad_norm": 7.439627170562744,
+      "learning_rate": 1.5969924812030076e-05,
+      "loss": 0.45540714263916016,
+      "num_input_tokens_seen": 132660,
+      "step": 408,
+      "train_runtime": 1088.101,
+      "train_tokens_per_second": 121.919
+    },
+    {
+      "epoch": 0.018804597701149426,
+      "grad_norm": 7.563625335693359,
+      "learning_rate": 1.5959899749373437e-05,
+      "loss": 0.40846192836761475,
+      "num_input_tokens_seen": 132908,
+      "step": 409,
+      "train_runtime": 1090.0297,
+      "train_tokens_per_second": 121.931
+    },
+    {
+      "epoch": 0.01885057471264368,
+      "grad_norm": 7.777848720550537,
+      "learning_rate": 1.5949874686716793e-05,
+      "loss": 0.5753679275512695,
+      "num_input_tokens_seen": 133400,
+      "step": 410,
+      "train_runtime": 1092.0325,
+      "train_tokens_per_second": 122.158
+    },
+    {
+      "epoch": 0.01889655172413793,
+      "grad_norm": 8.254167556762695,
+      "learning_rate": 1.593984962406015e-05,
+      "loss": 0.5449351072311401,
+      "num_input_tokens_seen": 133720,
+      "step": 411,
+      "train_runtime": 1093.9714,
+      "train_tokens_per_second": 122.234
+    },
+    {
+      "epoch": 0.018942528735632184,
+      "grad_norm": 10.308324813842773,
+      "learning_rate": 1.592982456140351e-05,
+      "loss": 0.500423789024353,
+      "num_input_tokens_seen": 133996,
+      "step": 412,
+      "train_runtime": 1095.9362,
+      "train_tokens_per_second": 122.266
+    },
+    {
+      "epoch": 0.018988505747126436,
+      "grad_norm": 6.994194507598877,
+      "learning_rate": 1.5919799498746867e-05,
+      "loss": 0.5234413146972656,
+      "num_input_tokens_seen": 134572,
+      "step": 413,
+      "train_runtime": 1098.0558,
+      "train_tokens_per_second": 122.555
+    },
+    {
+      "epoch": 0.01903448275862069,
+      "grad_norm": 9.316163063049316,
+      "learning_rate": 1.5909774436090227e-05,
+      "loss": 0.5487146377563477,
+      "num_input_tokens_seen": 134854,
+      "step": 414,
+      "train_runtime": 1100.0275,
+      "train_tokens_per_second": 122.591
+    },
+    {
+      "epoch": 0.01908045977011494,
+      "grad_norm": 7.727750778198242,
+      "learning_rate": 1.5899749373433588e-05,
+      "loss": 0.3887421488761902,
+      "num_input_tokens_seen": 135158,
+      "step": 415,
+      "train_runtime": 1102.0303,
+      "train_tokens_per_second": 122.645
+    },
+    {
+      "epoch": 0.019126436781609194,
+      "grad_norm": 6.617283344268799,
+      "learning_rate": 1.5889724310776944e-05,
+      "loss": 0.4084390699863434,
+      "num_input_tokens_seen": 135414,
+      "step": 416,
+      "train_runtime": 1103.9422,
+      "train_tokens_per_second": 122.664
+    },
+    {
+      "epoch": 0.019172413793103447,
+      "grad_norm": 8.78074836730957,
+      "learning_rate": 1.58796992481203e-05,
+      "loss": 0.5863832831382751,
+      "num_input_tokens_seen": 135808,
+      "step": 417,
+      "train_runtime": 1105.899,
+      "train_tokens_per_second": 122.803
+    },
+    {
+      "epoch": 0.019218390804597703,
+      "grad_norm": 8.403968811035156,
+      "learning_rate": 1.5869674185463658e-05,
+      "loss": 0.4228143095970154,
+      "num_input_tokens_seen": 136044,
+      "step": 418,
+      "train_runtime": 1107.8257,
+      "train_tokens_per_second": 122.803
+    },
+    {
+      "epoch": 0.019264367816091955,
+      "grad_norm": 11.255876541137695,
+      "learning_rate": 1.5859649122807018e-05,
+      "loss": 0.418620765209198,
+      "num_input_tokens_seen": 136322,
+      "step": 419,
+      "train_runtime": 1109.7699,
+      "train_tokens_per_second": 122.838
+    },
+    {
+      "epoch": 0.019310344827586208,
+      "grad_norm": 9.40054702758789,
+      "learning_rate": 1.5849624060150378e-05,
+      "loss": 0.5212907791137695,
+      "num_input_tokens_seen": 136602,
+      "step": 420,
+      "train_runtime": 1111.696,
+      "train_tokens_per_second": 122.877
+    },
+    {
+      "epoch": 0.01935632183908046,
+      "grad_norm": 8.371843338012695,
+      "learning_rate": 1.5839598997493735e-05,
+      "loss": 0.5342715978622437,
+      "num_input_tokens_seen": 136842,
+      "step": 421,
+      "train_runtime": 1127.3407,
+      "train_tokens_per_second": 121.385
+    },
+    {
+      "epoch": 0.019402298850574713,
+      "grad_norm": 7.767118453979492,
+      "learning_rate": 1.5829573934837095e-05,
+      "loss": 0.4537525773048401,
+      "num_input_tokens_seen": 137110,
+      "step": 422,
+      "train_runtime": 1129.271,
+      "train_tokens_per_second": 121.415
+    },
+    {
+      "epoch": 0.019448275862068966,
+      "grad_norm": 7.4775261878967285,
+      "learning_rate": 1.5819548872180452e-05,
+      "loss": 0.5792855620384216,
+      "num_input_tokens_seen": 137550,
+      "step": 423,
+      "train_runtime": 1131.2949,
+      "train_tokens_per_second": 121.586
+    },
+    {
+      "epoch": 0.01949425287356322,
+      "grad_norm": 8.903519630432129,
+      "learning_rate": 1.580952380952381e-05,
+      "loss": 0.5991487503051758,
+      "num_input_tokens_seen": 137820,
+      "step": 424,
+      "train_runtime": 1133.2398,
+      "train_tokens_per_second": 121.616
+    },
+    {
+      "epoch": 0.01954022988505747,
+      "grad_norm": 6.718785285949707,
+      "learning_rate": 1.579949874686717e-05,
+      "loss": 0.32860735058784485,
+      "num_input_tokens_seen": 138092,
+      "step": 425,
+      "train_runtime": 1135.1767,
+      "train_tokens_per_second": 121.648
+    },
+    {
+      "epoch": 0.019586206896551724,
+      "grad_norm": 9.034529685974121,
+      "learning_rate": 1.578947368421053e-05,
+      "loss": 0.5332455635070801,
+      "num_input_tokens_seen": 138342,
+      "step": 426,
+      "train_runtime": 1137.1264,
+      "train_tokens_per_second": 121.659
+    },
+    {
+      "epoch": 0.019632183908045976,
+      "grad_norm": 8.134893417358398,
+      "learning_rate": 1.5779448621553886e-05,
+      "loss": 0.34569254517555237,
+      "num_input_tokens_seen": 138584,
+      "step": 427,
+      "train_runtime": 1139.0641,
+      "train_tokens_per_second": 121.665
+    },
+    {
+      "epoch": 0.01967816091954023,
+      "grad_norm": 7.84832239151001,
+      "learning_rate": 1.5769423558897243e-05,
+      "loss": 0.5754497051239014,
+      "num_input_tokens_seen": 138904,
+      "step": 428,
+      "train_runtime": 1141.019,
+      "train_tokens_per_second": 121.737
+    },
+    {
+      "epoch": 0.01972413793103448,
+      "grad_norm": 8.042332649230957,
+      "learning_rate": 1.5759398496240603e-05,
+      "loss": 0.5238772034645081,
+      "num_input_tokens_seen": 139274,
+      "step": 429,
+      "train_runtime": 1143.0173,
+      "train_tokens_per_second": 121.848
+    },
+    {
+      "epoch": 0.019770114942528734,
+      "grad_norm": 8.774839401245117,
+      "learning_rate": 1.574937343358396e-05,
+      "loss": 0.5461223125457764,
+      "num_input_tokens_seen": 139716,
+      "step": 430,
+      "train_runtime": 1145.026,
+      "train_tokens_per_second": 122.02
+    },
+    {
+      "epoch": 0.01981609195402299,
+      "grad_norm": 9.017069816589355,
+      "learning_rate": 1.573934837092732e-05,
+      "loss": 0.45460045337677,
+      "num_input_tokens_seen": 139990,
+      "step": 431,
+      "train_runtime": 1147.0367,
+      "train_tokens_per_second": 122.045
+    },
+    {
+      "epoch": 0.019862068965517243,
+      "grad_norm": 7.076886177062988,
+      "learning_rate": 1.572932330827068e-05,
+      "loss": 0.45128387212753296,
+      "num_input_tokens_seen": 140418,
+      "step": 432,
+      "train_runtime": 1149.0105,
+      "train_tokens_per_second": 122.208
+    },
+    {
+      "epoch": 0.019908045977011495,
+      "grad_norm": 8.149660110473633,
+      "learning_rate": 1.5719298245614037e-05,
+      "loss": 0.5153030157089233,
+      "num_input_tokens_seen": 140734,
+      "step": 433,
+      "train_runtime": 1150.9352,
+      "train_tokens_per_second": 122.278
+    },
+    {
+      "epoch": 0.019954022988505748,
+      "grad_norm": 7.91249418258667,
+      "learning_rate": 1.5709273182957394e-05,
+      "loss": 0.4741940200328827,
+      "num_input_tokens_seen": 141004,
+      "step": 434,
+      "train_runtime": 1152.8614,
+      "train_tokens_per_second": 122.308
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 9.027854919433594,
+      "learning_rate": 1.5699248120300754e-05,
+      "loss": 0.5704158544540405,
+      "num_input_tokens_seen": 141316,
+      "step": 435,
+      "train_runtime": 1154.7862,
+      "train_tokens_per_second": 122.374
+    },
+    {
+      "epoch": 0.020045977011494253,
+      "grad_norm": 7.727769374847412,
+      "learning_rate": 1.568922305764411e-05,
+      "loss": 0.3855794370174408,
+      "num_input_tokens_seen": 141574,
+      "step": 436,
+      "train_runtime": 1156.7188,
+      "train_tokens_per_second": 122.393
+    },
+    {
+      "epoch": 0.020091954022988506,
+      "grad_norm": 7.8650593757629395,
+      "learning_rate": 1.567919799498747e-05,
+      "loss": 0.39530476927757263,
+      "num_input_tokens_seen": 141808,
+      "step": 437,
+      "train_runtime": 1158.6305,
+      "train_tokens_per_second": 122.393
+    },
+    {
+      "epoch": 0.020137931034482758,
+      "grad_norm": 8.397459983825684,
+      "learning_rate": 1.5669172932330828e-05,
+      "loss": 0.4957209825515747,
+      "num_input_tokens_seen": 142082,
+      "step": 438,
+      "train_runtime": 1160.5466,
+      "train_tokens_per_second": 122.427
+    },
+    {
+      "epoch": 0.02018390804597701,
+      "grad_norm": 7.495139122009277,
+      "learning_rate": 1.5659147869674188e-05,
+      "loss": 0.4659573435783386,
+      "num_input_tokens_seen": 142390,
+      "step": 439,
+      "train_runtime": 1162.4756,
+      "train_tokens_per_second": 122.489
+    },
+    {
+      "epoch": 0.020229885057471263,
+      "grad_norm": 9.757864952087402,
+      "learning_rate": 1.5649122807017545e-05,
+      "loss": 0.553784966468811,
+      "num_input_tokens_seen": 142764,
+      "step": 440,
+      "train_runtime": 1164.41,
+      "train_tokens_per_second": 122.606
+    },
+    {
+      "epoch": 0.020275862068965516,
+      "grad_norm": 8.840959548950195,
+      "learning_rate": 1.56390977443609e-05,
+      "loss": 0.4977976083755493,
+      "num_input_tokens_seen": 143074,
+      "step": 441,
+      "train_runtime": 1166.4194,
+      "train_tokens_per_second": 122.661
+    },
+    {
+      "epoch": 0.02032183908045977,
+      "grad_norm": 8.150773048400879,
+      "learning_rate": 1.562907268170426e-05,
+      "loss": 0.5242696404457092,
+      "num_input_tokens_seen": 143474,
+      "step": 442,
+      "train_runtime": 1168.4102,
+      "train_tokens_per_second": 122.794
+    },
+    {
+      "epoch": 0.020367816091954025,
+      "grad_norm": 8.866066932678223,
+      "learning_rate": 1.5619047619047622e-05,
+      "loss": 0.641364574432373,
+      "num_input_tokens_seen": 143864,
+      "step": 443,
+      "train_runtime": 1170.3661,
+      "train_tokens_per_second": 122.922
+    },
+    {
+      "epoch": 0.020413793103448277,
+      "grad_norm": 8.029109001159668,
+      "learning_rate": 1.560902255639098e-05,
+      "loss": 0.5271466374397278,
+      "num_input_tokens_seen": 144184,
+      "step": 444,
+      "train_runtime": 1172.2926,
+      "train_tokens_per_second": 122.993
+    },
+    {
+      "epoch": 0.02045977011494253,
+      "grad_norm": 6.696882724761963,
+      "learning_rate": 1.559899749373434e-05,
+      "loss": 0.29163163900375366,
+      "num_input_tokens_seen": 144416,
+      "step": 445,
+      "train_runtime": 1174.2064,
+      "train_tokens_per_second": 122.99
+    },
+    {
+      "epoch": 0.020505747126436782,
+      "grad_norm": 8.14375114440918,
+      "learning_rate": 1.5588972431077696e-05,
+      "loss": 0.469174325466156,
+      "num_input_tokens_seen": 144744,
+      "step": 446,
+      "train_runtime": 1176.1654,
+      "train_tokens_per_second": 123.064
+    },
+    {
+      "epoch": 0.020551724137931035,
+      "grad_norm": 8.158989906311035,
+      "learning_rate": 1.5578947368421052e-05,
+      "loss": 0.4768478274345398,
+      "num_input_tokens_seen": 145042,
+      "step": 447,
+      "train_runtime": 1178.1014,
+      "train_tokens_per_second": 123.115
+    },
+    {
+      "epoch": 0.020597701149425288,
+      "grad_norm": 7.018171310424805,
+      "learning_rate": 1.5568922305764413e-05,
+      "loss": 0.3784666061401367,
+      "num_input_tokens_seen": 145354,
+      "step": 448,
+      "train_runtime": 1180.0288,
+      "train_tokens_per_second": 123.178
+    },
+    {
+      "epoch": 0.02064367816091954,
+      "grad_norm": 9.718257904052734,
+      "learning_rate": 1.5558897243107773e-05,
+      "loss": 0.431374192237854,
+      "num_input_tokens_seen": 145630,
+      "step": 449,
+      "train_runtime": 1181.9543,
+      "train_tokens_per_second": 123.211
+    },
+    {
+      "epoch": 0.020689655172413793,
+      "grad_norm": 8.367993354797363,
+      "learning_rate": 1.554887218045113e-05,
+      "loss": 0.503094494342804,
+      "num_input_tokens_seen": 145952,
+      "step": 450,
+      "train_runtime": 1183.8786,
+      "train_tokens_per_second": 123.283
+    },
+    {
+      "epoch": 0.020735632183908045,
+      "grad_norm": 8.881491661071777,
+      "learning_rate": 1.5538847117794486e-05,
+      "loss": 0.6508010625839233,
+      "num_input_tokens_seen": 146236,
+      "step": 451,
+      "train_runtime": 1200.4898,
+      "train_tokens_per_second": 121.814
+    },
+    {
+      "epoch": 0.020781609195402298,
+      "grad_norm": 8.647028923034668,
+      "learning_rate": 1.5528822055137847e-05,
+      "loss": 0.5306245684623718,
+      "num_input_tokens_seen": 146644,
+      "step": 452,
+      "train_runtime": 1202.4537,
+      "train_tokens_per_second": 121.954
+    },
+    {
+      "epoch": 0.02082758620689655,
+      "grad_norm": 9.862062454223633,
+      "learning_rate": 1.5518796992481203e-05,
+      "loss": 0.5040091276168823,
+      "num_input_tokens_seen": 146924,
+      "step": 453,
+      "train_runtime": 1204.3853,
+      "train_tokens_per_second": 121.991
+    },
+    {
+      "epoch": 0.020873563218390803,
+      "grad_norm": 7.58740234375,
+      "learning_rate": 1.5508771929824563e-05,
+      "loss": 0.3113386929035187,
+      "num_input_tokens_seen": 147174,
+      "step": 454,
+      "train_runtime": 1206.2932,
+      "train_tokens_per_second": 122.005
+    },
+    {
+      "epoch": 0.020919540229885056,
+      "grad_norm": 7.78328800201416,
+      "learning_rate": 1.5498746867167924e-05,
+      "loss": 0.4073692858219147,
+      "num_input_tokens_seen": 147410,
+      "step": 455,
+      "train_runtime": 1208.2038,
+      "train_tokens_per_second": 122.008
+    },
+    {
+      "epoch": 0.020965517241379312,
+      "grad_norm": 7.565345764160156,
+      "learning_rate": 1.548872180451128e-05,
+      "loss": 0.40836089849472046,
+      "num_input_tokens_seen": 147656,
+      "step": 456,
+      "train_runtime": 1210.1109,
+      "train_tokens_per_second": 122.019
+    },
+    {
+      "epoch": 0.021011494252873564,
+      "grad_norm": 7.796747207641602,
+      "learning_rate": 1.5478696741854637e-05,
+      "loss": 0.5215970277786255,
+      "num_input_tokens_seen": 148002,
+      "step": 457,
+      "train_runtime": 1212.0942,
+      "train_tokens_per_second": 122.104
+    },
+    {
+      "epoch": 0.021057471264367817,
+      "grad_norm": 7.469287872314453,
+      "learning_rate": 1.5468671679197994e-05,
+      "loss": 0.4521618187427521,
+      "num_input_tokens_seen": 148314,
+      "step": 458,
+      "train_runtime": 1214.1412,
+      "train_tokens_per_second": 122.155
+    },
+    {
+      "epoch": 0.02110344827586207,
+      "grad_norm": 8.264181137084961,
+      "learning_rate": 1.5458646616541354e-05,
+      "loss": 0.4832594692707062,
+      "num_input_tokens_seen": 148574,
+      "step": 459,
+      "train_runtime": 1216.2168,
+      "train_tokens_per_second": 122.161
+    },
+    {
+      "epoch": 0.021149425287356322,
+      "grad_norm": 8.08668041229248,
+      "learning_rate": 1.5448621553884714e-05,
+      "loss": 0.46651777625083923,
+      "num_input_tokens_seen": 148954,
+      "step": 460,
+      "train_runtime": 1218.2745,
+      "train_tokens_per_second": 122.266
+    },
+    {
+      "epoch": 0.021195402298850575,
+      "grad_norm": 8.778621673583984,
+      "learning_rate": 1.543859649122807e-05,
+      "loss": 0.46269941329956055,
+      "num_input_tokens_seen": 149236,
+      "step": 461,
+      "train_runtime": 1220.2478,
+      "train_tokens_per_second": 122.3
+    },
+    {
+      "epoch": 0.021241379310344827,
+      "grad_norm": 7.521136283874512,
+      "learning_rate": 1.542857142857143e-05,
+      "loss": 0.39253920316696167,
+      "num_input_tokens_seen": 149522,
+      "step": 462,
+      "train_runtime": 1222.3107,
+      "train_tokens_per_second": 122.327
+    },
+    {
+      "epoch": 0.02128735632183908,
+      "grad_norm": 9.666208267211914,
+      "learning_rate": 1.5418546365914788e-05,
+      "loss": 0.5184369087219238,
+      "num_input_tokens_seen": 149780,
+      "step": 463,
+      "train_runtime": 1224.2963,
+      "train_tokens_per_second": 122.34
+    },
+    {
+      "epoch": 0.021333333333333333,
+      "grad_norm": 6.660348892211914,
+      "learning_rate": 1.5408521303258145e-05,
+      "loss": 0.43838030099868774,
+      "num_input_tokens_seen": 150366,
+      "step": 464,
+      "train_runtime": 1226.3605,
+      "train_tokens_per_second": 122.612
+    },
+    {
+      "epoch": 0.021379310344827585,
+      "grad_norm": 7.8536529541015625,
+      "learning_rate": 1.5398496240601505e-05,
+      "loss": 0.4633280038833618,
+      "num_input_tokens_seen": 150656,
+      "step": 465,
+      "train_runtime": 1228.346,
+      "train_tokens_per_second": 122.649
+    },
+    {
+      "epoch": 0.021425287356321838,
+      "grad_norm": 9.269664764404297,
+      "learning_rate": 1.5388471177944865e-05,
+      "loss": 0.4660486578941345,
+      "num_input_tokens_seen": 150930,
+      "step": 466,
+      "train_runtime": 1230.3816,
+      "train_tokens_per_second": 122.669
+    },
+    {
+      "epoch": 0.02147126436781609,
+      "grad_norm": 8.85814094543457,
+      "learning_rate": 1.5378446115288222e-05,
+      "loss": 0.6247583627700806,
+      "num_input_tokens_seen": 151268,
+      "step": 467,
+      "train_runtime": 1232.4881,
+      "train_tokens_per_second": 122.734
+    },
+    {
+      "epoch": 0.021517241379310346,
+      "grad_norm": 7.665519714355469,
+      "learning_rate": 1.536842105263158e-05,
+      "loss": 0.49776655435562134,
+      "num_input_tokens_seen": 151534,
+      "step": 468,
+      "train_runtime": 1234.4868,
+      "train_tokens_per_second": 122.751
+    },
+    {
+      "epoch": 0.0215632183908046,
+      "grad_norm": 8.703502655029297,
+      "learning_rate": 1.535839598997494e-05,
+      "loss": 0.4543405771255493,
+      "num_input_tokens_seen": 151818,
+      "step": 469,
+      "train_runtime": 1236.4857,
+      "train_tokens_per_second": 122.782
+    },
+    {
+      "epoch": 0.02160919540229885,
+      "grad_norm": 9.28618335723877,
+      "learning_rate": 1.5348370927318296e-05,
+      "loss": 0.5027074217796326,
+      "num_input_tokens_seen": 152176,
+      "step": 470,
+      "train_runtime": 1238.5203,
+      "train_tokens_per_second": 122.869
+    },
+    {
+      "epoch": 0.021655172413793104,
+      "grad_norm": 8.177092552185059,
+      "learning_rate": 1.5338345864661656e-05,
+      "loss": 0.5191943049430847,
+      "num_input_tokens_seen": 152496,
+      "step": 471,
+      "train_runtime": 1240.5355,
+      "train_tokens_per_second": 122.928
+    },
+    {
+      "epoch": 0.021701149425287357,
+      "grad_norm": 9.28865909576416,
+      "learning_rate": 1.5328320802005013e-05,
+      "loss": 0.4770624041557312,
+      "num_input_tokens_seen": 152886,
+      "step": 472,
+      "train_runtime": 1242.6231,
+      "train_tokens_per_second": 123.035
+    },
+    {
+      "epoch": 0.02174712643678161,
+      "grad_norm": 8.051939010620117,
+      "learning_rate": 1.5318295739348373e-05,
+      "loss": 0.468537300825119,
+      "num_input_tokens_seen": 153142,
+      "step": 473,
+      "train_runtime": 1244.6413,
+      "train_tokens_per_second": 123.041
+    },
+    {
+      "epoch": 0.021793103448275862,
+      "grad_norm": 7.96160888671875,
+      "learning_rate": 1.530827067669173e-05,
+      "loss": 0.47067826986312866,
+      "num_input_tokens_seen": 153408,
+      "step": 474,
+      "train_runtime": 1246.6788,
+      "train_tokens_per_second": 123.053
+    },
+    {
+      "epoch": 0.021839080459770115,
+      "grad_norm": 7.926995754241943,
+      "learning_rate": 1.529824561403509e-05,
+      "loss": 0.5530025959014893,
+      "num_input_tokens_seen": 153868,
+      "step": 475,
+      "train_runtime": 1248.7781,
+      "train_tokens_per_second": 123.215
+    },
+    {
+      "epoch": 0.021885057471264367,
+      "grad_norm": 7.6453022956848145,
+      "learning_rate": 1.5288220551378447e-05,
+      "loss": 0.46475687623023987,
+      "num_input_tokens_seen": 154140,
+      "step": 476,
+      "train_runtime": 1250.7882,
+      "train_tokens_per_second": 123.234
+    },
+    {
+      "epoch": 0.02193103448275862,
+      "grad_norm": 6.370632171630859,
+      "learning_rate": 1.5278195488721807e-05,
+      "loss": 0.3243369460105896,
+      "num_input_tokens_seen": 154390,
+      "step": 477,
+      "train_runtime": 1252.8214,
+      "train_tokens_per_second": 123.234
+    },
+    {
+      "epoch": 0.021977011494252872,
+      "grad_norm": 8.575563430786133,
+      "learning_rate": 1.5268170426065164e-05,
+      "loss": 0.5438640713691711,
+      "num_input_tokens_seen": 154672,
+      "step": 478,
+      "train_runtime": 1254.8355,
+      "train_tokens_per_second": 123.261
+    },
+    {
+      "epoch": 0.022022988505747125,
+      "grad_norm": 8.427260398864746,
+      "learning_rate": 1.5258145363408522e-05,
+      "loss": 0.4582708179950714,
+      "num_input_tokens_seen": 154988,
+      "step": 479,
+      "train_runtime": 1256.8219,
+      "train_tokens_per_second": 123.317
+    },
+    {
+      "epoch": 0.022068965517241378,
+      "grad_norm": 7.727510452270508,
+      "learning_rate": 1.524812030075188e-05,
+      "loss": 0.4847640097141266,
+      "num_input_tokens_seen": 155260,
+      "step": 480,
+      "train_runtime": 1258.7875,
+      "train_tokens_per_second": 123.341
+    },
+    {
+      "epoch": 0.022114942528735634,
+      "grad_norm": 8.2589111328125,
+      "learning_rate": 1.523809523809524e-05,
+      "loss": 0.4815359115600586,
+      "num_input_tokens_seen": 155586,
+      "step": 481,
+      "train_runtime": 1273.9387,
+      "train_tokens_per_second": 122.13
+    },
+    {
+      "epoch": 0.022160919540229886,
+      "grad_norm": 6.174463748931885,
+      "learning_rate": 1.5228070175438598e-05,
+      "loss": 0.5501198172569275,
+      "num_input_tokens_seen": 156492,
+      "step": 482,
+      "train_runtime": 1276.1628,
+      "train_tokens_per_second": 122.627
+    },
+    {
+      "epoch": 0.02220689655172414,
+      "grad_norm": 8.82485580444336,
+      "learning_rate": 1.5218045112781956e-05,
+      "loss": 0.45321643352508545,
+      "num_input_tokens_seen": 156780,
+      "step": 483,
+      "train_runtime": 1278.1967,
+      "train_tokens_per_second": 122.657
+    },
+    {
+      "epoch": 0.02225287356321839,
+      "grad_norm": 6.12175178527832,
+      "learning_rate": 1.5208020050125315e-05,
+      "loss": 0.3978089690208435,
+      "num_input_tokens_seen": 157088,
+      "step": 484,
+      "train_runtime": 1280.2145,
+      "train_tokens_per_second": 122.704
+    },
+    {
+      "epoch": 0.022298850574712644,
+      "grad_norm": 8.363214492797852,
+      "learning_rate": 1.5197994987468673e-05,
+      "loss": 0.5779292583465576,
+      "num_input_tokens_seen": 157444,
+      "step": 485,
+      "train_runtime": 1282.2865,
+      "train_tokens_per_second": 122.784
+    },
+    {
+      "epoch": 0.022344827586206897,
+      "grad_norm": 7.858548164367676,
+      "learning_rate": 1.5187969924812032e-05,
+      "loss": 0.4645827114582062,
+      "num_input_tokens_seen": 157718,
+      "step": 486,
+      "train_runtime": 1284.2629,
+      "train_tokens_per_second": 122.808
+    },
+    {
+      "epoch": 0.02239080459770115,
+      "grad_norm": 8.053011894226074,
+      "learning_rate": 1.5177944862155388e-05,
+      "loss": 0.5309751033782959,
+      "num_input_tokens_seen": 158040,
+      "step": 487,
+      "train_runtime": 1286.332,
+      "train_tokens_per_second": 122.861
+    },
+    {
+      "epoch": 0.022436781609195402,
+      "grad_norm": 8.499578475952148,
+      "learning_rate": 1.5167919799498747e-05,
+      "loss": 0.5499032139778137,
+      "num_input_tokens_seen": 158352,
+      "step": 488,
+      "train_runtime": 1288.3653,
+      "train_tokens_per_second": 122.909
+    },
+    {
+      "epoch": 0.022482758620689654,
+      "grad_norm": 8.27188491821289,
+      "learning_rate": 1.5157894736842107e-05,
+      "loss": 0.5076043009757996,
+      "num_input_tokens_seen": 158756,
+      "step": 489,
+      "train_runtime": 1290.469,
+      "train_tokens_per_second": 123.022
+    },
+    {
+      "epoch": 0.022528735632183907,
+      "grad_norm": 9.326817512512207,
+      "learning_rate": 1.5147869674185464e-05,
+      "loss": 0.5182700157165527,
+      "num_input_tokens_seen": 159020,
+      "step": 490,
+      "train_runtime": 1292.4903,
+      "train_tokens_per_second": 123.034
+    },
+    {
+      "epoch": 0.02257471264367816,
+      "grad_norm": 8.826271057128906,
+      "learning_rate": 1.5137844611528822e-05,
+      "loss": 0.491726279258728,
+      "num_input_tokens_seen": 159272,
+      "step": 491,
+      "train_runtime": 1294.603,
+      "train_tokens_per_second": 123.028
+    },
+    {
+      "epoch": 0.022620689655172412,
+      "grad_norm": 8.384391784667969,
+      "learning_rate": 1.5127819548872183e-05,
+      "loss": 0.6266180276870728,
+      "num_input_tokens_seen": 159538,
+      "step": 492,
+      "train_runtime": 1296.5724,
+      "train_tokens_per_second": 123.046
+    },
+    {
+      "epoch": 0.02266666666666667,
+      "grad_norm": 9.36903190612793,
+      "learning_rate": 1.511779448621554e-05,
+      "loss": 0.48814085125923157,
+      "num_input_tokens_seen": 159798,
+      "step": 493,
+      "train_runtime": 1298.5435,
+      "train_tokens_per_second": 123.059
+    },
+    {
+      "epoch": 0.02271264367816092,
+      "grad_norm": 8.908138275146484,
+      "learning_rate": 1.5107769423558898e-05,
+      "loss": 0.6812343597412109,
+      "num_input_tokens_seen": 160200,
+      "step": 494,
+      "train_runtime": 1300.7681,
+      "train_tokens_per_second": 123.158
+    },
+    {
+      "epoch": 0.022758620689655173,
+      "grad_norm": 8.59496784210205,
+      "learning_rate": 1.5097744360902258e-05,
+      "loss": 0.4920065104961395,
+      "num_input_tokens_seen": 160482,
+      "step": 495,
+      "train_runtime": 1302.7478,
+      "train_tokens_per_second": 123.187
+    },
+    {
+      "epoch": 0.022804597701149426,
+      "grad_norm": 7.333454132080078,
+      "learning_rate": 1.5087719298245615e-05,
+      "loss": 0.4726922810077667,
+      "num_input_tokens_seen": 160778,
+      "step": 496,
+      "train_runtime": 1304.7802,
+      "train_tokens_per_second": 123.222
+    },
+    {
+      "epoch": 0.02285057471264368,
+      "grad_norm": 6.41697883605957,
+      "learning_rate": 1.5077694235588973e-05,
+      "loss": 0.4380318522453308,
+      "num_input_tokens_seen": 161048,
+      "step": 497,
+      "train_runtime": 1306.7805,
+      "train_tokens_per_second": 123.24
+    },
+    {
+      "epoch": 0.02289655172413793,
+      "grad_norm": 8.461562156677246,
+      "learning_rate": 1.506766917293233e-05,
+      "loss": 0.48116976022720337,
+      "num_input_tokens_seen": 161414,
+      "step": 498,
+      "train_runtime": 1308.7561,
+      "train_tokens_per_second": 123.334
+    },
+    {
+      "epoch": 0.022942528735632184,
+      "grad_norm": 7.610356330871582,
+      "learning_rate": 1.505764411027569e-05,
+      "loss": 0.5558478236198425,
+      "num_input_tokens_seen": 161802,
+      "step": 499,
+      "train_runtime": 1310.7941,
+      "train_tokens_per_second": 123.438
+    },
+    {
+      "epoch": 0.022988505747126436,
+      "grad_norm": 7.735250949859619,
+      "learning_rate": 1.5047619047619049e-05,
+      "loss": 0.49599307775497437,
+      "num_input_tokens_seen": 162112,
+      "step": 500,
+      "train_runtime": 1312.7596,
+      "train_tokens_per_second": 123.489
+    },
+    {
+      "epoch": 0.02303448275862069,
+      "grad_norm": 9.797597885131836,
+      "learning_rate": 1.5037593984962406e-05,
+      "loss": 0.4995342195034027,
+      "num_input_tokens_seen": 162346,
+      "step": 501,
+      "train_runtime": 1314.8405,
+      "train_tokens_per_second": 123.472
+    },
+    {
+      "epoch": 0.02308045977011494,
+      "grad_norm": 7.869985103607178,
+      "learning_rate": 1.5027568922305766e-05,
+      "loss": 0.4657493531703949,
+      "num_input_tokens_seen": 162578,
+      "step": 502,
+      "train_runtime": 1316.8643,
+      "train_tokens_per_second": 123.458
+    },
+    {
+      "epoch": 0.023126436781609194,
+      "grad_norm": 10.171713829040527,
+      "learning_rate": 1.5017543859649124e-05,
+      "loss": 0.5351496934890747,
+      "num_input_tokens_seen": 162992,
+      "step": 503,
+      "train_runtime": 1318.9091,
+      "train_tokens_per_second": 123.581
+    },
+    {
+      "epoch": 0.023172413793103447,
+      "grad_norm": 10.212080955505371,
+      "learning_rate": 1.5007518796992481e-05,
+      "loss": 0.4960668087005615,
+      "num_input_tokens_seen": 163262,
+      "step": 504,
+      "train_runtime": 1320.9201,
+      "train_tokens_per_second": 123.597
+    },
+    {
+      "epoch": 0.0232183908045977,
+      "grad_norm": 7.753813743591309,
+      "learning_rate": 1.4997493734335841e-05,
+      "loss": 0.5769193172454834,
+      "num_input_tokens_seen": 163516,
+      "step": 505,
+      "train_runtime": 1322.9128,
+      "train_tokens_per_second": 123.603
+    },
+    {
+      "epoch": 0.023264367816091956,
+      "grad_norm": 7.388118267059326,
+      "learning_rate": 1.49874686716792e-05,
+      "loss": 0.39474964141845703,
+      "num_input_tokens_seen": 163770,
+      "step": 506,
+      "train_runtime": 1324.8763,
+      "train_tokens_per_second": 123.612
+    },
+    {
+      "epoch": 0.023310344827586208,
+      "grad_norm": 7.857269287109375,
+      "learning_rate": 1.4977443609022557e-05,
+      "loss": 0.4135737419128418,
+      "num_input_tokens_seen": 164132,
+      "step": 507,
+      "train_runtime": 1326.922,
+      "train_tokens_per_second": 123.694
+    },
+    {
+      "epoch": 0.02335632183908046,
+      "grad_norm": 9.931657791137695,
+      "learning_rate": 1.4967418546365915e-05,
+      "loss": 0.5334938168525696,
+      "num_input_tokens_seen": 164442,
+      "step": 508,
+      "train_runtime": 1328.9535,
+      "train_tokens_per_second": 123.738
+    },
+    {
+      "epoch": 0.023402298850574713,
+      "grad_norm": 7.529313087463379,
+      "learning_rate": 1.4957393483709275e-05,
+      "loss": 0.5453150272369385,
+      "num_input_tokens_seen": 164852,
+      "step": 509,
+      "train_runtime": 1330.952,
+      "train_tokens_per_second": 123.86
+    },
+    {
+      "epoch": 0.023448275862068966,
+      "grad_norm": 8.382284164428711,
+      "learning_rate": 1.4947368421052632e-05,
+      "loss": 0.5961440205574036,
+      "num_input_tokens_seen": 165228,
+      "step": 510,
+      "train_runtime": 1332.9383,
+      "train_tokens_per_second": 123.958
+    },
+    {
+      "epoch": 0.02349425287356322,
+      "grad_norm": 8.244186401367188,
+      "learning_rate": 1.493734335839599e-05,
+      "loss": 0.6272380948066711,
+      "num_input_tokens_seen": 165700,
+      "step": 511,
+      "train_runtime": 1347.7563,
+      "train_tokens_per_second": 122.945
+    },
+    {
+      "epoch": 0.02354022988505747,
+      "grad_norm": 6.809892177581787,
+      "learning_rate": 1.492731829573935e-05,
+      "loss": 0.40072810649871826,
+      "num_input_tokens_seen": 165926,
+      "step": 512,
+      "train_runtime": 1349.6926,
+      "train_tokens_per_second": 122.936
+    },
+    {
+      "epoch": 0.023586206896551724,
+      "grad_norm": 6.770306587219238,
+      "learning_rate": 1.4917293233082707e-05,
+      "loss": 0.34396111965179443,
+      "num_input_tokens_seen": 166178,
+      "step": 513,
+      "train_runtime": 1351.6389,
+      "train_tokens_per_second": 122.946
+    },
+    {
+      "epoch": 0.023632183908045976,
+      "grad_norm": 8.51698112487793,
+      "learning_rate": 1.4907268170426066e-05,
+      "loss": 0.44032377004623413,
+      "num_input_tokens_seen": 166540,
+      "step": 514,
+      "train_runtime": 1353.6505,
+      "train_tokens_per_second": 123.03
+    },
+    {
+      "epoch": 0.02367816091954023,
+      "grad_norm": 7.331808567047119,
+      "learning_rate": 1.4897243107769426e-05,
+      "loss": 0.4877912700176239,
+      "num_input_tokens_seen": 166802,
+      "step": 515,
+      "train_runtime": 1355.853,
+      "train_tokens_per_second": 123.024
+    },
+    {
+      "epoch": 0.02372413793103448,
+      "grad_norm": 7.553277492523193,
+      "learning_rate": 1.4887218045112783e-05,
+      "loss": 0.5095426440238953,
+      "num_input_tokens_seen": 167146,
+      "step": 516,
+      "train_runtime": 1357.8276,
+      "train_tokens_per_second": 123.098
+    },
+    {
+      "epoch": 0.023770114942528734,
+      "grad_norm": 9.049818992614746,
+      "learning_rate": 1.4877192982456141e-05,
+      "loss": 0.5703203678131104,
+      "num_input_tokens_seen": 167454,
+      "step": 517,
+      "train_runtime": 1359.7847,
+      "train_tokens_per_second": 123.147
+    },
+    {
+      "epoch": 0.02381609195402299,
+      "grad_norm": 8.191668510437012,
+      "learning_rate": 1.4867167919799498e-05,
+      "loss": 0.4943529963493347,
+      "num_input_tokens_seen": 167748,
+      "step": 518,
+      "train_runtime": 1361.7375,
+      "train_tokens_per_second": 123.187
+    },
+    {
+      "epoch": 0.023862068965517243,
+      "grad_norm": 8.651224136352539,
+      "learning_rate": 1.4857142857142858e-05,
+      "loss": 0.5686701536178589,
+      "num_input_tokens_seen": 168030,
+      "step": 519,
+      "train_runtime": 1363.6875,
+      "train_tokens_per_second": 123.217
+    },
+    {
+      "epoch": 0.023908045977011495,
+      "grad_norm": 7.486079692840576,
+      "learning_rate": 1.4847117794486217e-05,
+      "loss": 0.6263479590415955,
+      "num_input_tokens_seen": 168316,
+      "step": 520,
+      "train_runtime": 1365.641,
+      "train_tokens_per_second": 123.251
+    },
+    {
+      "epoch": 0.023954022988505748,
+      "grad_norm": 7.435495853424072,
+      "learning_rate": 1.4837092731829574e-05,
+      "loss": 0.463203102350235,
+      "num_input_tokens_seen": 168730,
+      "step": 521,
+      "train_runtime": 1367.6218,
+      "train_tokens_per_second": 123.375
+    },
+    {
+      "epoch": 0.024,
+      "grad_norm": 9.241541862487793,
+      "learning_rate": 1.4827067669172934e-05,
+      "loss": 0.31005528569221497,
+      "num_input_tokens_seen": 168992,
+      "step": 522,
+      "train_runtime": 1369.5807,
+      "train_tokens_per_second": 123.39
+    },
+    {
+      "epoch": 0.024045977011494253,
+      "grad_norm": 7.447315692901611,
+      "learning_rate": 1.4817042606516292e-05,
+      "loss": 0.5688130259513855,
+      "num_input_tokens_seen": 169278,
+      "step": 523,
+      "train_runtime": 1371.5402,
+      "train_tokens_per_second": 123.422
+    },
+    {
+      "epoch": 0.024091954022988506,
+      "grad_norm": 8.900172233581543,
+      "learning_rate": 1.4807017543859649e-05,
+      "loss": 0.4629487991333008,
+      "num_input_tokens_seen": 169518,
+      "step": 524,
+      "train_runtime": 1373.4644,
+      "train_tokens_per_second": 123.424
+    },
+    {
+      "epoch": 0.02413793103448276,
+      "grad_norm": 7.42686653137207,
+      "learning_rate": 1.479699248120301e-05,
+      "loss": 0.677523672580719,
+      "num_input_tokens_seen": 170060,
+      "step": 525,
+      "train_runtime": 1375.4921,
+      "train_tokens_per_second": 123.636
+    },
+    {
+      "epoch": 0.02418390804597701,
+      "grad_norm": 7.977572917938232,
+      "learning_rate": 1.4786967418546368e-05,
+      "loss": 0.4927060902118683,
+      "num_input_tokens_seen": 170368,
+      "step": 526,
+      "train_runtime": 1377.4365,
+      "train_tokens_per_second": 123.685
+    },
+    {
+      "epoch": 0.024229885057471263,
+      "grad_norm": 7.594063758850098,
+      "learning_rate": 1.4776942355889725e-05,
+      "loss": 0.5512229204177856,
+      "num_input_tokens_seen": 170688,
+      "step": 527,
+      "train_runtime": 1379.438,
+      "train_tokens_per_second": 123.737
+    },
+    {
+      "epoch": 0.024275862068965516,
+      "grad_norm": 10.225115776062012,
+      "learning_rate": 1.4766917293233083e-05,
+      "loss": 0.45086759328842163,
+      "num_input_tokens_seen": 170938,
+      "step": 528,
+      "train_runtime": 1381.404,
+      "train_tokens_per_second": 123.742
+    },
+    {
+      "epoch": 0.02432183908045977,
+      "grad_norm": 8.310455322265625,
+      "learning_rate": 1.4756892230576443e-05,
+      "loss": 0.5467156767845154,
+      "num_input_tokens_seen": 171270,
+      "step": 529,
+      "train_runtime": 1383.3557,
+      "train_tokens_per_second": 123.808
+    },
+    {
+      "epoch": 0.02436781609195402,
+      "grad_norm": 8.420825004577637,
+      "learning_rate": 1.47468671679198e-05,
+      "loss": 0.5078372955322266,
+      "num_input_tokens_seen": 171548,
+      "step": 530,
+      "train_runtime": 1385.3145,
+      "train_tokens_per_second": 123.833
+    },
+    {
+      "epoch": 0.024413793103448277,
+      "grad_norm": 6.346220016479492,
+      "learning_rate": 1.4736842105263159e-05,
+      "loss": 0.3089922368526459,
+      "num_input_tokens_seen": 171804,
+      "step": 531,
+      "train_runtime": 1387.2652,
+      "train_tokens_per_second": 123.844
+    },
+    {
+      "epoch": 0.02445977011494253,
+      "grad_norm": 7.621101379394531,
+      "learning_rate": 1.4726817042606519e-05,
+      "loss": 0.5899746417999268,
+      "num_input_tokens_seen": 172092,
+      "step": 532,
+      "train_runtime": 1389.2172,
+      "train_tokens_per_second": 123.877
+    },
+    {
+      "epoch": 0.024505747126436783,
+      "grad_norm": 9.290635108947754,
+      "learning_rate": 1.4716791979949876e-05,
+      "loss": 0.43958860635757446,
+      "num_input_tokens_seen": 172364,
+      "step": 533,
+      "train_runtime": 1391.169,
+      "train_tokens_per_second": 123.899
+    },
+    {
+      "epoch": 0.024551724137931035,
+      "grad_norm": 6.2526445388793945,
+      "learning_rate": 1.4706766917293234e-05,
+      "loss": 0.41999557614326477,
+      "num_input_tokens_seen": 172768,
+      "step": 534,
+      "train_runtime": 1393.1593,
+      "train_tokens_per_second": 124.012
+    },
+    {
+      "epoch": 0.024597701149425288,
+      "grad_norm": 7.623930931091309,
+      "learning_rate": 1.4696741854636594e-05,
+      "loss": 0.4332103133201599,
+      "num_input_tokens_seen": 173244,
+      "step": 535,
+      "train_runtime": 1395.1728,
+      "train_tokens_per_second": 124.174
+    },
+    {
+      "epoch": 0.02464367816091954,
+      "grad_norm": 7.97459077835083,
+      "learning_rate": 1.4686716791979951e-05,
+      "loss": 0.5894046425819397,
+      "num_input_tokens_seen": 173594,
+      "step": 536,
+      "train_runtime": 1397.1384,
+      "train_tokens_per_second": 124.25
+    },
+    {
+      "epoch": 0.024689655172413793,
+      "grad_norm": 8.448247909545898,
+      "learning_rate": 1.467669172932331e-05,
+      "loss": 0.3817816376686096,
+      "num_input_tokens_seen": 173896,
+      "step": 537,
+      "train_runtime": 1399.0896,
+      "train_tokens_per_second": 124.292
+    },
+    {
+      "epoch": 0.024735632183908045,
+      "grad_norm": 9.263516426086426,
+      "learning_rate": 1.4666666666666666e-05,
+      "loss": 0.5733354091644287,
+      "num_input_tokens_seen": 174264,
+      "step": 538,
+      "train_runtime": 1401.0423,
+      "train_tokens_per_second": 124.382
+    },
+    {
+      "epoch": 0.024781609195402298,
+      "grad_norm": 8.889558792114258,
+      "learning_rate": 1.4656641604010026e-05,
+      "loss": 0.5188465118408203,
+      "num_input_tokens_seen": 174560,
+      "step": 539,
+      "train_runtime": 1402.999,
+      "train_tokens_per_second": 124.419
+    },
+    {
+      "epoch": 0.02482758620689655,
+      "grad_norm": 6.875885963439941,
+      "learning_rate": 1.4646616541353385e-05,
+      "loss": 0.4254898130893707,
+      "num_input_tokens_seen": 174834,
+      "step": 540,
+      "train_runtime": 1404.9637,
+      "train_tokens_per_second": 124.44
+    },
+    {
+      "epoch": 0.024873563218390803,
+      "grad_norm": 8.637484550476074,
+      "learning_rate": 1.4636591478696742e-05,
+      "loss": 0.6016808748245239,
+      "num_input_tokens_seen": 175226,
+      "step": 541,
+      "train_runtime": 1420.1909,
+      "train_tokens_per_second": 123.382
+    },
+    {
+      "epoch": 0.024919540229885056,
+      "grad_norm": 8.453070640563965,
+      "learning_rate": 1.4626566416040102e-05,
+      "loss": 0.5227655172348022,
+      "num_input_tokens_seen": 175528,
+      "step": 542,
+      "train_runtime": 1422.1567,
+      "train_tokens_per_second": 123.424
+    },
+    {
+      "epoch": 0.024965517241379312,
+      "grad_norm": 7.210037708282471,
+      "learning_rate": 1.461654135338346e-05,
+      "loss": 0.46860626339912415,
+      "num_input_tokens_seen": 175844,
+      "step": 543,
+      "train_runtime": 1424.1666,
+      "train_tokens_per_second": 123.472
+    },
+    {
+      "epoch": 0.025011494252873565,
+      "grad_norm": 7.584609508514404,
+      "learning_rate": 1.4606516290726817e-05,
+      "loss": 0.49540987610816956,
+      "num_input_tokens_seen": 176076,
+      "step": 544,
+      "train_runtime": 1426.1361,
+      "train_tokens_per_second": 123.464
+    },
+    {
+      "epoch": 0.025057471264367817,
+      "grad_norm": 10.090349197387695,
+      "learning_rate": 1.4596491228070177e-05,
+      "loss": 0.6143651008605957,
+      "num_input_tokens_seen": 176392,
+      "step": 545,
+      "train_runtime": 1428.0895,
+      "train_tokens_per_second": 123.516
+    },
+    {
+      "epoch": 0.02510344827586207,
+      "grad_norm": 7.89303731918335,
+      "learning_rate": 1.4586466165413536e-05,
+      "loss": 0.4946277141571045,
+      "num_input_tokens_seen": 176704,
+      "step": 546,
+      "train_runtime": 1430.0562,
+      "train_tokens_per_second": 123.564
+    },
+    {
+      "epoch": 0.025149425287356322,
+      "grad_norm": 8.414692878723145,
+      "learning_rate": 1.4576441102756893e-05,
+      "loss": 0.3559797406196594,
+      "num_input_tokens_seen": 176988,
+      "step": 547,
+      "train_runtime": 1432.1087,
+      "train_tokens_per_second": 123.586
+    },
+    {
+      "epoch": 0.025195402298850575,
+      "grad_norm": 9.346529006958008,
+      "learning_rate": 1.4566416040100251e-05,
+      "loss": 0.5622787475585938,
+      "num_input_tokens_seen": 177306,
+      "step": 548,
+      "train_runtime": 1434.0917,
+      "train_tokens_per_second": 123.636
+    },
+    {
+      "epoch": 0.025241379310344828,
+      "grad_norm": 8.749399185180664,
+      "learning_rate": 1.4556390977443611e-05,
+      "loss": 0.5060253143310547,
+      "num_input_tokens_seen": 177578,
+      "step": 549,
+      "train_runtime": 1436.0715,
+      "train_tokens_per_second": 123.655
+    },
+    {
+      "epoch": 0.02528735632183908,
+      "grad_norm": 7.449631214141846,
+      "learning_rate": 1.4546365914786968e-05,
+      "loss": 0.46381819248199463,
+      "num_input_tokens_seen": 177886,
+      "step": 550,
+      "train_runtime": 1438.0329,
+      "train_tokens_per_second": 123.701
+    },
+    {
+      "epoch": 0.025333333333333333,
+      "grad_norm": 6.663496017456055,
+      "learning_rate": 1.4536340852130327e-05,
+      "loss": 0.42148929834365845,
+      "num_input_tokens_seen": 178454,
+      "step": 551,
+      "train_runtime": 1440.0625,
+      "train_tokens_per_second": 123.921
+    },
+    {
+      "epoch": 0.025379310344827585,
+      "grad_norm": 7.610936164855957,
+      "learning_rate": 1.4526315789473687e-05,
+      "loss": 0.42065486311912537,
+      "num_input_tokens_seen": 178674,
+      "step": 552,
+      "train_runtime": 1442.0126,
+      "train_tokens_per_second": 123.906
+    },
+    {
+      "epoch": 0.025425287356321838,
+      "grad_norm": 7.692618370056152,
+      "learning_rate": 1.4516290726817044e-05,
+      "loss": 0.5608479976654053,
+      "num_input_tokens_seen": 179156,
+      "step": 553,
+      "train_runtime": 1444.0085,
+      "train_tokens_per_second": 124.069
+    },
+    {
+      "epoch": 0.02547126436781609,
+      "grad_norm": 8.434357643127441,
+      "learning_rate": 1.4506265664160402e-05,
+      "loss": 0.4691472053527832,
+      "num_input_tokens_seen": 179438,
+      "step": 554,
+      "train_runtime": 1446.0222,
+      "train_tokens_per_second": 124.091
+    },
+    {
+      "epoch": 0.025517241379310347,
+      "grad_norm": 8.212077140808105,
+      "learning_rate": 1.4496240601503762e-05,
+      "loss": 0.42718926072120667,
+      "num_input_tokens_seen": 179764,
+      "step": 555,
+      "train_runtime": 1448.1529,
+      "train_tokens_per_second": 124.133
+    },
+    {
+      "epoch": 0.0255632183908046,
+      "grad_norm": 10.93320083618164,
+      "learning_rate": 1.4486215538847119e-05,
+      "loss": 0.5911573171615601,
+      "num_input_tokens_seen": 180104,
+      "step": 556,
+      "train_runtime": 1450.1192,
+      "train_tokens_per_second": 124.199
+    },
+    {
+      "epoch": 0.025609195402298852,
+      "grad_norm": 8.527013778686523,
+      "learning_rate": 1.4476190476190478e-05,
+      "loss": 0.44520294666290283,
+      "num_input_tokens_seen": 180362,
+      "step": 557,
+      "train_runtime": 1452.0722,
+      "train_tokens_per_second": 124.21
+    },
+    {
+      "epoch": 0.025655172413793104,
+      "grad_norm": 8.891039848327637,
+      "learning_rate": 1.4466165413533834e-05,
+      "loss": 0.6029215455055237,
+      "num_input_tokens_seen": 180728,
+      "step": 558,
+      "train_runtime": 1454.036,
+      "train_tokens_per_second": 124.294
+    },
+    {
+      "epoch": 0.025701149425287357,
+      "grad_norm": 9.30233097076416,
+      "learning_rate": 1.4456140350877195e-05,
+      "loss": 0.42037826776504517,
+      "num_input_tokens_seen": 181030,
+      "step": 559,
+      "train_runtime": 1456.0526,
+      "train_tokens_per_second": 124.329
+    },
+    {
+      "epoch": 0.02574712643678161,
+      "grad_norm": 8.955443382263184,
+      "learning_rate": 1.4446115288220553e-05,
+      "loss": 0.5944176316261292,
+      "num_input_tokens_seen": 181490,
+      "step": 560,
+      "train_runtime": 1458.045,
+      "train_tokens_per_second": 124.475
+    },
+    {
+      "epoch": 0.025793103448275862,
+      "grad_norm": 7.375617504119873,
+      "learning_rate": 1.443609022556391e-05,
+      "loss": 0.5741697549819946,
+      "num_input_tokens_seen": 181822,
+      "step": 561,
+      "train_runtime": 1460.0485,
+      "train_tokens_per_second": 124.531
+    },
+    {
+      "epoch": 0.025839080459770115,
+      "grad_norm": 8.91295337677002,
+      "learning_rate": 1.442606516290727e-05,
+      "loss": 0.6684648990631104,
+      "num_input_tokens_seen": 182246,
+      "step": 562,
+      "train_runtime": 1462.0756,
+      "train_tokens_per_second": 124.649
+    },
+    {
+      "epoch": 0.025885057471264367,
+      "grad_norm": 10.833512306213379,
+      "learning_rate": 1.4416040100250628e-05,
+      "loss": 0.496807724237442,
+      "num_input_tokens_seen": 182530,
+      "step": 563,
+      "train_runtime": 1464.0502,
+      "train_tokens_per_second": 124.675
+    },
+    {
+      "epoch": 0.02593103448275862,
+      "grad_norm": 10.345937728881836,
+      "learning_rate": 1.4406015037593985e-05,
+      "loss": 0.5157930254936218,
+      "num_input_tokens_seen": 182780,
+      "step": 564,
+      "train_runtime": 1466.012,
+      "train_tokens_per_second": 124.678
+    },
+    {
+      "epoch": 0.025977011494252873,
+      "grad_norm": 9.86699104309082,
+      "learning_rate": 1.4395989974937345e-05,
+      "loss": 0.4712110161781311,
+      "num_input_tokens_seen": 183010,
+      "step": 565,
+      "train_runtime": 1467.9732,
+      "train_tokens_per_second": 124.668
+    },
+    {
+      "epoch": 0.026022988505747125,
+      "grad_norm": 9.507144927978516,
+      "learning_rate": 1.4385964912280704e-05,
+      "loss": 0.5361157655715942,
+      "num_input_tokens_seen": 183318,
+      "step": 566,
+      "train_runtime": 1469.9285,
+      "train_tokens_per_second": 124.712
+    },
+    {
+      "epoch": 0.026068965517241378,
+      "grad_norm": 6.866785526275635,
+      "learning_rate": 1.437593984962406e-05,
+      "loss": 0.47417891025543213,
+      "num_input_tokens_seen": 183616,
+      "step": 567,
+      "train_runtime": 1471.8777,
+      "train_tokens_per_second": 124.749
+    },
+    {
+      "epoch": 0.026114942528735634,
+      "grad_norm": 7.958737373352051,
+      "learning_rate": 1.436591478696742e-05,
+      "loss": 0.4230140149593353,
+      "num_input_tokens_seen": 183896,
+      "step": 568,
+      "train_runtime": 1473.9123,
+      "train_tokens_per_second": 124.767
+    },
+    {
+      "epoch": 0.026160919540229886,
+      "grad_norm": 8.128926277160645,
+      "learning_rate": 1.435588972431078e-05,
+      "loss": 0.44361788034439087,
+      "num_input_tokens_seen": 184138,
+      "step": 569,
+      "train_runtime": 1475.8843,
+      "train_tokens_per_second": 124.765
+    },
+    {
+      "epoch": 0.02620689655172414,
+      "grad_norm": 7.628896236419678,
+      "learning_rate": 1.4345864661654136e-05,
+      "loss": 0.5449480414390564,
+      "num_input_tokens_seen": 184428,
+      "step": 570,
+      "train_runtime": 1477.8452,
+      "train_tokens_per_second": 124.795
+    },
+    {
+      "epoch": 0.02625287356321839,
+      "grad_norm": 7.5131988525390625,
+      "learning_rate": 1.4335839598997495e-05,
+      "loss": 0.4566008150577545,
+      "num_input_tokens_seen": 184716,
+      "step": 571,
+      "train_runtime": 1492.8443,
+      "train_tokens_per_second": 123.734
+    },
+    {
+      "epoch": 0.026298850574712644,
+      "grad_norm": 8.671358108520508,
+      "learning_rate": 1.4325814536340855e-05,
+      "loss": 0.4438135027885437,
+      "num_input_tokens_seen": 185020,
+      "step": 572,
+      "train_runtime": 1494.9389,
+      "train_tokens_per_second": 123.764
+    },
+    {
+      "epoch": 0.026344827586206897,
+      "grad_norm": 7.389676570892334,
+      "learning_rate": 1.4315789473684212e-05,
+      "loss": 0.48613792657852173,
+      "num_input_tokens_seen": 185470,
+      "step": 573,
+      "train_runtime": 1497.0691,
+      "train_tokens_per_second": 123.889
+    },
+    {
+      "epoch": 0.02639080459770115,
+      "grad_norm": 6.6500139236450195,
+      "learning_rate": 1.430576441102757e-05,
+      "loss": 0.3395925760269165,
+      "num_input_tokens_seen": 185710,
+      "step": 574,
+      "train_runtime": 1499.023,
+      "train_tokens_per_second": 123.887
+    },
+    {
+      "epoch": 0.026436781609195402,
+      "grad_norm": 8.409253120422363,
+      "learning_rate": 1.4295739348370929e-05,
+      "loss": 0.5716714859008789,
+      "num_input_tokens_seen": 186002,
+      "step": 575,
+      "train_runtime": 1501.1847,
+      "train_tokens_per_second": 123.903
+    },
+    {
+      "epoch": 0.026482758620689655,
+      "grad_norm": 6.835357666015625,
+      "learning_rate": 1.4285714285714287e-05,
+      "loss": 0.4041152596473694,
+      "num_input_tokens_seen": 186314,
+      "step": 576,
+      "train_runtime": 1503.2963,
+      "train_tokens_per_second": 123.937
+    },
+    {
+      "epoch": 0.026528735632183907,
+      "grad_norm": 9.923406600952148,
+      "learning_rate": 1.4275689223057646e-05,
+      "loss": 0.43302974104881287,
+      "num_input_tokens_seen": 186570,
+      "step": 577,
+      "train_runtime": 1505.3873,
+      "train_tokens_per_second": 123.935
+    },
+    {
+      "epoch": 0.02657471264367816,
+      "grad_norm": 7.4889655113220215,
+      "learning_rate": 1.4265664160401002e-05,
+      "loss": 0.4781152009963989,
+      "num_input_tokens_seen": 186848,
+      "step": 578,
+      "train_runtime": 1507.4908,
+      "train_tokens_per_second": 123.946
+    },
+    {
+      "epoch": 0.026620689655172412,
+      "grad_norm": 6.901342391967773,
+      "learning_rate": 1.4255639097744363e-05,
+      "loss": 0.4876782298088074,
+      "num_input_tokens_seen": 187148,
+      "step": 579,
+      "train_runtime": 1509.4421,
+      "train_tokens_per_second": 123.985
+    },
+    {
+      "epoch": 0.02666666666666667,
+      "grad_norm": 7.2897257804870605,
+      "learning_rate": 1.4245614035087721e-05,
+      "loss": 0.4617875814437866,
+      "num_input_tokens_seen": 187454,
+      "step": 580,
+      "train_runtime": 1511.411,
+      "train_tokens_per_second": 124.026
+    },
+    {
+      "epoch": 0.02671264367816092,
+      "grad_norm": 7.180599212646484,
+      "learning_rate": 1.4235588972431078e-05,
+      "loss": 0.42690128087997437,
+      "num_input_tokens_seen": 187736,
+      "step": 581,
+      "train_runtime": 1513.3662,
+      "train_tokens_per_second": 124.052
+    },
+    {
+      "epoch": 0.026758620689655174,
+      "grad_norm": 7.27871561050415,
+      "learning_rate": 1.4225563909774438e-05,
+      "loss": 0.4229961931705475,
+      "num_input_tokens_seen": 188106,
+      "step": 582,
+      "train_runtime": 1515.3431,
+      "train_tokens_per_second": 124.134
+    },
+    {
+      "epoch": 0.026804597701149426,
+      "grad_norm": 8.241268157958984,
+      "learning_rate": 1.4215538847117797e-05,
+      "loss": 0.3600757122039795,
+      "num_input_tokens_seen": 188354,
+      "step": 583,
+      "train_runtime": 1517.3128,
+      "train_tokens_per_second": 124.137
+    },
+    {
+      "epoch": 0.02685057471264368,
+      "grad_norm": 8.746855735778809,
+      "learning_rate": 1.4205513784461153e-05,
+      "loss": 0.6030400991439819,
+      "num_input_tokens_seen": 188624,
+      "step": 584,
+      "train_runtime": 1519.2549,
+      "train_tokens_per_second": 124.156
+    },
+    {
+      "epoch": 0.02689655172413793,
+      "grad_norm": 8.419955253601074,
+      "learning_rate": 1.4195488721804514e-05,
+      "loss": 0.473530650138855,
+      "num_input_tokens_seen": 188890,
+      "step": 585,
+      "train_runtime": 1521.1927,
+      "train_tokens_per_second": 124.172
+    },
+    {
+      "epoch": 0.026942528735632184,
+      "grad_norm": 6.791111469268799,
+      "learning_rate": 1.4185463659147872e-05,
+      "loss": 0.4334329664707184,
+      "num_input_tokens_seen": 189168,
+      "step": 586,
+      "train_runtime": 1523.136,
+      "train_tokens_per_second": 124.196
+    },
+    {
+      "epoch": 0.026988505747126437,
+      "grad_norm": 8.257518768310547,
+      "learning_rate": 1.4175438596491229e-05,
+      "loss": 0.4887130856513977,
+      "num_input_tokens_seen": 189486,
+      "step": 587,
+      "train_runtime": 1525.0838,
+      "train_tokens_per_second": 124.246
+    },
+    {
+      "epoch": 0.02703448275862069,
+      "grad_norm": 7.643197059631348,
+      "learning_rate": 1.4165413533834587e-05,
+      "loss": 0.45197662711143494,
+      "num_input_tokens_seen": 189752,
+      "step": 588,
+      "train_runtime": 1527.0541,
+      "train_tokens_per_second": 124.26
+    },
+    {
+      "epoch": 0.027080459770114942,
+      "grad_norm": 7.756775379180908,
+      "learning_rate": 1.4155388471177946e-05,
+      "loss": 0.33536070585250854,
+      "num_input_tokens_seen": 189996,
+      "step": 589,
+      "train_runtime": 1529.0022,
+      "train_tokens_per_second": 124.261
+    },
+    {
+      "epoch": 0.027126436781609194,
+      "grad_norm": 8.067883491516113,
+      "learning_rate": 1.4145363408521304e-05,
+      "loss": 0.438079297542572,
+      "num_input_tokens_seen": 190362,
+      "step": 590,
+      "train_runtime": 1530.98,
+      "train_tokens_per_second": 124.34
+    },
+    {
+      "epoch": 0.027172413793103447,
+      "grad_norm": 8.168673515319824,
+      "learning_rate": 1.4135338345864663e-05,
+      "loss": 0.48280560970306396,
+      "num_input_tokens_seen": 190668,
+      "step": 591,
+      "train_runtime": 1532.9365,
+      "train_tokens_per_second": 124.381
+    },
+    {
+      "epoch": 0.0272183908045977,
+      "grad_norm": 8.670255661010742,
+      "learning_rate": 1.4125313283208021e-05,
+      "loss": 0.2864142954349518,
+      "num_input_tokens_seen": 190886,
+      "step": 592,
+      "train_runtime": 1534.8748,
+      "train_tokens_per_second": 124.366
+    },
+    {
+      "epoch": 0.027264367816091956,
+      "grad_norm": 7.708620548248291,
+      "learning_rate": 1.411528822055138e-05,
+      "loss": 0.5283160209655762,
+      "num_input_tokens_seen": 191146,
+      "step": 593,
+      "train_runtime": 1536.9675,
+      "train_tokens_per_second": 124.366
+    },
+    {
+      "epoch": 0.027310344827586208,
+      "grad_norm": 6.311042785644531,
+      "learning_rate": 1.4105263157894738e-05,
+      "loss": 0.3583904206752777,
+      "num_input_tokens_seen": 191414,
+      "step": 594,
+      "train_runtime": 1538.9637,
+      "train_tokens_per_second": 124.379
+    },
+    {
+      "epoch": 0.02735632183908046,
+      "grad_norm": 7.285036563873291,
+      "learning_rate": 1.4095238095238097e-05,
+      "loss": 0.48402440547943115,
+      "num_input_tokens_seen": 191676,
+      "step": 595,
+      "train_runtime": 1540.9128,
+      "train_tokens_per_second": 124.391
+    },
+    {
+      "epoch": 0.027402298850574713,
+      "grad_norm": 7.623536586761475,
+      "learning_rate": 1.4085213032581455e-05,
+      "loss": 0.513954222202301,
+      "num_input_tokens_seen": 191994,
+      "step": 596,
+      "train_runtime": 1542.8809,
+      "train_tokens_per_second": 124.439
+    },
+    {
+      "epoch": 0.027448275862068966,
+      "grad_norm": 9.271323204040527,
+      "learning_rate": 1.4075187969924814e-05,
+      "loss": 0.542148232460022,
+      "num_input_tokens_seen": 192296,
+      "step": 597,
+      "train_runtime": 1544.8451,
+      "train_tokens_per_second": 124.476
+    },
+    {
+      "epoch": 0.02749425287356322,
+      "grad_norm": 8.006948471069336,
+      "learning_rate": 1.406516290726817e-05,
+      "loss": 0.32228660583496094,
+      "num_input_tokens_seen": 192524,
+      "step": 598,
+      "train_runtime": 1546.8082,
+      "train_tokens_per_second": 124.465
+    },
+    {
+      "epoch": 0.02754022988505747,
+      "grad_norm": 7.93782377243042,
+      "learning_rate": 1.405513784461153e-05,
+      "loss": 0.46611687541007996,
+      "num_input_tokens_seen": 192918,
+      "step": 599,
+      "train_runtime": 1548.8247,
+      "train_tokens_per_second": 124.558
+    },
+    {
+      "epoch": 0.027586206896551724,
+      "grad_norm": 8.049100875854492,
+      "learning_rate": 1.4045112781954887e-05,
+      "loss": 0.5331852436065674,
+      "num_input_tokens_seen": 193350,
+      "step": 600,
+      "train_runtime": 1550.823,
+      "train_tokens_per_second": 124.676
+    },
+    {
+      "epoch": 0.027586206896551724,
+      "eval_loss": 1.7835500240325928,
+      "eval_runtime": 54.7603,
+      "eval_samples_per_second": 18.261,
+      "eval_steps_per_second": 9.131,
+      "num_input_tokens_seen": 193350,
+      "step": 600
+    },
+    {
+      "epoch": 0.027632183908045976,
+      "grad_norm": 7.058294773101807,
+      "learning_rate": 1.4035087719298246e-05,
+      "loss": 0.38884398341178894,
+      "num_input_tokens_seen": 193646,
+      "step": 601,
+      "train_runtime": 1620.5228,
+      "train_tokens_per_second": 119.496
+    },
+    {
+      "epoch": 0.02767816091954023,
+      "grad_norm": 7.018427848815918,
+      "learning_rate": 1.4025062656641606e-05,
+      "loss": 0.35183781385421753,
+      "num_input_tokens_seen": 193842,
+      "step": 602,
+      "train_runtime": 1622.4683,
+      "train_tokens_per_second": 119.474
+    },
+    {
+      "epoch": 0.02772413793103448,
+      "grad_norm": 7.211747169494629,
+      "learning_rate": 1.4015037593984963e-05,
+      "loss": 0.5095769166946411,
+      "num_input_tokens_seen": 194138,
+      "step": 603,
+      "train_runtime": 1624.4303,
+      "train_tokens_per_second": 119.511
+    },
+    {
+      "epoch": 0.027770114942528734,
+      "grad_norm": 6.4296064376831055,
+      "learning_rate": 1.4005012531328321e-05,
+      "loss": 0.3702557682991028,
+      "num_input_tokens_seen": 194386,
+      "step": 604,
+      "train_runtime": 1626.3889,
+      "train_tokens_per_second": 119.52
+    },
+    {
+      "epoch": 0.02781609195402299,
+      "grad_norm": 7.240848064422607,
+      "learning_rate": 1.3994987468671682e-05,
+      "loss": 0.47583869099617004,
+      "num_input_tokens_seen": 194768,
+      "step": 605,
+      "train_runtime": 1628.4039,
+      "train_tokens_per_second": 119.607
+    },
+    {
+      "epoch": 0.027862068965517243,
+      "grad_norm": 7.305069923400879,
+      "learning_rate": 1.3984962406015038e-05,
+      "loss": 0.3479224443435669,
+      "num_input_tokens_seen": 195064,
+      "step": 606,
+      "train_runtime": 1630.4025,
+      "train_tokens_per_second": 119.642
+    },
+    {
+      "epoch": 0.027908045977011495,
+      "grad_norm": 13.084811210632324,
+      "learning_rate": 1.3974937343358397e-05,
+      "loss": 0.6538516879081726,
+      "num_input_tokens_seen": 195348,
+      "step": 607,
+      "train_runtime": 1632.362,
+      "train_tokens_per_second": 119.672
+    },
+    {
+      "epoch": 0.027954022988505748,
+      "grad_norm": 6.224904537200928,
+      "learning_rate": 1.3964912280701755e-05,
+      "loss": 0.34925347566604614,
+      "num_input_tokens_seen": 195622,
+      "step": 608,
+      "train_runtime": 1634.3551,
+      "train_tokens_per_second": 119.694
+    },
+    {
+      "epoch": 0.028,
+      "grad_norm": 8.950102806091309,
+      "learning_rate": 1.3954887218045114e-05,
+      "loss": 0.4487617611885071,
+      "num_input_tokens_seen": 195882,
+      "step": 609,
+      "train_runtime": 1636.3192,
+      "train_tokens_per_second": 119.709
+    },
+    {
+      "epoch": 0.028045977011494253,
+      "grad_norm": 5.966568946838379,
+      "learning_rate": 1.3944862155388472e-05,
+      "loss": 0.3355112671852112,
+      "num_input_tokens_seen": 196228,
+      "step": 610,
+      "train_runtime": 1638.3262,
+      "train_tokens_per_second": 119.773
+    },
+    {
+      "epoch": 0.028091954022988506,
+      "grad_norm": 11.133930206298828,
+      "learning_rate": 1.3934837092731829e-05,
+      "loss": 0.5155819654464722,
+      "num_input_tokens_seen": 196598,
+      "step": 611,
+      "train_runtime": 1640.3305,
+      "train_tokens_per_second": 119.853
+    },
+    {
+      "epoch": 0.02813793103448276,
+      "grad_norm": 7.472561836242676,
+      "learning_rate": 1.392481203007519e-05,
+      "loss": 0.5794181227684021,
+      "num_input_tokens_seen": 196916,
+      "step": 612,
+      "train_runtime": 1642.3962,
+      "train_tokens_per_second": 119.896
+    },
+    {
+      "epoch": 0.02818390804597701,
+      "grad_norm": 7.21183967590332,
+      "learning_rate": 1.3914786967418548e-05,
+      "loss": 0.4837031662464142,
+      "num_input_tokens_seen": 197190,
+      "step": 613,
+      "train_runtime": 1644.549,
+      "train_tokens_per_second": 119.905
+    },
+    {
+      "epoch": 0.028229885057471264,
+      "grad_norm": 9.273664474487305,
+      "learning_rate": 1.3904761904761905e-05,
+      "loss": 0.3903869390487671,
+      "num_input_tokens_seen": 197480,
+      "step": 614,
+      "train_runtime": 1646.6588,
+      "train_tokens_per_second": 119.928
+    },
+    {
+      "epoch": 0.028275862068965516,
+      "grad_norm": 9.718978881835938,
+      "learning_rate": 1.3894736842105265e-05,
+      "loss": 0.6804496049880981,
+      "num_input_tokens_seen": 197816,
+      "step": 615,
+      "train_runtime": 1648.6269,
+      "train_tokens_per_second": 119.988
+    },
+    {
+      "epoch": 0.02832183908045977,
+      "grad_norm": 8.028841972351074,
+      "learning_rate": 1.3884711779448623e-05,
+      "loss": 0.47770822048187256,
+      "num_input_tokens_seen": 198178,
+      "step": 616,
+      "train_runtime": 1650.6149,
+      "train_tokens_per_second": 120.063
+    },
+    {
+      "epoch": 0.02836781609195402,
+      "grad_norm": 9.076641082763672,
+      "learning_rate": 1.387468671679198e-05,
+      "loss": 0.4381833076477051,
+      "num_input_tokens_seen": 198434,
+      "step": 617,
+      "train_runtime": 1652.564,
+      "train_tokens_per_second": 120.076
+    },
+    {
+      "epoch": 0.028413793103448277,
+      "grad_norm": 8.21583080291748,
+      "learning_rate": 1.3864661654135339e-05,
+      "loss": 0.48954522609710693,
+      "num_input_tokens_seen": 198828,
+      "step": 618,
+      "train_runtime": 1654.5543,
+      "train_tokens_per_second": 120.17
+    },
+    {
+      "epoch": 0.02845977011494253,
+      "grad_norm": 8.827544212341309,
+      "learning_rate": 1.3854636591478699e-05,
+      "loss": 0.47703176736831665,
+      "num_input_tokens_seen": 199120,
+      "step": 619,
+      "train_runtime": 1656.5184,
+      "train_tokens_per_second": 120.204
+    },
+    {
+      "epoch": 0.028505747126436783,
+      "grad_norm": 6.700578212738037,
+      "learning_rate": 1.3844611528822056e-05,
+      "loss": 0.44256260991096497,
+      "num_input_tokens_seen": 199418,
+      "step": 620,
+      "train_runtime": 1658.4664,
+      "train_tokens_per_second": 120.242
+    },
+    {
+      "epoch": 0.028551724137931035,
+      "grad_norm": 10.429258346557617,
+      "learning_rate": 1.3834586466165414e-05,
+      "loss": 0.5015515089035034,
+      "num_input_tokens_seen": 199744,
+      "step": 621,
+      "train_runtime": 1660.5068,
+      "train_tokens_per_second": 120.291
+    },
+    {
+      "epoch": 0.028597701149425288,
+      "grad_norm": 6.868889331817627,
+      "learning_rate": 1.3824561403508774e-05,
+      "loss": 0.43400683999061584,
+      "num_input_tokens_seen": 200088,
+      "step": 622,
+      "train_runtime": 1662.5546,
+      "train_tokens_per_second": 120.35
+    },
+    {
+      "epoch": 0.02864367816091954,
+      "grad_norm": 8.866421699523926,
+      "learning_rate": 1.3814536340852131e-05,
+      "loss": 0.3786638677120209,
+      "num_input_tokens_seen": 200356,
+      "step": 623,
+      "train_runtime": 1664.5177,
+      "train_tokens_per_second": 120.369
+    },
+    {
+      "epoch": 0.028689655172413793,
+      "grad_norm": 7.716616153717041,
+      "learning_rate": 1.380451127819549e-05,
+      "loss": 0.5063319206237793,
+      "num_input_tokens_seen": 200636,
+      "step": 624,
+      "train_runtime": 1666.4893,
+      "train_tokens_per_second": 120.394
+    },
+    {
+      "epoch": 0.028735632183908046,
+      "grad_norm": 7.845930576324463,
+      "learning_rate": 1.379448621553885e-05,
+      "loss": 0.5226324796676636,
+      "num_input_tokens_seen": 201086,
+      "step": 625,
+      "train_runtime": 1668.502,
+      "train_tokens_per_second": 120.519
+    },
+    {
+      "epoch": 0.028781609195402298,
+      "grad_norm": 8.0442476272583,
+      "learning_rate": 1.3784461152882206e-05,
+      "loss": 0.5390662550926208,
+      "num_input_tokens_seen": 201526,
+      "step": 626,
+      "train_runtime": 1670.499,
+      "train_tokens_per_second": 120.638
+    },
+    {
+      "epoch": 0.02882758620689655,
+      "grad_norm": 7.399601459503174,
+      "learning_rate": 1.3774436090225565e-05,
+      "loss": 0.42423149943351746,
+      "num_input_tokens_seen": 201884,
+      "step": 627,
+      "train_runtime": 1672.4675,
+      "train_tokens_per_second": 120.71
+    },
+    {
+      "epoch": 0.028873563218390803,
+      "grad_norm": 7.964285850524902,
+      "learning_rate": 1.3764411027568922e-05,
+      "loss": 0.4824962019920349,
+      "num_input_tokens_seen": 202188,
+      "step": 628,
+      "train_runtime": 1674.4263,
+      "train_tokens_per_second": 120.751
+    },
+    {
+      "epoch": 0.028919540229885056,
+      "grad_norm": 8.684676170349121,
+      "learning_rate": 1.3754385964912282e-05,
+      "loss": 0.587040901184082,
+      "num_input_tokens_seen": 202648,
+      "step": 629,
+      "train_runtime": 1676.4966,
+      "train_tokens_per_second": 120.876
+    },
+    {
+      "epoch": 0.028965517241379312,
+      "grad_norm": 7.471075057983398,
+      "learning_rate": 1.374436090225564e-05,
+      "loss": 0.48896095156669617,
+      "num_input_tokens_seen": 202972,
+      "step": 630,
+      "train_runtime": 1678.4583,
+      "train_tokens_per_second": 120.928
+    },
+    {
+      "epoch": 0.029011494252873565,
+      "grad_norm": 7.523292541503906,
+      "learning_rate": 1.3734335839598997e-05,
+      "loss": 0.49141925573349,
+      "num_input_tokens_seen": 203302,
+      "step": 631,
+      "train_runtime": 1693.1094,
+      "train_tokens_per_second": 120.076
+    },
+    {
+      "epoch": 0.029057471264367817,
+      "grad_norm": 7.045520782470703,
+      "learning_rate": 1.3724310776942357e-05,
+      "loss": 0.515127420425415,
+      "num_input_tokens_seen": 203906,
+      "step": 632,
+      "train_runtime": 1695.2235,
+      "train_tokens_per_second": 120.283
+    },
+    {
+      "epoch": 0.02910344827586207,
+      "grad_norm": 6.967346668243408,
+      "learning_rate": 1.3714285714285716e-05,
+      "loss": 0.4676874876022339,
+      "num_input_tokens_seen": 204196,
+      "step": 633,
+      "train_runtime": 1697.2192,
+      "train_tokens_per_second": 120.312
+    },
+    {
+      "epoch": 0.029149425287356322,
+      "grad_norm": 7.842945575714111,
+      "learning_rate": 1.3704260651629073e-05,
+      "loss": 0.5667434334754944,
+      "num_input_tokens_seen": 204538,
+      "step": 634,
+      "train_runtime": 1699.2623,
+      "train_tokens_per_second": 120.369
+    },
+    {
+      "epoch": 0.029195402298850575,
+      "grad_norm": 10.040627479553223,
+      "learning_rate": 1.3694235588972433e-05,
+      "loss": 0.28221577405929565,
+      "num_input_tokens_seen": 204780,
+      "step": 635,
+      "train_runtime": 1701.2622,
+      "train_tokens_per_second": 120.369
+    },
+    {
+      "epoch": 0.029241379310344828,
+      "grad_norm": 8.622676849365234,
+      "learning_rate": 1.3684210526315791e-05,
+      "loss": 0.5731203556060791,
+      "num_input_tokens_seen": 205136,
+      "step": 636,
+      "train_runtime": 1703.31,
+      "train_tokens_per_second": 120.434
+    },
+    {
+      "epoch": 0.02928735632183908,
+      "grad_norm": 9.108221054077148,
+      "learning_rate": 1.3674185463659148e-05,
+      "loss": 0.5550015568733215,
+      "num_input_tokens_seen": 205418,
+      "step": 637,
+      "train_runtime": 1705.2624,
+      "train_tokens_per_second": 120.461
+    },
+    {
+      "epoch": 0.029333333333333333,
+      "grad_norm": 6.611140251159668,
+      "learning_rate": 1.3664160401002507e-05,
+      "loss": 0.3855544924736023,
+      "num_input_tokens_seen": 205746,
+      "step": 638,
+      "train_runtime": 1707.3309,
+      "train_tokens_per_second": 120.507
+    },
+    {
+      "epoch": 0.029379310344827585,
+      "grad_norm": 6.947152137756348,
+      "learning_rate": 1.3654135338345867e-05,
+      "loss": 0.399821937084198,
+      "num_input_tokens_seen": 206056,
+      "step": 639,
+      "train_runtime": 1709.3312,
+      "train_tokens_per_second": 120.548
+    },
+    {
+      "epoch": 0.029425287356321838,
+      "grad_norm": 7.046751022338867,
+      "learning_rate": 1.3644110275689224e-05,
+      "loss": 0.6221940517425537,
+      "num_input_tokens_seen": 206572,
+      "step": 640,
+      "train_runtime": 1711.3431,
+      "train_tokens_per_second": 120.708
+    },
+    {
+      "epoch": 0.02947126436781609,
+      "grad_norm": 8.085256576538086,
+      "learning_rate": 1.3634085213032582e-05,
+      "loss": 0.5442467927932739,
+      "num_input_tokens_seen": 206826,
+      "step": 641,
+      "train_runtime": 1713.276,
+      "train_tokens_per_second": 120.72
+    },
+    {
+      "epoch": 0.029517241379310343,
+      "grad_norm": 9.777151107788086,
+      "learning_rate": 1.3624060150375942e-05,
+      "loss": 0.5079631209373474,
+      "num_input_tokens_seen": 207134,
+      "step": 642,
+      "train_runtime": 1715.221,
+      "train_tokens_per_second": 120.762
+    },
+    {
+      "epoch": 0.0295632183908046,
+      "grad_norm": 8.128308296203613,
+      "learning_rate": 1.3614035087719299e-05,
+      "loss": 0.4377592206001282,
+      "num_input_tokens_seen": 207388,
+      "step": 643,
+      "train_runtime": 1717.174,
+      "train_tokens_per_second": 120.773
+    },
+    {
+      "epoch": 0.029609195402298852,
+      "grad_norm": 7.2447662353515625,
+      "learning_rate": 1.3604010025062658e-05,
+      "loss": 0.6402813792228699,
+      "num_input_tokens_seen": 207848,
+      "step": 644,
+      "train_runtime": 1719.1667,
+      "train_tokens_per_second": 120.9
+    },
+    {
+      "epoch": 0.029655172413793104,
+      "grad_norm": 6.542693614959717,
+      "learning_rate": 1.3593984962406018e-05,
+      "loss": 0.3917843997478485,
+      "num_input_tokens_seen": 208120,
+      "step": 645,
+      "train_runtime": 1721.1133,
+      "train_tokens_per_second": 120.922
+    },
+    {
+      "epoch": 0.029701149425287357,
+      "grad_norm": 9.332645416259766,
+      "learning_rate": 1.3583959899749374e-05,
+      "loss": 0.3541853129863739,
+      "num_input_tokens_seen": 208334,
+      "step": 646,
+      "train_runtime": 1723.0478,
+      "train_tokens_per_second": 120.91
+    },
+    {
+      "epoch": 0.02974712643678161,
+      "grad_norm": 8.380578994750977,
+      "learning_rate": 1.3573934837092733e-05,
+      "loss": 0.49306660890579224,
+      "num_input_tokens_seen": 208630,
+      "step": 647,
+      "train_runtime": 1725.0058,
+      "train_tokens_per_second": 120.945
+    },
+    {
+      "epoch": 0.029793103448275862,
+      "grad_norm": 7.993964195251465,
+      "learning_rate": 1.356390977443609e-05,
+      "loss": 0.4118781089782715,
+      "num_input_tokens_seen": 208944,
+      "step": 648,
+      "train_runtime": 1726.986,
+      "train_tokens_per_second": 120.988
+    },
+    {
+      "epoch": 0.029839080459770115,
+      "grad_norm": 7.586103439331055,
+      "learning_rate": 1.355388471177945e-05,
+      "loss": 0.4668120741844177,
+      "num_input_tokens_seen": 209338,
+      "step": 649,
+      "train_runtime": 1728.9893,
+      "train_tokens_per_second": 121.075
+    },
+    {
+      "epoch": 0.029885057471264367,
+      "grad_norm": 6.804312229156494,
+      "learning_rate": 1.3543859649122808e-05,
+      "loss": 0.4309317171573639,
+      "num_input_tokens_seen": 209632,
+      "step": 650,
+      "train_runtime": 1731.0188,
+      "train_tokens_per_second": 121.103
+    },
+    {
+      "epoch": 0.02993103448275862,
+      "grad_norm": 8.213543891906738,
+      "learning_rate": 1.3533834586466165e-05,
+      "loss": 0.4724840521812439,
+      "num_input_tokens_seen": 210066,
+      "step": 651,
+      "train_runtime": 1733.0485,
+      "train_tokens_per_second": 121.212
+    },
+    {
+      "epoch": 0.029977011494252873,
+      "grad_norm": 7.234861373901367,
+      "learning_rate": 1.3523809523809525e-05,
+      "loss": 0.4469287693500519,
+      "num_input_tokens_seen": 210416,
+      "step": 652,
+      "train_runtime": 1735.0305,
+      "train_tokens_per_second": 121.275
+    },
+    {
+      "epoch": 0.030022988505747125,
+      "grad_norm": 8.447935104370117,
+      "learning_rate": 1.3513784461152884e-05,
+      "loss": 0.4970450699329376,
+      "num_input_tokens_seen": 210684,
+      "step": 653,
+      "train_runtime": 1737.0663,
+      "train_tokens_per_second": 121.287
+    },
+    {
+      "epoch": 0.030068965517241378,
+      "grad_norm": 6.736108303070068,
+      "learning_rate": 1.350375939849624e-05,
+      "loss": 0.42876240611076355,
+      "num_input_tokens_seen": 211004,
+      "step": 654,
+      "train_runtime": 1739.0767,
+      "train_tokens_per_second": 121.331
+    },
+    {
+      "epoch": 0.030114942528735634,
+      "grad_norm": 8.359664916992188,
+      "learning_rate": 1.3493734335839601e-05,
+      "loss": 0.4727824330329895,
+      "num_input_tokens_seen": 211294,
+      "step": 655,
+      "train_runtime": 1741.0635,
+      "train_tokens_per_second": 121.359
+    },
+    {
+      "epoch": 0.030160919540229886,
+      "grad_norm": 6.793044090270996,
+      "learning_rate": 1.348370927318296e-05,
+      "loss": 0.4805864095687866,
+      "num_input_tokens_seen": 211812,
+      "step": 656,
+      "train_runtime": 1743.0757,
+      "train_tokens_per_second": 121.516
+    },
+    {
+      "epoch": 0.03020689655172414,
+      "grad_norm": 8.12999153137207,
+      "learning_rate": 1.3473684210526316e-05,
+      "loss": 0.5290884971618652,
+      "num_input_tokens_seen": 212122,
+      "step": 657,
+      "train_runtime": 1745.1085,
+      "train_tokens_per_second": 121.552
+    },
+    {
+      "epoch": 0.03025287356321839,
+      "grad_norm": 9.024924278259277,
+      "learning_rate": 1.3463659147869675e-05,
+      "loss": 0.4053267240524292,
+      "num_input_tokens_seen": 212400,
+      "step": 658,
+      "train_runtime": 1747.0701,
+      "train_tokens_per_second": 121.575
+    },
+    {
+      "epoch": 0.030298850574712644,
+      "grad_norm": 7.814553260803223,
+      "learning_rate": 1.3453634085213035e-05,
+      "loss": 0.41010433435440063,
+      "num_input_tokens_seen": 212696,
+      "step": 659,
+      "train_runtime": 1749.0533,
+      "train_tokens_per_second": 121.606
+    },
+    {
+      "epoch": 0.030344827586206897,
+      "grad_norm": 7.647894859313965,
+      "learning_rate": 1.3443609022556392e-05,
+      "loss": 0.5449205040931702,
+      "num_input_tokens_seen": 213122,
+      "step": 660,
+      "train_runtime": 1751.0473,
+      "train_tokens_per_second": 121.711
+    },
+    {
+      "epoch": 0.03039080459770115,
+      "grad_norm": 7.343043327331543,
+      "learning_rate": 1.343358395989975e-05,
+      "loss": 0.4413844645023346,
+      "num_input_tokens_seen": 213422,
+      "step": 661,
+      "train_runtime": 1766.4526,
+      "train_tokens_per_second": 120.82
+    },
+    {
+      "epoch": 0.030436781609195402,
+      "grad_norm": 7.271646976470947,
+      "learning_rate": 1.342355889724311e-05,
+      "loss": 0.4377230107784271,
+      "num_input_tokens_seen": 213646,
+      "step": 662,
+      "train_runtime": 1768.5021,
+      "train_tokens_per_second": 120.806
+    },
+    {
+      "epoch": 0.030482758620689655,
+      "grad_norm": 9.043072700500488,
+      "learning_rate": 1.3413533834586467e-05,
+      "loss": 0.6550434231758118,
+      "num_input_tokens_seen": 213968,
+      "step": 663,
+      "train_runtime": 1770.4631,
+      "train_tokens_per_second": 120.854
+    },
+    {
+      "epoch": 0.030528735632183907,
+      "grad_norm": 7.99835205078125,
+      "learning_rate": 1.3403508771929826e-05,
+      "loss": 0.4773554503917694,
+      "num_input_tokens_seen": 214184,
+      "step": 664,
+      "train_runtime": 1772.4789,
+      "train_tokens_per_second": 120.839
+    },
+    {
+      "epoch": 0.03057471264367816,
+      "grad_norm": 7.9440484046936035,
+      "learning_rate": 1.3393483709273186e-05,
+      "loss": 0.43905895948410034,
+      "num_input_tokens_seen": 214490,
+      "step": 665,
+      "train_runtime": 1774.4847,
+      "train_tokens_per_second": 120.875
+    },
+    {
+      "epoch": 0.030620689655172412,
+      "grad_norm": 9.001914024353027,
+      "learning_rate": 1.3383458646616543e-05,
+      "loss": 0.4448626637458801,
+      "num_input_tokens_seen": 214836,
+      "step": 666,
+      "train_runtime": 1776.5169,
+      "train_tokens_per_second": 120.931
+    },
+    {
+      "epoch": 0.030666666666666665,
+      "grad_norm": 6.8270487785339355,
+      "learning_rate": 1.3373433583959901e-05,
+      "loss": 0.31435269117355347,
+      "num_input_tokens_seen": 215078,
+      "step": 667,
+      "train_runtime": 1778.4645,
+      "train_tokens_per_second": 120.935
+    },
+    {
+      "epoch": 0.03071264367816092,
+      "grad_norm": 6.213423252105713,
+      "learning_rate": 1.3363408521303258e-05,
+      "loss": 0.41751205921173096,
+      "num_input_tokens_seen": 215408,
+      "step": 668,
+      "train_runtime": 1780.4957,
+      "train_tokens_per_second": 120.982
+    },
+    {
+      "epoch": 0.030758620689655174,
+      "grad_norm": 8.144896507263184,
+      "learning_rate": 1.3353383458646618e-05,
+      "loss": 0.439815878868103,
+      "num_input_tokens_seen": 215810,
+      "step": 669,
+      "train_runtime": 1782.4866,
+      "train_tokens_per_second": 121.072
+    },
+    {
+      "epoch": 0.030804597701149426,
+      "grad_norm": 6.508113861083984,
+      "learning_rate": 1.3343358395989977e-05,
+      "loss": 0.44854140281677246,
+      "num_input_tokens_seen": 216114,
+      "step": 670,
+      "train_runtime": 1784.4499,
+      "train_tokens_per_second": 121.11
+    },
+    {
+      "epoch": 0.03085057471264368,
+      "grad_norm": 7.181475639343262,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.458507776260376,
+      "num_input_tokens_seen": 216458,
+      "step": 671,
+      "train_runtime": 1786.4038,
+      "train_tokens_per_second": 121.17
+    },
+    {
+      "epoch": 0.03089655172413793,
+      "grad_norm": 7.080460071563721,
+      "learning_rate": 1.3323308270676693e-05,
+      "loss": 0.4886537790298462,
+      "num_input_tokens_seen": 216726,
+      "step": 672,
+      "train_runtime": 1788.3541,
+      "train_tokens_per_second": 121.187
+    },
+    {
+      "epoch": 0.030942528735632184,
+      "grad_norm": 8.234830856323242,
+      "learning_rate": 1.3313283208020052e-05,
+      "loss": 0.4340991973876953,
+      "num_input_tokens_seen": 217042,
+      "step": 673,
+      "train_runtime": 1790.3403,
+      "train_tokens_per_second": 121.229
+    },
+    {
+      "epoch": 0.030988505747126437,
+      "grad_norm": 7.746073246002197,
+      "learning_rate": 1.3303258145363409e-05,
+      "loss": 0.4000415802001953,
+      "num_input_tokens_seen": 217282,
+      "step": 674,
+      "train_runtime": 1792.3022,
+      "train_tokens_per_second": 121.231
+    },
+    {
+      "epoch": 0.03103448275862069,
+      "grad_norm": 6.61154842376709,
+      "learning_rate": 1.3293233082706769e-05,
+      "loss": 0.4018608033657074,
+      "num_input_tokens_seen": 217612,
+      "step": 675,
+      "train_runtime": 1794.2664,
+      "train_tokens_per_second": 121.282
+    },
+    {
+      "epoch": 0.031080459770114942,
+      "grad_norm": 6.929211616516113,
+      "learning_rate": 1.3283208020050127e-05,
+      "loss": 0.4212467074394226,
+      "num_input_tokens_seen": 217870,
+      "step": 676,
+      "train_runtime": 1796.2241,
+      "train_tokens_per_second": 121.293
+    },
+    {
+      "epoch": 0.031126436781609194,
+      "grad_norm": 7.709577560424805,
+      "learning_rate": 1.3273182957393484e-05,
+      "loss": 0.5369236469268799,
+      "num_input_tokens_seen": 218218,
+      "step": 677,
+      "train_runtime": 1798.1968,
+      "train_tokens_per_second": 121.354
+    },
+    {
+      "epoch": 0.031172413793103447,
+      "grad_norm": 7.48252534866333,
+      "learning_rate": 1.3263157894736843e-05,
+      "loss": 0.44973501563072205,
+      "num_input_tokens_seen": 218512,
+      "step": 678,
+      "train_runtime": 1800.1588,
+      "train_tokens_per_second": 121.385
+    },
+    {
+      "epoch": 0.0312183908045977,
+      "grad_norm": 9.012537956237793,
+      "learning_rate": 1.3253132832080203e-05,
+      "loss": 0.49290916323661804,
+      "num_input_tokens_seen": 218816,
+      "step": 679,
+      "train_runtime": 1802.1136,
+      "train_tokens_per_second": 121.422
+    },
+    {
+      "epoch": 0.03126436781609195,
+      "grad_norm": 6.668360710144043,
+      "learning_rate": 1.324310776942356e-05,
+      "loss": 0.4302211105823517,
+      "num_input_tokens_seen": 219348,
+      "step": 680,
+      "train_runtime": 1804.1024,
+      "train_tokens_per_second": 121.583
+    },
+    {
+      "epoch": 0.031310344827586205,
+      "grad_norm": 7.369109630584717,
+      "learning_rate": 1.3233082706766918e-05,
+      "loss": 0.42139798402786255,
+      "num_input_tokens_seen": 219638,
+      "step": 681,
+      "train_runtime": 1806.0623,
+      "train_tokens_per_second": 121.612
+    },
+    {
+      "epoch": 0.03135632183908046,
+      "grad_norm": 7.988426208496094,
+      "learning_rate": 1.3223057644110278e-05,
+      "loss": 0.47020840644836426,
+      "num_input_tokens_seen": 219914,
+      "step": 682,
+      "train_runtime": 1808.0479,
+      "train_tokens_per_second": 121.631
+    },
+    {
+      "epoch": 0.03140229885057471,
+      "grad_norm": 7.562366962432861,
+      "learning_rate": 1.3213032581453635e-05,
+      "loss": 0.43431341648101807,
+      "num_input_tokens_seen": 220154,
+      "step": 683,
+      "train_runtime": 1810.007,
+      "train_tokens_per_second": 121.632
+    },
+    {
+      "epoch": 0.03144827586206896,
+      "grad_norm": 7.61648416519165,
+      "learning_rate": 1.3203007518796994e-05,
+      "loss": 0.49406546354293823,
+      "num_input_tokens_seen": 220442,
+      "step": 684,
+      "train_runtime": 1811.9616,
+      "train_tokens_per_second": 121.659
+    },
+    {
+      "epoch": 0.031494252873563215,
+      "grad_norm": 6.642371654510498,
+      "learning_rate": 1.3192982456140354e-05,
+      "loss": 0.4200516641139984,
+      "num_input_tokens_seen": 220808,
+      "step": 685,
+      "train_runtime": 1813.9428,
+      "train_tokens_per_second": 121.728
+    },
+    {
+      "epoch": 0.031540229885057475,
+      "grad_norm": 7.991518020629883,
+      "learning_rate": 1.318295739348371e-05,
+      "loss": 0.3903042674064636,
+      "num_input_tokens_seen": 221096,
+      "step": 686,
+      "train_runtime": 1815.8973,
+      "train_tokens_per_second": 121.756
+    },
+    {
+      "epoch": 0.03158620689655173,
+      "grad_norm": 8.292240142822266,
+      "learning_rate": 1.3172932330827069e-05,
+      "loss": 0.4748704135417938,
+      "num_input_tokens_seen": 221604,
+      "step": 687,
+      "train_runtime": 1817.9524,
+      "train_tokens_per_second": 121.898
+    },
+    {
+      "epoch": 0.03163218390804598,
+      "grad_norm": 8.80848503112793,
+      "learning_rate": 1.3162907268170426e-05,
+      "loss": 0.5277791619300842,
+      "num_input_tokens_seen": 221920,
+      "step": 688,
+      "train_runtime": 1819.9307,
+      "train_tokens_per_second": 121.939
+    },
+    {
+      "epoch": 0.03167816091954023,
+      "grad_norm": 11.570622444152832,
+      "learning_rate": 1.3152882205513786e-05,
+      "loss": 0.4864501953125,
+      "num_input_tokens_seen": 222230,
+      "step": 689,
+      "train_runtime": 1821.917,
+      "train_tokens_per_second": 121.976
+    },
+    {
+      "epoch": 0.031724137931034485,
+      "grad_norm": 9.256178855895996,
+      "learning_rate": 1.3142857142857145e-05,
+      "loss": 0.5053106546401978,
+      "num_input_tokens_seen": 222514,
+      "step": 690,
+      "train_runtime": 1823.933,
+      "train_tokens_per_second": 121.997
+    },
+    {
+      "epoch": 0.03177011494252874,
+      "grad_norm": 7.844667911529541,
+      "learning_rate": 1.3132832080200501e-05,
+      "loss": 0.5310298800468445,
+      "num_input_tokens_seen": 222826,
+      "step": 691,
+      "train_runtime": 1838.7833,
+      "train_tokens_per_second": 121.181
+    },
+    {
+      "epoch": 0.03181609195402299,
+      "grad_norm": 8.256976127624512,
+      "learning_rate": 1.3122807017543862e-05,
+      "loss": 0.48716580867767334,
+      "num_input_tokens_seen": 223188,
+      "step": 692,
+      "train_runtime": 1840.7512,
+      "train_tokens_per_second": 121.248
+    },
+    {
+      "epoch": 0.03186206896551724,
+      "grad_norm": 7.259397983551025,
+      "learning_rate": 1.311278195488722e-05,
+      "loss": 0.36275187134742737,
+      "num_input_tokens_seen": 223438,
+      "step": 693,
+      "train_runtime": 1842.7643,
+      "train_tokens_per_second": 121.252
+    },
+    {
+      "epoch": 0.031908045977011495,
+      "grad_norm": 7.258387565612793,
+      "learning_rate": 1.3102756892230577e-05,
+      "loss": 0.45881396532058716,
+      "num_input_tokens_seen": 223742,
+      "step": 694,
+      "train_runtime": 1844.7174,
+      "train_tokens_per_second": 121.288
+    },
+    {
+      "epoch": 0.03195402298850575,
+      "grad_norm": 9.610076904296875,
+      "learning_rate": 1.3092731829573937e-05,
+      "loss": 0.48475897312164307,
+      "num_input_tokens_seen": 224166,
+      "step": 695,
+      "train_runtime": 1846.714,
+      "train_tokens_per_second": 121.386
+    },
+    {
+      "epoch": 0.032,
+      "grad_norm": 8.098437309265137,
+      "learning_rate": 1.3082706766917295e-05,
+      "loss": 0.5809595584869385,
+      "num_input_tokens_seen": 224542,
+      "step": 696,
+      "train_runtime": 1848.7338,
+      "train_tokens_per_second": 121.457
+    },
+    {
+      "epoch": 0.03204597701149425,
+      "grad_norm": 8.92160415649414,
+      "learning_rate": 1.3072681704260652e-05,
+      "loss": 0.5239991545677185,
+      "num_input_tokens_seen": 224844,
+      "step": 697,
+      "train_runtime": 1850.6894,
+      "train_tokens_per_second": 121.492
+    },
+    {
+      "epoch": 0.032091954022988506,
+      "grad_norm": 7.095712184906006,
+      "learning_rate": 1.306265664160401e-05,
+      "loss": 0.3632602095603943,
+      "num_input_tokens_seen": 225148,
+      "step": 698,
+      "train_runtime": 1852.6457,
+      "train_tokens_per_second": 121.528
+    },
+    {
+      "epoch": 0.03213793103448276,
+      "grad_norm": 7.637246608734131,
+      "learning_rate": 1.305263157894737e-05,
+      "loss": 0.34871622920036316,
+      "num_input_tokens_seen": 225412,
+      "step": 699,
+      "train_runtime": 1854.6345,
+      "train_tokens_per_second": 121.54
+    },
+    {
+      "epoch": 0.03218390804597701,
+      "grad_norm": 8.343865394592285,
+      "learning_rate": 1.3042606516290728e-05,
+      "loss": 0.44183212518692017,
+      "num_input_tokens_seen": 225652,
+      "step": 700,
+      "train_runtime": 1856.5783,
+      "train_tokens_per_second": 121.542
+    },
+    {
+      "epoch": 0.032229885057471264,
+      "grad_norm": 8.810320854187012,
+      "learning_rate": 1.3032581453634086e-05,
+      "loss": 0.43702125549316406,
+      "num_input_tokens_seen": 225910,
+      "step": 701,
+      "train_runtime": 1858.5502,
+      "train_tokens_per_second": 121.552
+    },
+    {
+      "epoch": 0.032275862068965516,
+      "grad_norm": 7.387729167938232,
+      "learning_rate": 1.3022556390977445e-05,
+      "loss": 0.4800367057323456,
+      "num_input_tokens_seen": 226198,
+      "step": 702,
+      "train_runtime": 1860.6122,
+      "train_tokens_per_second": 121.572
+    },
+    {
+      "epoch": 0.03232183908045977,
+      "grad_norm": 7.299515724182129,
+      "learning_rate": 1.3012531328320803e-05,
+      "loss": 0.5099941492080688,
+      "num_input_tokens_seen": 226596,
+      "step": 703,
+      "train_runtime": 1862.7084,
+      "train_tokens_per_second": 121.649
+    },
+    {
+      "epoch": 0.03236781609195402,
+      "grad_norm": 7.442165374755859,
+      "learning_rate": 1.3002506265664162e-05,
+      "loss": 0.43732163310050964,
+      "num_input_tokens_seen": 226820,
+      "step": 704,
+      "train_runtime": 1864.6852,
+      "train_tokens_per_second": 121.64
+    },
+    {
+      "epoch": 0.032413793103448274,
+      "grad_norm": 7.282296657562256,
+      "learning_rate": 1.299248120300752e-05,
+      "loss": 0.5136317014694214,
+      "num_input_tokens_seen": 227138,
+      "step": 705,
+      "train_runtime": 1866.6787,
+      "train_tokens_per_second": 121.68
+    },
+    {
+      "epoch": 0.03245977011494253,
+      "grad_norm": 7.085516452789307,
+      "learning_rate": 1.2982456140350879e-05,
+      "loss": 0.5167177319526672,
+      "num_input_tokens_seen": 227500,
+      "step": 706,
+      "train_runtime": 1868.7934,
+      "train_tokens_per_second": 121.736
+    },
+    {
+      "epoch": 0.03250574712643678,
+      "grad_norm": 7.478757381439209,
+      "learning_rate": 1.2972431077694237e-05,
+      "loss": 0.6133139133453369,
+      "num_input_tokens_seen": 227880,
+      "step": 707,
+      "train_runtime": 1870.8257,
+      "train_tokens_per_second": 121.807
+    },
+    {
+      "epoch": 0.03255172413793103,
+      "grad_norm": 7.987599849700928,
+      "learning_rate": 1.2962406015037594e-05,
+      "loss": 0.5924161076545715,
+      "num_input_tokens_seen": 228290,
+      "step": 708,
+      "train_runtime": 1872.8476,
+      "train_tokens_per_second": 121.895
+    },
+    {
+      "epoch": 0.032597701149425284,
+      "grad_norm": 7.35952615737915,
+      "learning_rate": 1.2952380952380954e-05,
+      "loss": 0.37251582741737366,
+      "num_input_tokens_seen": 228566,
+      "step": 709,
+      "train_runtime": 1874.7947,
+      "train_tokens_per_second": 121.915
+    },
+    {
+      "epoch": 0.03264367816091954,
+      "grad_norm": 8.810981750488281,
+      "learning_rate": 1.2942355889724313e-05,
+      "loss": 0.47795727849006653,
+      "num_input_tokens_seen": 228802,
+      "step": 710,
+      "train_runtime": 1876.7411,
+      "train_tokens_per_second": 121.915
+    },
+    {
+      "epoch": 0.0326896551724138,
+      "grad_norm": 7.858730316162109,
+      "learning_rate": 1.293233082706767e-05,
+      "loss": 0.5269241333007812,
+      "num_input_tokens_seen": 229110,
+      "step": 711,
+      "train_runtime": 1878.7187,
+      "train_tokens_per_second": 121.95
+    },
+    {
+      "epoch": 0.03273563218390805,
+      "grad_norm": 8.294468879699707,
+      "learning_rate": 1.292230576441103e-05,
+      "loss": 0.45605868101119995,
+      "num_input_tokens_seen": 229390,
+      "step": 712,
+      "train_runtime": 1880.6647,
+      "train_tokens_per_second": 121.973
+    },
+    {
+      "epoch": 0.0327816091954023,
+      "grad_norm": 10.320055961608887,
+      "learning_rate": 1.2912280701754386e-05,
+      "loss": 0.5948427319526672,
+      "num_input_tokens_seen": 229800,
+      "step": 713,
+      "train_runtime": 1882.6606,
+      "train_tokens_per_second": 122.061
+    },
+    {
+      "epoch": 0.032827586206896554,
+      "grad_norm": 7.579660892486572,
+      "learning_rate": 1.2902255639097745e-05,
+      "loss": 0.550399899482727,
+      "num_input_tokens_seen": 230112,
+      "step": 714,
+      "train_runtime": 1884.6154,
+      "train_tokens_per_second": 122.1
+    },
+    {
+      "epoch": 0.03287356321839081,
+      "grad_norm": 6.343344211578369,
+      "learning_rate": 1.2892230576441105e-05,
+      "loss": 0.4625878930091858,
+      "num_input_tokens_seen": 230484,
+      "step": 715,
+      "train_runtime": 1886.5964,
+      "train_tokens_per_second": 122.169
+    },
+    {
+      "epoch": 0.03291954022988506,
+      "grad_norm": 6.889019966125488,
+      "learning_rate": 1.2882205513784462e-05,
+      "loss": 0.4150872230529785,
+      "num_input_tokens_seen": 230778,
+      "step": 716,
+      "train_runtime": 1888.5437,
+      "train_tokens_per_second": 122.199
+    },
+    {
+      "epoch": 0.03296551724137931,
+      "grad_norm": 7.413138389587402,
+      "learning_rate": 1.287218045112782e-05,
+      "loss": 0.4648476243019104,
+      "num_input_tokens_seen": 231080,
+      "step": 717,
+      "train_runtime": 1890.5258,
+      "train_tokens_per_second": 122.231
+    },
+    {
+      "epoch": 0.033011494252873565,
+      "grad_norm": 5.9188995361328125,
+      "learning_rate": 1.2862155388471179e-05,
+      "loss": 0.27306920289993286,
+      "num_input_tokens_seen": 231318,
+      "step": 718,
+      "train_runtime": 1892.4846,
+      "train_tokens_per_second": 122.23
+    },
+    {
+      "epoch": 0.03305747126436782,
+      "grad_norm": 7.059986114501953,
+      "learning_rate": 1.2852130325814537e-05,
+      "loss": 0.4230448305606842,
+      "num_input_tokens_seen": 231628,
+      "step": 719,
+      "train_runtime": 1894.4493,
+      "train_tokens_per_second": 122.267
+    },
+    {
+      "epoch": 0.03310344827586207,
+      "grad_norm": 8.006654739379883,
+      "learning_rate": 1.2842105263157896e-05,
+      "loss": 0.5292198061943054,
+      "num_input_tokens_seen": 231958,
+      "step": 720,
+      "train_runtime": 1896.4044,
+      "train_tokens_per_second": 122.315
+    },
+    {
+      "epoch": 0.03314942528735632,
+      "grad_norm": 8.713791847229004,
+      "learning_rate": 1.2832080200501254e-05,
+      "loss": 0.4419120252132416,
+      "num_input_tokens_seen": 232242,
+      "step": 721,
+      "train_runtime": 1911.0895,
+      "train_tokens_per_second": 121.523
+    },
+    {
+      "epoch": 0.033195402298850575,
+      "grad_norm": 7.501795768737793,
+      "learning_rate": 1.2822055137844613e-05,
+      "loss": 0.47290876507759094,
+      "num_input_tokens_seen": 232524,
+      "step": 722,
+      "train_runtime": 1913.0366,
+      "train_tokens_per_second": 121.547
+    },
+    {
+      "epoch": 0.03324137931034483,
+      "grad_norm": 6.713019847869873,
+      "learning_rate": 1.2812030075187971e-05,
+      "loss": 0.5273690223693848,
+      "num_input_tokens_seen": 233128,
+      "step": 723,
+      "train_runtime": 1915.0886,
+      "train_tokens_per_second": 121.732
+    },
+    {
+      "epoch": 0.03328735632183908,
+      "grad_norm": 6.822129249572754,
+      "learning_rate": 1.2802005012531328e-05,
+      "loss": 0.4337116479873657,
+      "num_input_tokens_seen": 233550,
+      "step": 724,
+      "train_runtime": 1917.1347,
+      "train_tokens_per_second": 121.822
+    },
+    {
+      "epoch": 0.03333333333333333,
+      "grad_norm": 6.972231388092041,
+      "learning_rate": 1.2791979949874688e-05,
+      "loss": 0.4706537127494812,
+      "num_input_tokens_seen": 233794,
+      "step": 725,
+      "train_runtime": 1919.1695,
+      "train_tokens_per_second": 121.82
+    },
+    {
+      "epoch": 0.033379310344827585,
+      "grad_norm": 6.156042575836182,
+      "learning_rate": 1.2781954887218047e-05,
+      "loss": 0.3718259334564209,
+      "num_input_tokens_seen": 234110,
+      "step": 726,
+      "train_runtime": 1921.183,
+      "train_tokens_per_second": 121.857
+    },
+    {
+      "epoch": 0.03342528735632184,
+      "grad_norm": 7.888618469238281,
+      "learning_rate": 1.2771929824561404e-05,
+      "loss": 0.36061882972717285,
+      "num_input_tokens_seen": 234418,
+      "step": 727,
+      "train_runtime": 1923.2046,
+      "train_tokens_per_second": 121.889
+    },
+    {
+      "epoch": 0.03347126436781609,
+      "grad_norm": 7.973721981048584,
+      "learning_rate": 1.2761904761904762e-05,
+      "loss": 0.3296748399734497,
+      "num_input_tokens_seen": 234710,
+      "step": 728,
+      "train_runtime": 1925.1532,
+      "train_tokens_per_second": 121.918
+    },
+    {
+      "epoch": 0.03351724137931034,
+      "grad_norm": 6.52110481262207,
+      "learning_rate": 1.2751879699248122e-05,
+      "loss": 0.454277902841568,
+      "num_input_tokens_seen": 234990,
+      "step": 729,
+      "train_runtime": 1927.1264,
+      "train_tokens_per_second": 121.938
+    },
+    {
+      "epoch": 0.033563218390804596,
+      "grad_norm": 7.006773948669434,
+      "learning_rate": 1.2741854636591479e-05,
+      "loss": 0.4588077664375305,
+      "num_input_tokens_seen": 235280,
+      "step": 730,
+      "train_runtime": 1929.1047,
+      "train_tokens_per_second": 121.963
+    },
+    {
+      "epoch": 0.03360919540229885,
+      "grad_norm": 7.811942100524902,
+      "learning_rate": 1.2731829573934837e-05,
+      "loss": 0.5093876123428345,
+      "num_input_tokens_seen": 235590,
+      "step": 731,
+      "train_runtime": 1931.0677,
+      "train_tokens_per_second": 122.0
+    },
+    {
+      "epoch": 0.0336551724137931,
+      "grad_norm": 9.06869125366211,
+      "learning_rate": 1.2721804511278198e-05,
+      "loss": 0.501866340637207,
+      "num_input_tokens_seen": 235878,
+      "step": 732,
+      "train_runtime": 1933.0266,
+      "train_tokens_per_second": 122.025
+    },
+    {
+      "epoch": 0.033701149425287354,
+      "grad_norm": 7.744815349578857,
+      "learning_rate": 1.2711779448621554e-05,
+      "loss": 0.5169853568077087,
+      "num_input_tokens_seen": 236118,
+      "step": 733,
+      "train_runtime": 1934.9909,
+      "train_tokens_per_second": 122.025
+    },
+    {
+      "epoch": 0.033747126436781606,
+      "grad_norm": 8.02056884765625,
+      "learning_rate": 1.2701754385964913e-05,
+      "loss": 0.4525764286518097,
+      "num_input_tokens_seen": 236362,
+      "step": 734,
+      "train_runtime": 1936.9662,
+      "train_tokens_per_second": 122.027
+    },
+    {
+      "epoch": 0.03379310344827586,
+      "grad_norm": 7.841916561126709,
+      "learning_rate": 1.2691729323308273e-05,
+      "loss": 0.44451504945755005,
+      "num_input_tokens_seen": 236674,
+      "step": 735,
+      "train_runtime": 1938.9437,
+      "train_tokens_per_second": 122.063
+    },
+    {
+      "epoch": 0.03383908045977012,
+      "grad_norm": 8.466221809387207,
+      "learning_rate": 1.268170426065163e-05,
+      "loss": 0.5051043033599854,
+      "num_input_tokens_seen": 236960,
+      "step": 736,
+      "train_runtime": 1940.9267,
+      "train_tokens_per_second": 122.086
+    },
+    {
+      "epoch": 0.03388505747126437,
+      "grad_norm": 8.845619201660156,
+      "learning_rate": 1.2671679197994988e-05,
+      "loss": 0.5814340710639954,
+      "num_input_tokens_seen": 237332,
+      "step": 737,
+      "train_runtime": 1942.9296,
+      "train_tokens_per_second": 122.152
+    },
+    {
+      "epoch": 0.033931034482758624,
+      "grad_norm": 8.570676803588867,
+      "learning_rate": 1.2661654135338345e-05,
+      "loss": 0.42572882771492004,
+      "num_input_tokens_seen": 237562,
+      "step": 738,
+      "train_runtime": 1944.9096,
+      "train_tokens_per_second": 122.146
+    },
+    {
+      "epoch": 0.033977011494252876,
+      "grad_norm": 8.907341957092285,
+      "learning_rate": 1.2651629072681705e-05,
+      "loss": 0.32531145215034485,
+      "num_input_tokens_seen": 237810,
+      "step": 739,
+      "train_runtime": 1946.8737,
+      "train_tokens_per_second": 122.15
+    },
+    {
+      "epoch": 0.03402298850574713,
+      "grad_norm": 7.892190933227539,
+      "learning_rate": 1.2641604010025064e-05,
+      "loss": 0.39891165494918823,
+      "num_input_tokens_seen": 238072,
+      "step": 740,
+      "train_runtime": 1948.8155,
+      "train_tokens_per_second": 122.162
+    },
+    {
+      "epoch": 0.03406896551724138,
+      "grad_norm": 8.642949104309082,
+      "learning_rate": 1.263157894736842e-05,
+      "loss": 0.4310913383960724,
+      "num_input_tokens_seen": 238384,
+      "step": 741,
+      "train_runtime": 1950.807,
+      "train_tokens_per_second": 122.198
+    },
+    {
+      "epoch": 0.034114942528735634,
+      "grad_norm": 7.021979331970215,
+      "learning_rate": 1.262155388471178e-05,
+      "loss": 0.30681487917900085,
+      "num_input_tokens_seen": 238654,
+      "step": 742,
+      "train_runtime": 1952.7593,
+      "train_tokens_per_second": 122.214
+    },
+    {
+      "epoch": 0.034160919540229887,
+      "grad_norm": 8.602571487426758,
+      "learning_rate": 1.261152882205514e-05,
+      "loss": 0.56992107629776,
+      "num_input_tokens_seen": 239044,
+      "step": 743,
+      "train_runtime": 1954.7455,
+      "train_tokens_per_second": 122.289
+    },
+    {
+      "epoch": 0.03420689655172414,
+      "grad_norm": 8.6575927734375,
+      "learning_rate": 1.2601503759398496e-05,
+      "loss": 0.5144814252853394,
+      "num_input_tokens_seen": 239334,
+      "step": 744,
+      "train_runtime": 1956.7685,
+      "train_tokens_per_second": 122.311
+    },
+    {
+      "epoch": 0.03425287356321839,
+      "grad_norm": 7.739847660064697,
+      "learning_rate": 1.2591478696741856e-05,
+      "loss": 0.43455153703689575,
+      "num_input_tokens_seen": 239608,
+      "step": 745,
+      "train_runtime": 1958.8685,
+      "train_tokens_per_second": 122.32
+    },
+    {
+      "epoch": 0.034298850574712644,
+      "grad_norm": 8.102360725402832,
+      "learning_rate": 1.2581453634085215e-05,
+      "loss": 0.5057243704795837,
+      "num_input_tokens_seen": 239894,
+      "step": 746,
+      "train_runtime": 1960.9181,
+      "train_tokens_per_second": 122.338
+    },
+    {
+      "epoch": 0.0343448275862069,
+      "grad_norm": 7.547651290893555,
+      "learning_rate": 1.2571428571428572e-05,
+      "loss": 0.5827213525772095,
+      "num_input_tokens_seen": 240464,
+      "step": 747,
+      "train_runtime": 1962.9819,
+      "train_tokens_per_second": 122.499
+    },
+    {
+      "epoch": 0.03439080459770115,
+      "grad_norm": 8.792771339416504,
+      "learning_rate": 1.256140350877193e-05,
+      "loss": 0.5615402460098267,
+      "num_input_tokens_seen": 240774,
+      "step": 748,
+      "train_runtime": 1964.9332,
+      "train_tokens_per_second": 122.535
+    },
+    {
+      "epoch": 0.0344367816091954,
+      "grad_norm": 9.596569061279297,
+      "learning_rate": 1.255137844611529e-05,
+      "loss": 0.3781774640083313,
+      "num_input_tokens_seen": 241046,
+      "step": 749,
+      "train_runtime": 1966.9508,
+      "train_tokens_per_second": 122.548
+    },
+    {
+      "epoch": 0.034482758620689655,
+      "grad_norm": 8.446516036987305,
+      "learning_rate": 1.2541353383458647e-05,
+      "loss": 0.5088064074516296,
+      "num_input_tokens_seen": 241522,
+      "step": 750,
+      "train_runtime": 1968.9682,
+      "train_tokens_per_second": 122.664
+    },
+    {
+      "epoch": 0.03452873563218391,
+      "grad_norm": 7.82168436050415,
+      "learning_rate": 1.2531328320802006e-05,
+      "loss": 0.4287633001804352,
+      "num_input_tokens_seen": 241790,
+      "step": 751,
+      "train_runtime": 1983.9727,
+      "train_tokens_per_second": 121.872
+    },
+    {
+      "epoch": 0.03457471264367816,
+      "grad_norm": 7.16981840133667,
+      "learning_rate": 1.2521303258145366e-05,
+      "loss": 0.486166775226593,
+      "num_input_tokens_seen": 242142,
+      "step": 752,
+      "train_runtime": 1985.9572,
+      "train_tokens_per_second": 121.927
+    },
+    {
+      "epoch": 0.03462068965517241,
+      "grad_norm": 8.633017539978027,
+      "learning_rate": 1.2511278195488723e-05,
+      "loss": 0.39653506875038147,
+      "num_input_tokens_seen": 242408,
+      "step": 753,
+      "train_runtime": 1987.9791,
+      "train_tokens_per_second": 121.937
+    },
+    {
+      "epoch": 0.034666666666666665,
+      "grad_norm": 8.903992652893066,
+      "learning_rate": 1.2501253132832081e-05,
+      "loss": 0.4707050323486328,
+      "num_input_tokens_seen": 242926,
+      "step": 754,
+      "train_runtime": 1990.017,
+      "train_tokens_per_second": 122.072
+    },
+    {
+      "epoch": 0.03471264367816092,
+      "grad_norm": 10.9977388381958,
+      "learning_rate": 1.2491228070175441e-05,
+      "loss": 0.3755106031894684,
+      "num_input_tokens_seen": 243162,
+      "step": 755,
+      "train_runtime": 1992.034,
+      "train_tokens_per_second": 122.067
+    },
+    {
+      "epoch": 0.03475862068965517,
+      "grad_norm": 8.29912281036377,
+      "learning_rate": 1.2481203007518798e-05,
+      "loss": 0.5001455545425415,
+      "num_input_tokens_seen": 243460,
+      "step": 756,
+      "train_runtime": 1994.0761,
+      "train_tokens_per_second": 122.092
+    },
+    {
+      "epoch": 0.03480459770114942,
+      "grad_norm": 7.330634117126465,
+      "learning_rate": 1.2471177944862156e-05,
+      "loss": 0.529636561870575,
+      "num_input_tokens_seen": 243786,
+      "step": 757,
+      "train_runtime": 1996.0372,
+      "train_tokens_per_second": 122.135
+    },
+    {
+      "epoch": 0.034850574712643675,
+      "grad_norm": 6.908694267272949,
+      "learning_rate": 1.2461152882205513e-05,
+      "loss": 0.287724107503891,
+      "num_input_tokens_seen": 244026,
+      "step": 758,
+      "train_runtime": 1998.0191,
+      "train_tokens_per_second": 122.134
+    },
+    {
+      "epoch": 0.03489655172413793,
+      "grad_norm": 7.529731750488281,
+      "learning_rate": 1.2451127819548873e-05,
+      "loss": 0.5130578875541687,
+      "num_input_tokens_seen": 244366,
+      "step": 759,
+      "train_runtime": 2000.0437,
+      "train_tokens_per_second": 122.18
+    },
+    {
+      "epoch": 0.03494252873563218,
+      "grad_norm": 8.070693016052246,
+      "learning_rate": 1.2441102756892232e-05,
+      "loss": 0.39174437522888184,
+      "num_input_tokens_seen": 244746,
+      "step": 760,
+      "train_runtime": 2002.1149,
+      "train_tokens_per_second": 122.244
+    },
+    {
+      "epoch": 0.03498850574712644,
+      "grad_norm": 8.24476432800293,
+      "learning_rate": 1.2431077694235589e-05,
+      "loss": 0.5158138275146484,
+      "num_input_tokens_seen": 245056,
+      "step": 761,
+      "train_runtime": 2004.1891,
+      "train_tokens_per_second": 122.272
+    },
+    {
+      "epoch": 0.03503448275862069,
+      "grad_norm": 6.291467189788818,
+      "learning_rate": 1.2421052631578949e-05,
+      "loss": 0.3103332817554474,
+      "num_input_tokens_seen": 245382,
+      "step": 762,
+      "train_runtime": 2006.1648,
+      "train_tokens_per_second": 122.314
+    },
+    {
+      "epoch": 0.035080459770114945,
+      "grad_norm": 7.742890357971191,
+      "learning_rate": 1.2411027568922307e-05,
+      "loss": 0.5137875080108643,
+      "num_input_tokens_seen": 245714,
+      "step": 763,
+      "train_runtime": 2008.1302,
+      "train_tokens_per_second": 122.36
+    },
+    {
+      "epoch": 0.0351264367816092,
+      "grad_norm": 6.73689079284668,
+      "learning_rate": 1.2401002506265664e-05,
+      "loss": 0.3042636811733246,
+      "num_input_tokens_seen": 245956,
+      "step": 764,
+      "train_runtime": 2010.2565,
+      "train_tokens_per_second": 122.351
+    },
+    {
+      "epoch": 0.03517241379310345,
+      "grad_norm": 8.467912673950195,
+      "learning_rate": 1.2390977443609024e-05,
+      "loss": 0.5755658745765686,
+      "num_input_tokens_seen": 246240,
+      "step": 765,
+      "train_runtime": 2012.2645,
+      "train_tokens_per_second": 122.37
+    },
+    {
+      "epoch": 0.0352183908045977,
+      "grad_norm": 7.699549674987793,
+      "learning_rate": 1.2380952380952383e-05,
+      "loss": 0.5804579257965088,
+      "num_input_tokens_seen": 246638,
+      "step": 766,
+      "train_runtime": 2014.2625,
+      "train_tokens_per_second": 122.446
+    },
+    {
+      "epoch": 0.035264367816091956,
+      "grad_norm": 7.961452484130859,
+      "learning_rate": 1.237092731829574e-05,
+      "loss": 0.4314732253551483,
+      "num_input_tokens_seen": 246968,
+      "step": 767,
+      "train_runtime": 2016.2306,
+      "train_tokens_per_second": 122.49
+    },
+    {
+      "epoch": 0.03531034482758621,
+      "grad_norm": 7.277204990386963,
+      "learning_rate": 1.2360902255639098e-05,
+      "loss": 0.5331660509109497,
+      "num_input_tokens_seen": 247258,
+      "step": 768,
+      "train_runtime": 2018.2055,
+      "train_tokens_per_second": 122.514
+    },
+    {
+      "epoch": 0.03535632183908046,
+      "grad_norm": 5.924410820007324,
+      "learning_rate": 1.2350877192982458e-05,
+      "loss": 0.37716495990753174,
+      "num_input_tokens_seen": 247624,
+      "step": 769,
+      "train_runtime": 2020.1989,
+      "train_tokens_per_second": 122.574
+    },
+    {
+      "epoch": 0.035402298850574714,
+      "grad_norm": 8.82693862915039,
+      "learning_rate": 1.2340852130325815e-05,
+      "loss": 0.4919133186340332,
+      "num_input_tokens_seen": 247862,
+      "step": 770,
+      "train_runtime": 2022.162,
+      "train_tokens_per_second": 122.573
+    },
+    {
+      "epoch": 0.035448275862068966,
+      "grad_norm": 10.985586166381836,
+      "learning_rate": 1.2330827067669174e-05,
+      "loss": 0.4656320810317993,
+      "num_input_tokens_seen": 248190,
+      "step": 771,
+      "train_runtime": 2024.1127,
+      "train_tokens_per_second": 122.617
+    },
+    {
+      "epoch": 0.03549425287356322,
+      "grad_norm": 9.225574493408203,
+      "learning_rate": 1.2320802005012534e-05,
+      "loss": 0.6084414124488831,
+      "num_input_tokens_seen": 248462,
+      "step": 772,
+      "train_runtime": 2026.0559,
+      "train_tokens_per_second": 122.633
+    },
+    {
+      "epoch": 0.03554022988505747,
+      "grad_norm": 7.783392429351807,
+      "learning_rate": 1.231077694235589e-05,
+      "loss": 0.4332180917263031,
+      "num_input_tokens_seen": 248776,
+      "step": 773,
+      "train_runtime": 2028.0009,
+      "train_tokens_per_second": 122.671
+    },
+    {
+      "epoch": 0.035586206896551724,
+      "grad_norm": 7.647105693817139,
+      "learning_rate": 1.2300751879699249e-05,
+      "loss": 0.4509713649749756,
+      "num_input_tokens_seen": 249078,
+      "step": 774,
+      "train_runtime": 2029.9753,
+      "train_tokens_per_second": 122.7
+    },
+    {
+      "epoch": 0.035632183908045977,
+      "grad_norm": 6.6883625984191895,
+      "learning_rate": 1.229072681704261e-05,
+      "loss": 0.43819460272789,
+      "num_input_tokens_seen": 249340,
+      "step": 775,
+      "train_runtime": 2031.9451,
+      "train_tokens_per_second": 122.71
+    },
+    {
+      "epoch": 0.03567816091954023,
+      "grad_norm": 8.419464111328125,
+      "learning_rate": 1.2280701754385966e-05,
+      "loss": 0.6304515600204468,
+      "num_input_tokens_seen": 249640,
+      "step": 776,
+      "train_runtime": 2033.9021,
+      "train_tokens_per_second": 122.739
+    },
+    {
+      "epoch": 0.03572413793103448,
+      "grad_norm": 6.70818567276001,
+      "learning_rate": 1.2270676691729325e-05,
+      "loss": 0.36638951301574707,
+      "num_input_tokens_seen": 249950,
+      "step": 777,
+      "train_runtime": 2035.8967,
+      "train_tokens_per_second": 122.771
+    },
+    {
+      "epoch": 0.035770114942528734,
+      "grad_norm": 8.973495483398438,
+      "learning_rate": 1.2260651629072681e-05,
+      "loss": 0.4474521279335022,
+      "num_input_tokens_seen": 250274,
+      "step": 778,
+      "train_runtime": 2037.9072,
+      "train_tokens_per_second": 122.809
+    },
+    {
+      "epoch": 0.03581609195402299,
+      "grad_norm": 7.661665916442871,
+      "learning_rate": 1.2250626566416041e-05,
+      "loss": 0.424831360578537,
+      "num_input_tokens_seen": 250566,
+      "step": 779,
+      "train_runtime": 2039.9066,
+      "train_tokens_per_second": 122.832
+    },
+    {
+      "epoch": 0.03586206896551724,
+      "grad_norm": 7.2076897621154785,
+      "learning_rate": 1.22406015037594e-05,
+      "loss": 0.46515989303588867,
+      "num_input_tokens_seen": 250874,
+      "step": 780,
+      "train_runtime": 2041.9398,
+      "train_tokens_per_second": 122.861
+    },
+    {
+      "epoch": 0.03590804597701149,
+      "grad_norm": 8.688509941101074,
+      "learning_rate": 1.2230576441102757e-05,
+      "loss": 0.5213080048561096,
+      "num_input_tokens_seen": 251220,
+      "step": 781,
+      "train_runtime": 2056.584,
+      "train_tokens_per_second": 122.154
+    },
+    {
+      "epoch": 0.035954022988505745,
+      "grad_norm": 7.395033836364746,
+      "learning_rate": 1.2220551378446117e-05,
+      "loss": 0.5273518562316895,
+      "num_input_tokens_seen": 251498,
+      "step": 782,
+      "train_runtime": 2058.5801,
+      "train_tokens_per_second": 122.171
+    },
+    {
+      "epoch": 0.036,
+      "grad_norm": 7.278432846069336,
+      "learning_rate": 1.2210526315789475e-05,
+      "loss": 0.4046718180179596,
+      "num_input_tokens_seen": 251790,
+      "step": 783,
+      "train_runtime": 2060.5337,
+      "train_tokens_per_second": 122.196
+    },
+    {
+      "epoch": 0.03604597701149425,
+      "grad_norm": 7.578866481781006,
+      "learning_rate": 1.2200501253132832e-05,
+      "loss": 0.407046914100647,
+      "num_input_tokens_seen": 252110,
+      "step": 784,
+      "train_runtime": 2062.4952,
+      "train_tokens_per_second": 122.235
+    },
+    {
+      "epoch": 0.0360919540229885,
+      "grad_norm": 7.994529724121094,
+      "learning_rate": 1.2190476190476192e-05,
+      "loss": 0.4625910818576813,
+      "num_input_tokens_seen": 252400,
+      "step": 785,
+      "train_runtime": 2064.5309,
+      "train_tokens_per_second": 122.255
+    },
+    {
+      "epoch": 0.03613793103448276,
+      "grad_norm": 7.406494140625,
+      "learning_rate": 1.2180451127819551e-05,
+      "loss": 0.40049877762794495,
+      "num_input_tokens_seen": 252644,
+      "step": 786,
+      "train_runtime": 2066.563,
+      "train_tokens_per_second": 122.253
+    },
+    {
+      "epoch": 0.036183908045977015,
+      "grad_norm": 7.816009998321533,
+      "learning_rate": 1.2170426065162908e-05,
+      "loss": 0.43142542243003845,
+      "num_input_tokens_seen": 252940,
+      "step": 787,
+      "train_runtime": 2068.5445,
+      "train_tokens_per_second": 122.279
+    },
+    {
+      "epoch": 0.03622988505747127,
+      "grad_norm": 7.505245685577393,
+      "learning_rate": 1.2160401002506266e-05,
+      "loss": 0.5745660066604614,
+      "num_input_tokens_seen": 253202,
+      "step": 788,
+      "train_runtime": 2070.5511,
+      "train_tokens_per_second": 122.287
+    },
+    {
+      "epoch": 0.03627586206896552,
+      "grad_norm": 8.933348655700684,
+      "learning_rate": 1.2150375939849626e-05,
+      "loss": 0.4125485420227051,
+      "num_input_tokens_seen": 253450,
+      "step": 789,
+      "train_runtime": 2072.5072,
+      "train_tokens_per_second": 122.291
+    },
+    {
+      "epoch": 0.03632183908045977,
+      "grad_norm": 7.910968780517578,
+      "learning_rate": 1.2140350877192983e-05,
+      "loss": 0.42910078167915344,
+      "num_input_tokens_seen": 253722,
+      "step": 790,
+      "train_runtime": 2074.4758,
+      "train_tokens_per_second": 122.307
+    },
+    {
+      "epoch": 0.036367816091954025,
+      "grad_norm": 6.820565700531006,
+      "learning_rate": 1.2130325814536342e-05,
+      "loss": 0.43786799907684326,
+      "num_input_tokens_seen": 254020,
+      "step": 791,
+      "train_runtime": 2076.4855,
+      "train_tokens_per_second": 122.332
+    },
+    {
+      "epoch": 0.03641379310344828,
+      "grad_norm": 8.971982955932617,
+      "learning_rate": 1.2120300751879702e-05,
+      "loss": 0.46482279896736145,
+      "num_input_tokens_seen": 254282,
+      "step": 792,
+      "train_runtime": 2078.4249,
+      "train_tokens_per_second": 122.344
+    },
+    {
+      "epoch": 0.03645977011494253,
+      "grad_norm": 7.097390174865723,
+      "learning_rate": 1.2110275689223059e-05,
+      "loss": 0.5560008883476257,
+      "num_input_tokens_seen": 254906,
+      "step": 793,
+      "train_runtime": 2080.4848,
+      "train_tokens_per_second": 122.522
+    },
+    {
+      "epoch": 0.03650574712643678,
+      "grad_norm": 7.8160881996154785,
+      "learning_rate": 1.2100250626566417e-05,
+      "loss": 0.5623802542686462,
+      "num_input_tokens_seen": 255222,
+      "step": 794,
+      "train_runtime": 2082.4285,
+      "train_tokens_per_second": 122.56
+    },
+    {
+      "epoch": 0.036551724137931035,
+      "grad_norm": 8.788517951965332,
+      "learning_rate": 1.2090225563909777e-05,
+      "loss": 0.4851282835006714,
+      "num_input_tokens_seen": 255614,
+      "step": 795,
+      "train_runtime": 2084.4044,
+      "train_tokens_per_second": 122.632
+    },
+    {
+      "epoch": 0.03659770114942529,
+      "grad_norm": 7.661938190460205,
+      "learning_rate": 1.2080200501253134e-05,
+      "loss": 0.4763977527618408,
+      "num_input_tokens_seen": 256134,
+      "step": 796,
+      "train_runtime": 2086.4251,
+      "train_tokens_per_second": 122.762
+    },
+    {
+      "epoch": 0.03664367816091954,
+      "grad_norm": 7.841512680053711,
+      "learning_rate": 1.2070175438596493e-05,
+      "loss": 0.4477618932723999,
+      "num_input_tokens_seen": 256392,
+      "step": 797,
+      "train_runtime": 2088.3741,
+      "train_tokens_per_second": 122.771
+    },
+    {
+      "epoch": 0.03668965517241379,
+      "grad_norm": 8.038578987121582,
+      "learning_rate": 1.206015037593985e-05,
+      "loss": 0.4073207974433899,
+      "num_input_tokens_seen": 256756,
+      "step": 798,
+      "train_runtime": 2090.3709,
+      "train_tokens_per_second": 122.828
+    },
+    {
+      "epoch": 0.036735632183908046,
+      "grad_norm": 7.738718032836914,
+      "learning_rate": 1.205012531328321e-05,
+      "loss": 0.5579912662506104,
+      "num_input_tokens_seen": 257056,
+      "step": 799,
+      "train_runtime": 2092.3563,
+      "train_tokens_per_second": 122.855
+    },
+    {
+      "epoch": 0.0367816091954023,
+      "grad_norm": 12.660835266113281,
+      "learning_rate": 1.2040100250626568e-05,
+      "loss": 0.6363130211830139,
+      "num_input_tokens_seen": 257392,
+      "step": 800,
+      "train_runtime": 2094.3382,
+      "train_tokens_per_second": 122.899
+    },
+    {
+      "epoch": 0.0367816091954023,
+      "eval_loss": 1.7513116598129272,
+      "eval_runtime": 54.1869,
+      "eval_samples_per_second": 18.455,
+      "eval_steps_per_second": 9.227,
+      "num_input_tokens_seen": 257392,
+      "step": 800
+    },
+    {
+      "epoch": 0.03682758620689655,
+      "grad_norm": 8.01834487915039,
+      "learning_rate": 1.2030075187969925e-05,
+      "loss": 0.40306514501571655,
+      "num_input_tokens_seen": 257638,
+      "step": 801,
+      "train_runtime": 2150.5794,
+      "train_tokens_per_second": 119.799
+    },
+    {
+      "epoch": 0.036873563218390804,
+      "grad_norm": 8.324341773986816,
+      "learning_rate": 1.2020050125313285e-05,
+      "loss": 0.4507819712162018,
+      "num_input_tokens_seen": 258164,
+      "step": 802,
+      "train_runtime": 2152.578,
+      "train_tokens_per_second": 119.932
+    },
+    {
+      "epoch": 0.036919540229885056,
+      "grad_norm": 7.941473484039307,
+      "learning_rate": 1.2010025062656644e-05,
+      "loss": 0.4734458029270172,
+      "num_input_tokens_seen": 258454,
+      "step": 803,
+      "train_runtime": 2154.57,
+      "train_tokens_per_second": 119.956
+    },
+    {
+      "epoch": 0.03696551724137931,
+      "grad_norm": 7.906909465789795,
+      "learning_rate": 1.2e-05,
+      "loss": 0.3462047576904297,
+      "num_input_tokens_seen": 258712,
+      "step": 804,
+      "train_runtime": 2156.5256,
+      "train_tokens_per_second": 119.967
+    },
+    {
+      "epoch": 0.03701149425287356,
+      "grad_norm": 7.4157843589782715,
+      "learning_rate": 1.1989974937343359e-05,
+      "loss": 0.45490097999572754,
+      "num_input_tokens_seen": 258970,
+      "step": 805,
+      "train_runtime": 2158.4846,
+      "train_tokens_per_second": 119.978
+    },
+    {
+      "epoch": 0.037057471264367814,
+      "grad_norm": 6.880097389221191,
+      "learning_rate": 1.1979949874686719e-05,
+      "loss": 0.5047414302825928,
+      "num_input_tokens_seen": 259248,
+      "step": 806,
+      "train_runtime": 2160.4505,
+      "train_tokens_per_second": 119.997
+    },
+    {
+      "epoch": 0.037103448275862067,
+      "grad_norm": 8.169018745422363,
+      "learning_rate": 1.1969924812030076e-05,
+      "loss": 0.4830625057220459,
+      "num_input_tokens_seen": 259530,
+      "step": 807,
+      "train_runtime": 2162.4594,
+      "train_tokens_per_second": 120.016
+    },
+    {
+      "epoch": 0.03714942528735632,
+      "grad_norm": 7.684258937835693,
+      "learning_rate": 1.1959899749373434e-05,
+      "loss": 0.5238369107246399,
+      "num_input_tokens_seen": 259846,
+      "step": 808,
+      "train_runtime": 2164.4497,
+      "train_tokens_per_second": 120.052
+    },
+    {
+      "epoch": 0.03719540229885057,
+      "grad_norm": 6.430134296417236,
+      "learning_rate": 1.1949874686716794e-05,
+      "loss": 0.3272133767604828,
+      "num_input_tokens_seen": 260084,
+      "step": 809,
+      "train_runtime": 2166.3968,
+      "train_tokens_per_second": 120.054
+    },
+    {
+      "epoch": 0.037241379310344824,
+      "grad_norm": 7.486498832702637,
+      "learning_rate": 1.1939849624060151e-05,
+      "loss": 0.3699163496494293,
+      "num_input_tokens_seen": 260340,
+      "step": 810,
+      "train_runtime": 2168.3539,
+      "train_tokens_per_second": 120.063
+    },
+    {
+      "epoch": 0.037287356321839084,
+      "grad_norm": 7.046302795410156,
+      "learning_rate": 1.192982456140351e-05,
+      "loss": 0.4449416995048523,
+      "num_input_tokens_seen": 260596,
+      "step": 811,
+      "train_runtime": 2183.5737,
+      "train_tokens_per_second": 119.344
+    },
+    {
+      "epoch": 0.037333333333333336,
+      "grad_norm": 7.615978240966797,
+      "learning_rate": 1.1919799498746868e-05,
+      "loss": 0.39871329069137573,
+      "num_input_tokens_seen": 260892,
+      "step": 812,
+      "train_runtime": 2185.5393,
+      "train_tokens_per_second": 119.372
+    },
+    {
+      "epoch": 0.03737931034482759,
+      "grad_norm": 8.0755615234375,
+      "learning_rate": 1.1909774436090227e-05,
+      "loss": 0.5120543241500854,
+      "num_input_tokens_seen": 261184,
+      "step": 813,
+      "train_runtime": 2187.5246,
+      "train_tokens_per_second": 119.397
+    },
+    {
+      "epoch": 0.03742528735632184,
+      "grad_norm": 7.727443695068359,
+      "learning_rate": 1.1899749373433585e-05,
+      "loss": 0.4481019079685211,
+      "num_input_tokens_seen": 261510,
+      "step": 814,
+      "train_runtime": 2189.4837,
+      "train_tokens_per_second": 119.439
+    },
+    {
+      "epoch": 0.037471264367816094,
+      "grad_norm": 7.489780902862549,
+      "learning_rate": 1.1889724310776942e-05,
+      "loss": 0.5464756488800049,
+      "num_input_tokens_seen": 261832,
+      "step": 815,
+      "train_runtime": 2191.4881,
+      "train_tokens_per_second": 119.477
+    },
+    {
+      "epoch": 0.03751724137931035,
+      "grad_norm": 7.514716625213623,
+      "learning_rate": 1.1879699248120302e-05,
+      "loss": 0.40943312644958496,
+      "num_input_tokens_seen": 262070,
+      "step": 816,
+      "train_runtime": 2193.4286,
+      "train_tokens_per_second": 119.48
+    },
+    {
+      "epoch": 0.0375632183908046,
+      "grad_norm": 7.514858245849609,
+      "learning_rate": 1.186967418546366e-05,
+      "loss": 0.4800604581832886,
+      "num_input_tokens_seen": 262358,
+      "step": 817,
+      "train_runtime": 2195.3865,
+      "train_tokens_per_second": 119.504
+    },
+    {
+      "epoch": 0.03760919540229885,
+      "grad_norm": 7.993260860443115,
+      "learning_rate": 1.1859649122807017e-05,
+      "loss": 0.4908154010772705,
+      "num_input_tokens_seen": 262650,
+      "step": 818,
+      "train_runtime": 2197.3534,
+      "train_tokens_per_second": 119.53
+    },
+    {
+      "epoch": 0.037655172413793105,
+      "grad_norm": 6.254027366638184,
+      "learning_rate": 1.1849624060150378e-05,
+      "loss": 0.378743976354599,
+      "num_input_tokens_seen": 262892,
+      "step": 819,
+      "train_runtime": 2199.2982,
+      "train_tokens_per_second": 119.534
+    },
+    {
+      "epoch": 0.03770114942528736,
+      "grad_norm": 7.698431491851807,
+      "learning_rate": 1.1839598997493736e-05,
+      "loss": 0.4918335974216461,
+      "num_input_tokens_seen": 263186,
+      "step": 820,
+      "train_runtime": 2201.2494,
+      "train_tokens_per_second": 119.562
+    },
+    {
+      "epoch": 0.03774712643678161,
+      "grad_norm": 7.610840320587158,
+      "learning_rate": 1.1829573934837093e-05,
+      "loss": 0.4648166596889496,
+      "num_input_tokens_seen": 263482,
+      "step": 821,
+      "train_runtime": 2203.1831,
+      "train_tokens_per_second": 119.592
+    },
+    {
+      "epoch": 0.03779310344827586,
+      "grad_norm": 7.461253643035889,
+      "learning_rate": 1.1819548872180453e-05,
+      "loss": 0.5091919898986816,
+      "num_input_tokens_seen": 263802,
+      "step": 822,
+      "train_runtime": 2205.1372,
+      "train_tokens_per_second": 119.631
+    },
+    {
+      "epoch": 0.037839080459770115,
+      "grad_norm": 7.983728408813477,
+      "learning_rate": 1.180952380952381e-05,
+      "loss": 0.40278542041778564,
+      "num_input_tokens_seen": 264092,
+      "step": 823,
+      "train_runtime": 2207.1388,
+      "train_tokens_per_second": 119.654
+    },
+    {
+      "epoch": 0.03788505747126437,
+      "grad_norm": 8.200235366821289,
+      "learning_rate": 1.1799498746867168e-05,
+      "loss": 0.5143745541572571,
+      "num_input_tokens_seen": 264348,
+      "step": 824,
+      "train_runtime": 2209.1144,
+      "train_tokens_per_second": 119.662
+    },
+    {
+      "epoch": 0.03793103448275862,
+      "grad_norm": 7.840239524841309,
+      "learning_rate": 1.1789473684210527e-05,
+      "loss": 0.4168047606945038,
+      "num_input_tokens_seen": 264574,
+      "step": 825,
+      "train_runtime": 2211.0468,
+      "train_tokens_per_second": 119.66
+    },
+    {
+      "epoch": 0.03797701149425287,
+      "grad_norm": 6.6527791023254395,
+      "learning_rate": 1.1779448621553885e-05,
+      "loss": 0.3283456563949585,
+      "num_input_tokens_seen": 264806,
+      "step": 826,
+      "train_runtime": 2212.9852,
+      "train_tokens_per_second": 119.66
+    },
+    {
+      "epoch": 0.038022988505747125,
+      "grad_norm": 8.250972747802734,
+      "learning_rate": 1.1769423558897244e-05,
+      "loss": 0.530316948890686,
+      "num_input_tokens_seen": 265078,
+      "step": 827,
+      "train_runtime": 2214.9409,
+      "train_tokens_per_second": 119.677
+    },
+    {
+      "epoch": 0.03806896551724138,
+      "grad_norm": 7.472963333129883,
+      "learning_rate": 1.1759398496240602e-05,
+      "loss": 0.4070558547973633,
+      "num_input_tokens_seen": 265378,
+      "step": 828,
+      "train_runtime": 2216.8972,
+      "train_tokens_per_second": 119.707
+    },
+    {
+      "epoch": 0.03811494252873563,
+      "grad_norm": 7.2969069480896,
+      "learning_rate": 1.174937343358396e-05,
+      "loss": 0.4591655433177948,
+      "num_input_tokens_seen": 265692,
+      "step": 829,
+      "train_runtime": 2218.856,
+      "train_tokens_per_second": 119.743
+    },
+    {
+      "epoch": 0.03816091954022988,
+      "grad_norm": 8.33116626739502,
+      "learning_rate": 1.173934837092732e-05,
+      "loss": 0.4876357913017273,
+      "num_input_tokens_seen": 266040,
+      "step": 830,
+      "train_runtime": 2220.8274,
+      "train_tokens_per_second": 119.793
+    },
+    {
+      "epoch": 0.038206896551724136,
+      "grad_norm": 7.567378997802734,
+      "learning_rate": 1.1729323308270678e-05,
+      "loss": 0.40954798460006714,
+      "num_input_tokens_seen": 266276,
+      "step": 831,
+      "train_runtime": 2222.7668,
+      "train_tokens_per_second": 119.795
+    },
+    {
+      "epoch": 0.03825287356321839,
+      "grad_norm": 7.296380043029785,
+      "learning_rate": 1.1719298245614036e-05,
+      "loss": 0.5455539226531982,
+      "num_input_tokens_seen": 266548,
+      "step": 832,
+      "train_runtime": 2224.7492,
+      "train_tokens_per_second": 119.81
+    },
+    {
+      "epoch": 0.03829885057471264,
+      "grad_norm": 7.5871453285217285,
+      "learning_rate": 1.1709273182957395e-05,
+      "loss": 0.4466758370399475,
+      "num_input_tokens_seen": 266902,
+      "step": 833,
+      "train_runtime": 2226.7381,
+      "train_tokens_per_second": 119.862
+    },
+    {
+      "epoch": 0.038344827586206894,
+      "grad_norm": 7.718501567840576,
+      "learning_rate": 1.1699248120300753e-05,
+      "loss": 0.4605112671852112,
+      "num_input_tokens_seen": 267194,
+      "step": 834,
+      "train_runtime": 2228.7241,
+      "train_tokens_per_second": 119.887
+    },
+    {
+      "epoch": 0.038390804597701146,
+      "grad_norm": 8.58492660522461,
+      "learning_rate": 1.168922305764411e-05,
+      "loss": 0.4204022288322449,
+      "num_input_tokens_seen": 267506,
+      "step": 835,
+      "train_runtime": 2230.6938,
+      "train_tokens_per_second": 119.921
+    },
+    {
+      "epoch": 0.038436781609195406,
+      "grad_norm": 7.482681751251221,
+      "learning_rate": 1.167919799498747e-05,
+      "loss": 0.35398271679878235,
+      "num_input_tokens_seen": 267818,
+      "step": 836,
+      "train_runtime": 2232.6484,
+      "train_tokens_per_second": 119.955
+    },
+    {
+      "epoch": 0.03848275862068966,
+      "grad_norm": 10.523682594299316,
+      "learning_rate": 1.1669172932330827e-05,
+      "loss": 0.36216866970062256,
+      "num_input_tokens_seen": 268074,
+      "step": 837,
+      "train_runtime": 2234.5813,
+      "train_tokens_per_second": 119.966
+    },
+    {
+      "epoch": 0.03852873563218391,
+      "grad_norm": 8.485894203186035,
+      "learning_rate": 1.1659147869674185e-05,
+      "loss": 0.4826376140117645,
+      "num_input_tokens_seen": 268360,
+      "step": 838,
+      "train_runtime": 2236.5648,
+      "train_tokens_per_second": 119.988
+    },
+    {
+      "epoch": 0.03857471264367816,
+      "grad_norm": 7.533479690551758,
+      "learning_rate": 1.1649122807017546e-05,
+      "loss": 0.41024893522262573,
+      "num_input_tokens_seen": 268596,
+      "step": 839,
+      "train_runtime": 2238.5326,
+      "train_tokens_per_second": 119.988
+    },
+    {
+      "epoch": 0.038620689655172416,
+      "grad_norm": 7.251370429992676,
+      "learning_rate": 1.1639097744360902e-05,
+      "loss": 0.43169522285461426,
+      "num_input_tokens_seen": 268908,
+      "step": 840,
+      "train_runtime": 2240.5802,
+      "train_tokens_per_second": 120.017
+    },
+    {
+      "epoch": 0.03866666666666667,
+      "grad_norm": 6.323627471923828,
+      "learning_rate": 1.1629072681704261e-05,
+      "loss": 0.38648054003715515,
+      "num_input_tokens_seen": 269244,
+      "step": 841,
+      "train_runtime": 2254.3827,
+      "train_tokens_per_second": 119.431
+    },
+    {
+      "epoch": 0.03871264367816092,
+      "grad_norm": 6.9223737716674805,
+      "learning_rate": 1.1619047619047621e-05,
+      "loss": 0.35775619745254517,
+      "num_input_tokens_seen": 269566,
+      "step": 842,
+      "train_runtime": 2256.3703,
+      "train_tokens_per_second": 119.469
+    },
+    {
+      "epoch": 0.038758620689655174,
+      "grad_norm": 6.996086597442627,
+      "learning_rate": 1.1609022556390978e-05,
+      "loss": 0.3565095365047455,
+      "num_input_tokens_seen": 269794,
+      "step": 843,
+      "train_runtime": 2258.435,
+      "train_tokens_per_second": 119.461
+    },
+    {
+      "epoch": 0.038804597701149426,
+      "grad_norm": 8.16794490814209,
+      "learning_rate": 1.1598997493734336e-05,
+      "loss": 0.5579286813735962,
+      "num_input_tokens_seen": 270066,
+      "step": 844,
+      "train_runtime": 2260.4194,
+      "train_tokens_per_second": 119.476
+    },
+    {
+      "epoch": 0.03885057471264368,
+      "grad_norm": 7.90415096282959,
+      "learning_rate": 1.1588972431077695e-05,
+      "loss": 0.37357866764068604,
+      "num_input_tokens_seen": 270376,
+      "step": 845,
+      "train_runtime": 2262.5671,
+      "train_tokens_per_second": 119.5
+    },
+    {
+      "epoch": 0.03889655172413793,
+      "grad_norm": 6.86454963684082,
+      "learning_rate": 1.1578947368421053e-05,
+      "loss": 0.37914904952049255,
+      "num_input_tokens_seen": 270630,
+      "step": 846,
+      "train_runtime": 2264.5082,
+      "train_tokens_per_second": 119.509
+    },
+    {
+      "epoch": 0.038942528735632184,
+      "grad_norm": 7.4026007652282715,
+      "learning_rate": 1.1568922305764412e-05,
+      "loss": 0.50594562292099,
+      "num_input_tokens_seen": 270982,
+      "step": 847,
+      "train_runtime": 2266.4979,
+      "train_tokens_per_second": 119.56
+    },
+    {
+      "epoch": 0.03898850574712644,
+      "grad_norm": 8.02122974395752,
+      "learning_rate": 1.1558897243107769e-05,
+      "loss": 0.6458178758621216,
+      "num_input_tokens_seen": 271352,
+      "step": 848,
+      "train_runtime": 2268.456,
+      "train_tokens_per_second": 119.62
+    },
+    {
+      "epoch": 0.03903448275862069,
+      "grad_norm": 10.095008850097656,
+      "learning_rate": 1.1548872180451129e-05,
+      "loss": 0.48162248730659485,
+      "num_input_tokens_seen": 271590,
+      "step": 849,
+      "train_runtime": 2270.3935,
+      "train_tokens_per_second": 119.622
+    },
+    {
+      "epoch": 0.03908045977011494,
+      "grad_norm": 7.111009120941162,
+      "learning_rate": 1.1538847117794487e-05,
+      "loss": 0.4771837592124939,
+      "num_input_tokens_seen": 271902,
+      "step": 850,
+      "train_runtime": 2272.35,
+      "train_tokens_per_second": 119.657
+    },
+    {
+      "epoch": 0.039126436781609195,
+      "grad_norm": 8.130017280578613,
+      "learning_rate": 1.1528822055137844e-05,
+      "loss": 0.39255622029304504,
+      "num_input_tokens_seen": 272218,
+      "step": 851,
+      "train_runtime": 2274.2978,
+      "train_tokens_per_second": 119.693
+    },
+    {
+      "epoch": 0.03917241379310345,
+      "grad_norm": 7.996945381164551,
+      "learning_rate": 1.1518796992481204e-05,
+      "loss": 0.496831476688385,
+      "num_input_tokens_seen": 272484,
+      "step": 852,
+      "train_runtime": 2276.2473,
+      "train_tokens_per_second": 119.708
+    },
+    {
+      "epoch": 0.0392183908045977,
+      "grad_norm": 9.667489051818848,
+      "learning_rate": 1.1508771929824563e-05,
+      "loss": 0.528260350227356,
+      "num_input_tokens_seen": 272852,
+      "step": 853,
+      "train_runtime": 2278.2113,
+      "train_tokens_per_second": 119.766
+    },
+    {
+      "epoch": 0.03926436781609195,
+      "grad_norm": 8.4678955078125,
+      "learning_rate": 1.149874686716792e-05,
+      "loss": 0.4786686301231384,
+      "num_input_tokens_seen": 273258,
+      "step": 854,
+      "train_runtime": 2280.318,
+      "train_tokens_per_second": 119.833
+    },
+    {
+      "epoch": 0.039310344827586205,
+      "grad_norm": 8.95380973815918,
+      "learning_rate": 1.1488721804511278e-05,
+      "loss": 0.5270786285400391,
+      "num_input_tokens_seen": 273560,
+      "step": 855,
+      "train_runtime": 2282.2789,
+      "train_tokens_per_second": 119.863
+    },
+    {
+      "epoch": 0.03935632183908046,
+      "grad_norm": 8.605949401855469,
+      "learning_rate": 1.1478696741854638e-05,
+      "loss": 0.48096340894699097,
+      "num_input_tokens_seen": 273866,
+      "step": 856,
+      "train_runtime": 2284.2383,
+      "train_tokens_per_second": 119.894
+    },
+    {
+      "epoch": 0.03940229885057471,
+      "grad_norm": 7.588695526123047,
+      "learning_rate": 1.1468671679197995e-05,
+      "loss": 0.42437565326690674,
+      "num_input_tokens_seen": 274128,
+      "step": 857,
+      "train_runtime": 2286.1711,
+      "train_tokens_per_second": 119.907
+    },
+    {
+      "epoch": 0.03944827586206896,
+      "grad_norm": 7.444614410400391,
+      "learning_rate": 1.1458646616541354e-05,
+      "loss": 0.40262681245803833,
+      "num_input_tokens_seen": 274414,
+      "step": 858,
+      "train_runtime": 2288.1158,
+      "train_tokens_per_second": 119.93
+    },
+    {
+      "epoch": 0.039494252873563215,
+      "grad_norm": 5.969850063323975,
+      "learning_rate": 1.1448621553884714e-05,
+      "loss": 0.32492494583129883,
+      "num_input_tokens_seen": 274716,
+      "step": 859,
+      "train_runtime": 2290.0704,
+      "train_tokens_per_second": 119.96
+    },
+    {
+      "epoch": 0.03954022988505747,
+      "grad_norm": 10.765917778015137,
+      "learning_rate": 1.143859649122807e-05,
+      "loss": 0.4458027482032776,
+      "num_input_tokens_seen": 275008,
+      "step": 860,
+      "train_runtime": 2292.0179,
+      "train_tokens_per_second": 119.985
+    },
+    {
+      "epoch": 0.03958620689655173,
+      "grad_norm": 7.071032524108887,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 0.3341474235057831,
+      "num_input_tokens_seen": 275294,
+      "step": 861,
+      "train_runtime": 2293.9718,
+      "train_tokens_per_second": 120.008
+    },
+    {
+      "epoch": 0.03963218390804598,
+      "grad_norm": 8.292301177978516,
+      "learning_rate": 1.141854636591479e-05,
+      "loss": 0.45281994342803955,
+      "num_input_tokens_seen": 275578,
+      "step": 862,
+      "train_runtime": 2295.9487,
+      "train_tokens_per_second": 120.028
+    },
+    {
+      "epoch": 0.03967816091954023,
+      "grad_norm": 7.089531898498535,
+      "learning_rate": 1.1408521303258146e-05,
+      "loss": 0.4461163282394409,
+      "num_input_tokens_seen": 275854,
+      "step": 863,
+      "train_runtime": 2297.9113,
+      "train_tokens_per_second": 120.046
+    },
+    {
+      "epoch": 0.039724137931034485,
+      "grad_norm": 8.957581520080566,
+      "learning_rate": 1.1398496240601504e-05,
+      "loss": 0.5779985189437866,
+      "num_input_tokens_seen": 276178,
+      "step": 864,
+      "train_runtime": 2299.8734,
+      "train_tokens_per_second": 120.084
+    },
+    {
+      "epoch": 0.03977011494252874,
+      "grad_norm": 6.947776794433594,
+      "learning_rate": 1.1388471177944861e-05,
+      "loss": 0.4939615726470947,
+      "num_input_tokens_seen": 276454,
+      "step": 865,
+      "train_runtime": 2301.8316,
+      "train_tokens_per_second": 120.102
+    },
+    {
+      "epoch": 0.03981609195402299,
+      "grad_norm": 7.037495136260986,
+      "learning_rate": 1.1378446115288221e-05,
+      "loss": 0.4288913905620575,
+      "num_input_tokens_seen": 276780,
+      "step": 866,
+      "train_runtime": 2303.7821,
+      "train_tokens_per_second": 120.142
+    },
+    {
+      "epoch": 0.03986206896551724,
+      "grad_norm": 7.895960330963135,
+      "learning_rate": 1.136842105263158e-05,
+      "loss": 0.4161183536052704,
+      "num_input_tokens_seen": 277064,
+      "step": 867,
+      "train_runtime": 2305.7235,
+      "train_tokens_per_second": 120.164
+    },
+    {
+      "epoch": 0.039908045977011496,
+      "grad_norm": 8.328717231750488,
+      "learning_rate": 1.1358395989974937e-05,
+      "loss": 0.5347382426261902,
+      "num_input_tokens_seen": 277346,
+      "step": 868,
+      "train_runtime": 2307.6615,
+      "train_tokens_per_second": 120.185
+    },
+    {
+      "epoch": 0.03995402298850575,
+      "grad_norm": 7.7827372550964355,
+      "learning_rate": 1.1348370927318297e-05,
+      "loss": 0.45566022396087646,
+      "num_input_tokens_seen": 277628,
+      "step": 869,
+      "train_runtime": 2309.6053,
+      "train_tokens_per_second": 120.206
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 7.058058261871338,
+      "learning_rate": 1.1338345864661655e-05,
+      "loss": 0.5263017416000366,
+      "num_input_tokens_seen": 278352,
+      "step": 870,
+      "train_runtime": 2311.646,
+      "train_tokens_per_second": 120.413
+    },
+    {
+      "epoch": 0.04004597701149425,
+      "grad_norm": 9.117223739624023,
+      "learning_rate": 1.1328320802005012e-05,
+      "loss": 0.6244474649429321,
+      "num_input_tokens_seen": 278674,
+      "step": 871,
+      "train_runtime": 2327.0496,
+      "train_tokens_per_second": 119.754
+    },
+    {
+      "epoch": 0.040091954022988506,
+      "grad_norm": 7.346180438995361,
+      "learning_rate": 1.1318295739348372e-05,
+      "loss": 0.5163547396659851,
+      "num_input_tokens_seen": 279032,
+      "step": 872,
+      "train_runtime": 2329.0357,
+      "train_tokens_per_second": 119.806
+    },
+    {
+      "epoch": 0.04013793103448276,
+      "grad_norm": 9.150981903076172,
+      "learning_rate": 1.1308270676691731e-05,
+      "loss": 0.560847282409668,
+      "num_input_tokens_seen": 279340,
+      "step": 873,
+      "train_runtime": 2331.085,
+      "train_tokens_per_second": 119.833
+    },
+    {
+      "epoch": 0.04018390804597701,
+      "grad_norm": 7.573378562927246,
+      "learning_rate": 1.1298245614035088e-05,
+      "loss": 0.4473652243614197,
+      "num_input_tokens_seen": 279662,
+      "step": 874,
+      "train_runtime": 2333.0629,
+      "train_tokens_per_second": 119.869
+    },
+    {
+      "epoch": 0.040229885057471264,
+      "grad_norm": 8.060361862182617,
+      "learning_rate": 1.1288220551378446e-05,
+      "loss": 0.5227193236351013,
+      "num_input_tokens_seen": 280008,
+      "step": 875,
+      "train_runtime": 2335.0286,
+      "train_tokens_per_second": 119.916
+    },
+    {
+      "epoch": 0.040275862068965516,
+      "grad_norm": 8.859715461730957,
+      "learning_rate": 1.1278195488721806e-05,
+      "loss": 0.5025198459625244,
+      "num_input_tokens_seen": 280276,
+      "step": 876,
+      "train_runtime": 2336.9747,
+      "train_tokens_per_second": 119.931
+    },
+    {
+      "epoch": 0.04032183908045977,
+      "grad_norm": 7.789861679077148,
+      "learning_rate": 1.1268170426065163e-05,
+      "loss": 0.4694594144821167,
+      "num_input_tokens_seen": 280654,
+      "step": 877,
+      "train_runtime": 2338.9943,
+      "train_tokens_per_second": 119.989
+    },
+    {
+      "epoch": 0.04036781609195402,
+      "grad_norm": 7.73496675491333,
+      "learning_rate": 1.1258145363408522e-05,
+      "loss": 0.47190093994140625,
+      "num_input_tokens_seen": 280980,
+      "step": 878,
+      "train_runtime": 2340.9878,
+      "train_tokens_per_second": 120.026
+    },
+    {
+      "epoch": 0.040413793103448274,
+      "grad_norm": 7.591188907623291,
+      "learning_rate": 1.1248120300751882e-05,
+      "loss": 0.4663242995738983,
+      "num_input_tokens_seen": 281300,
+      "step": 879,
+      "train_runtime": 2342.9632,
+      "train_tokens_per_second": 120.062
+    },
+    {
+      "epoch": 0.04045977011494253,
+      "grad_norm": 8.221147537231445,
+      "learning_rate": 1.1238095238095239e-05,
+      "loss": 0.40577393770217896,
+      "num_input_tokens_seen": 281588,
+      "step": 880,
+      "train_runtime": 2344.9185,
+      "train_tokens_per_second": 120.084
+    },
+    {
+      "epoch": 0.04050574712643678,
+      "grad_norm": 7.800092697143555,
+      "learning_rate": 1.1228070175438597e-05,
+      "loss": 0.532424807548523,
+      "num_input_tokens_seen": 281884,
+      "step": 881,
+      "train_runtime": 2346.9048,
+      "train_tokens_per_second": 120.109
+    },
+    {
+      "epoch": 0.04055172413793103,
+      "grad_norm": 7.333950519561768,
+      "learning_rate": 1.1218045112781957e-05,
+      "loss": 0.37188905477523804,
+      "num_input_tokens_seen": 282176,
+      "step": 882,
+      "train_runtime": 2348.9234,
+      "train_tokens_per_second": 120.13
+    },
+    {
+      "epoch": 0.040597701149425285,
+      "grad_norm": 11.310759544372559,
+      "learning_rate": 1.1208020050125314e-05,
+      "loss": 0.4371299147605896,
+      "num_input_tokens_seen": 282428,
+      "step": 883,
+      "train_runtime": 2350.8924,
+      "train_tokens_per_second": 120.137
+    },
+    {
+      "epoch": 0.04064367816091954,
+      "grad_norm": 7.542466163635254,
+      "learning_rate": 1.1197994987468673e-05,
+      "loss": 0.40072888135910034,
+      "num_input_tokens_seen": 282724,
+      "step": 884,
+      "train_runtime": 2352.8432,
+      "train_tokens_per_second": 120.163
+    },
+    {
+      "epoch": 0.04068965517241379,
+      "grad_norm": 9.259745597839355,
+      "learning_rate": 1.118796992481203e-05,
+      "loss": 0.355293869972229,
+      "num_input_tokens_seen": 282978,
+      "step": 885,
+      "train_runtime": 2354.7723,
+      "train_tokens_per_second": 120.172
+    },
+    {
+      "epoch": 0.04073563218390805,
+      "grad_norm": 7.640303611755371,
+      "learning_rate": 1.117794486215539e-05,
+      "loss": 0.5113065838813782,
+      "num_input_tokens_seen": 283250,
+      "step": 886,
+      "train_runtime": 2356.7209,
+      "train_tokens_per_second": 120.188
+    },
+    {
+      "epoch": 0.0407816091954023,
+      "grad_norm": 6.655959606170654,
+      "learning_rate": 1.1167919799498748e-05,
+      "loss": 0.45531749725341797,
+      "num_input_tokens_seen": 283712,
+      "step": 887,
+      "train_runtime": 2358.7731,
+      "train_tokens_per_second": 120.279
+    },
+    {
+      "epoch": 0.040827586206896554,
+      "grad_norm": 7.468194484710693,
+      "learning_rate": 1.1157894736842105e-05,
+      "loss": 0.5452451109886169,
+      "num_input_tokens_seen": 284258,
+      "step": 888,
+      "train_runtime": 2360.8264,
+      "train_tokens_per_second": 120.406
+    },
+    {
+      "epoch": 0.04087356321839081,
+      "grad_norm": 9.536760330200195,
+      "learning_rate": 1.1147869674185465e-05,
+      "loss": 0.5002374649047852,
+      "num_input_tokens_seen": 284534,
+      "step": 889,
+      "train_runtime": 2362.8485,
+      "train_tokens_per_second": 120.42
+    },
+    {
+      "epoch": 0.04091954022988506,
+      "grad_norm": 8.694121360778809,
+      "learning_rate": 1.1137844611528823e-05,
+      "loss": 0.5382827520370483,
+      "num_input_tokens_seen": 284830,
+      "step": 890,
+      "train_runtime": 2364.8225,
+      "train_tokens_per_second": 120.445
+    },
+    {
+      "epoch": 0.04096551724137931,
+      "grad_norm": 7.214402198791504,
+      "learning_rate": 1.112781954887218e-05,
+      "loss": 0.5347087383270264,
+      "num_input_tokens_seen": 285268,
+      "step": 891,
+      "train_runtime": 2366.9102,
+      "train_tokens_per_second": 120.523
+    },
+    {
+      "epoch": 0.041011494252873565,
+      "grad_norm": 8.898858070373535,
+      "learning_rate": 1.111779448621554e-05,
+      "loss": 0.6471966505050659,
+      "num_input_tokens_seen": 285612,
+      "step": 892,
+      "train_runtime": 2368.8961,
+      "train_tokens_per_second": 120.568
+    },
+    {
+      "epoch": 0.04105747126436782,
+      "grad_norm": 8.152880668640137,
+      "learning_rate": 1.1107769423558899e-05,
+      "loss": 0.42676907777786255,
+      "num_input_tokens_seen": 285870,
+      "step": 893,
+      "train_runtime": 2371.0213,
+      "train_tokens_per_second": 120.568
+    },
+    {
+      "epoch": 0.04110344827586207,
+      "grad_norm": 7.1052069664001465,
+      "learning_rate": 1.1097744360902256e-05,
+      "loss": 0.25924891233444214,
+      "num_input_tokens_seen": 286108,
+      "step": 894,
+      "train_runtime": 2373.0105,
+      "train_tokens_per_second": 120.568
+    },
+    {
+      "epoch": 0.04114942528735632,
+      "grad_norm": 6.890835285186768,
+      "learning_rate": 1.1087719298245614e-05,
+      "loss": 0.29447272419929504,
+      "num_input_tokens_seen": 286324,
+      "step": 895,
+      "train_runtime": 2374.944,
+      "train_tokens_per_second": 120.56
+    },
+    {
+      "epoch": 0.041195402298850575,
+      "grad_norm": 9.2092866897583,
+      "learning_rate": 1.1077694235588974e-05,
+      "loss": 0.4511667490005493,
+      "num_input_tokens_seen": 286624,
+      "step": 896,
+      "train_runtime": 2376.8871,
+      "train_tokens_per_second": 120.588
+    },
+    {
+      "epoch": 0.04124137931034483,
+      "grad_norm": 7.098134517669678,
+      "learning_rate": 1.1067669172932331e-05,
+      "loss": 0.44858479499816895,
+      "num_input_tokens_seen": 286898,
+      "step": 897,
+      "train_runtime": 2378.8327,
+      "train_tokens_per_second": 120.605
+    },
+    {
+      "epoch": 0.04128735632183908,
+      "grad_norm": 8.426091194152832,
+      "learning_rate": 1.105764411027569e-05,
+      "loss": 0.450019896030426,
+      "num_input_tokens_seen": 287224,
+      "step": 898,
+      "train_runtime": 2380.7909,
+      "train_tokens_per_second": 120.642
+    },
+    {
+      "epoch": 0.04133333333333333,
+      "grad_norm": 7.69377326965332,
+      "learning_rate": 1.104761904761905e-05,
+      "loss": 0.4973233938217163,
+      "num_input_tokens_seen": 287490,
+      "step": 899,
+      "train_runtime": 2382.7414,
+      "train_tokens_per_second": 120.655
+    },
+    {
+      "epoch": 0.041379310344827586,
+      "grad_norm": 7.684279441833496,
+      "learning_rate": 1.1037593984962407e-05,
+      "loss": 0.44500482082366943,
+      "num_input_tokens_seen": 287906,
+      "step": 900,
+      "train_runtime": 2384.7822,
+      "train_tokens_per_second": 120.726
+    },
+    {
+      "epoch": 0.04142528735632184,
+      "grad_norm": 7.631931304931641,
+      "learning_rate": 1.1027568922305765e-05,
+      "loss": 0.46322792768478394,
+      "num_input_tokens_seen": 288130,
+      "step": 901,
+      "train_runtime": 2400.0767,
+      "train_tokens_per_second": 120.05
+    },
+    {
+      "epoch": 0.04147126436781609,
+      "grad_norm": 6.782238483428955,
+      "learning_rate": 1.1017543859649125e-05,
+      "loss": 0.3592989444732666,
+      "num_input_tokens_seen": 288456,
+      "step": 902,
+      "train_runtime": 2402.0953,
+      "train_tokens_per_second": 120.085
+    },
+    {
+      "epoch": 0.04151724137931034,
+      "grad_norm": 8.23485279083252,
+      "learning_rate": 1.1007518796992482e-05,
+      "loss": 0.630439281463623,
+      "num_input_tokens_seen": 288722,
+      "step": 903,
+      "train_runtime": 2404.0699,
+      "train_tokens_per_second": 120.097
+    },
+    {
+      "epoch": 0.041563218390804596,
+      "grad_norm": 8.228728294372559,
+      "learning_rate": 1.099749373433584e-05,
+      "loss": 0.4678223133087158,
+      "num_input_tokens_seen": 289046,
+      "step": 904,
+      "train_runtime": 2406.0529,
+      "train_tokens_per_second": 120.133
+    },
+    {
+      "epoch": 0.04160919540229885,
+      "grad_norm": 7.0378499031066895,
+      "learning_rate": 1.0987468671679197e-05,
+      "loss": 0.42920780181884766,
+      "num_input_tokens_seen": 289366,
+      "step": 905,
+      "train_runtime": 2408.0205,
+      "train_tokens_per_second": 120.168
+    },
+    {
+      "epoch": 0.0416551724137931,
+      "grad_norm": 8.889481544494629,
+      "learning_rate": 1.0977443609022558e-05,
+      "loss": 0.4088243246078491,
+      "num_input_tokens_seen": 289702,
+      "step": 906,
+      "train_runtime": 2409.987,
+      "train_tokens_per_second": 120.209
+    },
+    {
+      "epoch": 0.041701149425287354,
+      "grad_norm": 9.026168823242188,
+      "learning_rate": 1.0967418546365916e-05,
+      "loss": 0.34462451934814453,
+      "num_input_tokens_seen": 289954,
+      "step": 907,
+      "train_runtime": 2411.9324,
+      "train_tokens_per_second": 120.216
+    },
+    {
+      "epoch": 0.041747126436781606,
+      "grad_norm": 7.272502899169922,
+      "learning_rate": 1.0957393483709273e-05,
+      "loss": 0.4410274624824524,
+      "num_input_tokens_seen": 290284,
+      "step": 908,
+      "train_runtime": 2413.8913,
+      "train_tokens_per_second": 120.256
+    },
+    {
+      "epoch": 0.04179310344827586,
+      "grad_norm": 7.197849273681641,
+      "learning_rate": 1.0947368421052633e-05,
+      "loss": 0.380460262298584,
+      "num_input_tokens_seen": 290576,
+      "step": 909,
+      "train_runtime": 2415.8372,
+      "train_tokens_per_second": 120.28
+    },
+    {
+      "epoch": 0.04183908045977011,
+      "grad_norm": 7.162438869476318,
+      "learning_rate": 1.0937343358395992e-05,
+      "loss": 0.4622759222984314,
+      "num_input_tokens_seen": 290928,
+      "step": 910,
+      "train_runtime": 2417.8277,
+      "train_tokens_per_second": 120.326
+    },
+    {
+      "epoch": 0.04188505747126437,
+      "grad_norm": 7.091503143310547,
+      "learning_rate": 1.0927318295739348e-05,
+      "loss": 0.3691563308238983,
+      "num_input_tokens_seen": 291200,
+      "step": 911,
+      "train_runtime": 2419.7731,
+      "train_tokens_per_second": 120.342
+    },
+    {
+      "epoch": 0.041931034482758624,
+      "grad_norm": 7.902462482452393,
+      "learning_rate": 1.0917293233082709e-05,
+      "loss": 0.5885430574417114,
+      "num_input_tokens_seen": 291552,
+      "step": 912,
+      "train_runtime": 2421.7607,
+      "train_tokens_per_second": 120.388
+    },
+    {
+      "epoch": 0.041977011494252876,
+      "grad_norm": 8.314557075500488,
+      "learning_rate": 1.0907268170426067e-05,
+      "loss": 0.4684392213821411,
+      "num_input_tokens_seen": 291806,
+      "step": 913,
+      "train_runtime": 2423.6947,
+      "train_tokens_per_second": 120.397
+    },
+    {
+      "epoch": 0.04202298850574713,
+      "grad_norm": 8.574982643127441,
+      "learning_rate": 1.0897243107769424e-05,
+      "loss": 0.4393140375614166,
+      "num_input_tokens_seen": 292054,
+      "step": 914,
+      "train_runtime": 2425.6381,
+      "train_tokens_per_second": 120.403
+    },
+    {
+      "epoch": 0.04206896551724138,
+      "grad_norm": 6.826984882354736,
+      "learning_rate": 1.0887218045112782e-05,
+      "loss": 0.3470788598060608,
+      "num_input_tokens_seen": 292344,
+      "step": 915,
+      "train_runtime": 2427.5802,
+      "train_tokens_per_second": 120.426
+    },
+    {
+      "epoch": 0.042114942528735634,
+      "grad_norm": 7.039104461669922,
+      "learning_rate": 1.0877192982456142e-05,
+      "loss": 0.35542216897010803,
+      "num_input_tokens_seen": 292608,
+      "step": 916,
+      "train_runtime": 2429.5518,
+      "train_tokens_per_second": 120.437
+    },
+    {
+      "epoch": 0.04216091954022989,
+      "grad_norm": 8.844517707824707,
+      "learning_rate": 1.08671679197995e-05,
+      "loss": 0.5223974585533142,
+      "num_input_tokens_seen": 292898,
+      "step": 917,
+      "train_runtime": 2431.5009,
+      "train_tokens_per_second": 120.46
+    },
+    {
+      "epoch": 0.04220689655172414,
+      "grad_norm": 9.181806564331055,
+      "learning_rate": 1.0857142857142858e-05,
+      "loss": 0.337760865688324,
+      "num_input_tokens_seen": 293172,
+      "step": 918,
+      "train_runtime": 2433.4407,
+      "train_tokens_per_second": 120.476
+    },
+    {
+      "epoch": 0.04225287356321839,
+      "grad_norm": 7.520045280456543,
+      "learning_rate": 1.0847117794486218e-05,
+      "loss": 0.44798052310943604,
+      "num_input_tokens_seen": 293460,
+      "step": 919,
+      "train_runtime": 2435.384,
+      "train_tokens_per_second": 120.498
+    },
+    {
+      "epoch": 0.042298850574712644,
+      "grad_norm": 8.774689674377441,
+      "learning_rate": 1.0837092731829575e-05,
+      "loss": 0.4620504379272461,
+      "num_input_tokens_seen": 293752,
+      "step": 920,
+      "train_runtime": 2437.3368,
+      "train_tokens_per_second": 120.522
+    },
+    {
+      "epoch": 0.0423448275862069,
+      "grad_norm": 7.388124465942383,
+      "learning_rate": 1.0827067669172933e-05,
+      "loss": 0.4084808826446533,
+      "num_input_tokens_seen": 294076,
+      "step": 921,
+      "train_runtime": 2439.2907,
+      "train_tokens_per_second": 120.558
+    },
+    {
+      "epoch": 0.04239080459770115,
+      "grad_norm": 8.380356788635254,
+      "learning_rate": 1.0817042606516293e-05,
+      "loss": 0.43804672360420227,
+      "num_input_tokens_seen": 294346,
+      "step": 922,
+      "train_runtime": 2441.2794,
+      "train_tokens_per_second": 120.57
+    },
+    {
+      "epoch": 0.0424367816091954,
+      "grad_norm": 8.176957130432129,
+      "learning_rate": 1.080701754385965e-05,
+      "loss": 0.39138269424438477,
+      "num_input_tokens_seen": 294646,
+      "step": 923,
+      "train_runtime": 2443.2453,
+      "train_tokens_per_second": 120.596
+    },
+    {
+      "epoch": 0.042482758620689655,
+      "grad_norm": 8.583574295043945,
+      "learning_rate": 1.0796992481203009e-05,
+      "loss": 0.5286657810211182,
+      "num_input_tokens_seen": 294962,
+      "step": 924,
+      "train_runtime": 2445.2019,
+      "train_tokens_per_second": 120.629
+    },
+    {
+      "epoch": 0.04252873563218391,
+      "grad_norm": 7.618959903717041,
+      "learning_rate": 1.0786967418546365e-05,
+      "loss": 0.4618549942970276,
+      "num_input_tokens_seen": 295252,
+      "step": 925,
+      "train_runtime": 2447.1606,
+      "train_tokens_per_second": 120.651
+    },
+    {
+      "epoch": 0.04257471264367816,
+      "grad_norm": 6.271094799041748,
+      "learning_rate": 1.0776942355889726e-05,
+      "loss": 0.5403990745544434,
+      "num_input_tokens_seen": 296004,
+      "step": 926,
+      "train_runtime": 2449.3437,
+      "train_tokens_per_second": 120.85
+    },
+    {
+      "epoch": 0.04262068965517241,
+      "grad_norm": 7.08641242980957,
+      "learning_rate": 1.0766917293233084e-05,
+      "loss": 0.49688249826431274,
+      "num_input_tokens_seen": 296336,
+      "step": 927,
+      "train_runtime": 2451.3155,
+      "train_tokens_per_second": 120.889
+    },
+    {
+      "epoch": 0.042666666666666665,
+      "grad_norm": 7.1499762535095215,
+      "learning_rate": 1.0756892230576441e-05,
+      "loss": 0.4259389042854309,
+      "num_input_tokens_seen": 296658,
+      "step": 928,
+      "train_runtime": 2453.3065,
+      "train_tokens_per_second": 120.922
+    },
+    {
+      "epoch": 0.04271264367816092,
+      "grad_norm": 10.639947891235352,
+      "learning_rate": 1.0746867167919801e-05,
+      "loss": 0.5289331674575806,
+      "num_input_tokens_seen": 297002,
+      "step": 929,
+      "train_runtime": 2455.2521,
+      "train_tokens_per_second": 120.966
+    },
+    {
+      "epoch": 0.04275862068965517,
+      "grad_norm": 8.506094932556152,
+      "learning_rate": 1.073684210526316e-05,
+      "loss": 0.3833962082862854,
+      "num_input_tokens_seen": 297270,
+      "step": 930,
+      "train_runtime": 2457.256,
+      "train_tokens_per_second": 120.976
+    },
+    {
+      "epoch": 0.04280459770114942,
+      "grad_norm": 7.21469783782959,
+      "learning_rate": 1.0726817042606516e-05,
+      "loss": 0.46723437309265137,
+      "num_input_tokens_seen": 297520,
+      "step": 931,
+      "train_runtime": 2472.0078,
+      "train_tokens_per_second": 120.356
+    },
+    {
+      "epoch": 0.042850574712643676,
+      "grad_norm": 9.229071617126465,
+      "learning_rate": 1.0716791979949877e-05,
+      "loss": 0.5288851261138916,
+      "num_input_tokens_seen": 297830,
+      "step": 932,
+      "train_runtime": 2474.085,
+      "train_tokens_per_second": 120.38
+    },
+    {
+      "epoch": 0.04289655172413793,
+      "grad_norm": 7.219783782958984,
+      "learning_rate": 1.0706766917293235e-05,
+      "loss": 0.38349175453186035,
+      "num_input_tokens_seen": 298080,
+      "step": 933,
+      "train_runtime": 2476.1395,
+      "train_tokens_per_second": 120.381
+    },
+    {
+      "epoch": 0.04294252873563218,
+      "grad_norm": 8.181379318237305,
+      "learning_rate": 1.0696741854636592e-05,
+      "loss": 0.43957898020744324,
+      "num_input_tokens_seen": 298350,
+      "step": 934,
+      "train_runtime": 2478.1833,
+      "train_tokens_per_second": 120.391
+    },
+    {
+      "epoch": 0.04298850574712643,
+      "grad_norm": 6.5870866775512695,
+      "learning_rate": 1.068671679197995e-05,
+      "loss": 0.4357609152793884,
+      "num_input_tokens_seen": 298650,
+      "step": 935,
+      "train_runtime": 2480.148,
+      "train_tokens_per_second": 120.416
+    },
+    {
+      "epoch": 0.04303448275862069,
+      "grad_norm": 9.388429641723633,
+      "learning_rate": 1.0676691729323309e-05,
+      "loss": 0.47772032022476196,
+      "num_input_tokens_seen": 298952,
+      "step": 936,
+      "train_runtime": 2482.158,
+      "train_tokens_per_second": 120.44
+    },
+    {
+      "epoch": 0.043080459770114946,
+      "grad_norm": 7.635393142700195,
+      "learning_rate": 1.0666666666666667e-05,
+      "loss": 0.46329423785209656,
+      "num_input_tokens_seen": 299284,
+      "step": 937,
+      "train_runtime": 2484.2258,
+      "train_tokens_per_second": 120.474
+    },
+    {
+      "epoch": 0.0431264367816092,
+      "grad_norm": 8.625504493713379,
+      "learning_rate": 1.0656641604010026e-05,
+      "loss": 0.5440782308578491,
+      "num_input_tokens_seen": 299538,
+      "step": 938,
+      "train_runtime": 2486.2547,
+      "train_tokens_per_second": 120.478
+    },
+    {
+      "epoch": 0.04317241379310345,
+      "grad_norm": 8.2759428024292,
+      "learning_rate": 1.0646616541353384e-05,
+      "loss": 0.5514028668403625,
+      "num_input_tokens_seen": 299882,
+      "step": 939,
+      "train_runtime": 2488.2888,
+      "train_tokens_per_second": 120.517
+    },
+    {
+      "epoch": 0.0432183908045977,
+      "grad_norm": 9.492950439453125,
+      "learning_rate": 1.0636591478696743e-05,
+      "loss": 0.3638257682323456,
+      "num_input_tokens_seen": 300106,
+      "step": 940,
+      "train_runtime": 2490.3154,
+      "train_tokens_per_second": 120.509
+    },
+    {
+      "epoch": 0.043264367816091956,
+      "grad_norm": 8.512650489807129,
+      "learning_rate": 1.0626566416040101e-05,
+      "loss": 0.6001561880111694,
+      "num_input_tokens_seen": 300458,
+      "step": 941,
+      "train_runtime": 2492.3559,
+      "train_tokens_per_second": 120.552
+    },
+    {
+      "epoch": 0.04331034482758621,
+      "grad_norm": 7.094176292419434,
+      "learning_rate": 1.061654135338346e-05,
+      "loss": 0.3695550262928009,
+      "num_input_tokens_seen": 300726,
+      "step": 942,
+      "train_runtime": 2494.3025,
+      "train_tokens_per_second": 120.565
+    },
+    {
+      "epoch": 0.04335632183908046,
+      "grad_norm": 7.242445468902588,
+      "learning_rate": 1.0606516290726818e-05,
+      "loss": 0.3477405607700348,
+      "num_input_tokens_seen": 301004,
+      "step": 943,
+      "train_runtime": 2496.4506,
+      "train_tokens_per_second": 120.573
+    },
+    {
+      "epoch": 0.043402298850574714,
+      "grad_norm": 8.440631866455078,
+      "learning_rate": 1.0596491228070177e-05,
+      "loss": 0.5115600824356079,
+      "num_input_tokens_seen": 301344,
+      "step": 944,
+      "train_runtime": 2498.4938,
+      "train_tokens_per_second": 120.61
+    },
+    {
+      "epoch": 0.043448275862068966,
+      "grad_norm": 7.042312145233154,
+      "learning_rate": 1.0586466165413534e-05,
+      "loss": 0.3534718155860901,
+      "num_input_tokens_seen": 301626,
+      "step": 945,
+      "train_runtime": 2500.5479,
+      "train_tokens_per_second": 120.624
+    },
+    {
+      "epoch": 0.04349425287356322,
+      "grad_norm": 6.0229668617248535,
+      "learning_rate": 1.0576441102756894e-05,
+      "loss": 0.3125123381614685,
+      "num_input_tokens_seen": 301920,
+      "step": 946,
+      "train_runtime": 2502.5541,
+      "train_tokens_per_second": 120.645
+    },
+    {
+      "epoch": 0.04354022988505747,
+      "grad_norm": 6.676801681518555,
+      "learning_rate": 1.0566416040100252e-05,
+      "loss": 0.4015813171863556,
+      "num_input_tokens_seen": 302176,
+      "step": 947,
+      "train_runtime": 2504.5478,
+      "train_tokens_per_second": 120.651
+    },
+    {
+      "epoch": 0.043586206896551724,
+      "grad_norm": 8.421010971069336,
+      "learning_rate": 1.0556390977443609e-05,
+      "loss": 0.4361885190010071,
+      "num_input_tokens_seen": 302488,
+      "step": 948,
+      "train_runtime": 2506.6159,
+      "train_tokens_per_second": 120.676
+    },
+    {
+      "epoch": 0.04363218390804598,
+      "grad_norm": 8.230731010437012,
+      "learning_rate": 1.054636591478697e-05,
+      "loss": 0.48423004150390625,
+      "num_input_tokens_seen": 302742,
+      "step": 949,
+      "train_runtime": 2508.6251,
+      "train_tokens_per_second": 120.68
+    },
+    {
+      "epoch": 0.04367816091954023,
+      "grad_norm": 10.06833267211914,
+      "learning_rate": 1.0536340852130326e-05,
+      "loss": 0.6311561465263367,
+      "num_input_tokens_seen": 303026,
+      "step": 950,
+      "train_runtime": 2510.6008,
+      "train_tokens_per_second": 120.699
+    },
+    {
+      "epoch": 0.04372413793103448,
+      "grad_norm": 6.6952738761901855,
+      "learning_rate": 1.0526315789473684e-05,
+      "loss": 0.3988801836967468,
+      "num_input_tokens_seen": 303292,
+      "step": 951,
+      "train_runtime": 2512.5476,
+      "train_tokens_per_second": 120.711
+    },
+    {
+      "epoch": 0.043770114942528734,
+      "grad_norm": 8.504010200500488,
+      "learning_rate": 1.0516290726817045e-05,
+      "loss": 0.6308104991912842,
+      "num_input_tokens_seen": 303608,
+      "step": 952,
+      "train_runtime": 2514.5226,
+      "train_tokens_per_second": 120.742
+    },
+    {
+      "epoch": 0.04381609195402299,
+      "grad_norm": 7.483700275421143,
+      "learning_rate": 1.0506265664160401e-05,
+      "loss": 0.46924853324890137,
+      "num_input_tokens_seen": 303850,
+      "step": 953,
+      "train_runtime": 2516.4748,
+      "train_tokens_per_second": 120.744
+    },
+    {
+      "epoch": 0.04386206896551724,
+      "grad_norm": 6.958237171173096,
+      "learning_rate": 1.049624060150376e-05,
+      "loss": 0.43178248405456543,
+      "num_input_tokens_seen": 304108,
+      "step": 954,
+      "train_runtime": 2518.4213,
+      "train_tokens_per_second": 120.753
+    },
+    {
+      "epoch": 0.04390804597701149,
+      "grad_norm": 6.699312686920166,
+      "learning_rate": 1.0486215538847118e-05,
+      "loss": 0.3953361511230469,
+      "num_input_tokens_seen": 304424,
+      "step": 955,
+      "train_runtime": 2520.4513,
+      "train_tokens_per_second": 120.782
+    },
+    {
+      "epoch": 0.043954022988505745,
+      "grad_norm": 6.884195804595947,
+      "learning_rate": 1.0476190476190477e-05,
+      "loss": 0.46768489480018616,
+      "num_input_tokens_seen": 304944,
+      "step": 956,
+      "train_runtime": 2522.5017,
+      "train_tokens_per_second": 120.89
+    },
+    {
+      "epoch": 0.044,
+      "grad_norm": 8.453701972961426,
+      "learning_rate": 1.0466165413533835e-05,
+      "loss": 0.49336737394332886,
+      "num_input_tokens_seen": 305222,
+      "step": 957,
+      "train_runtime": 2524.4489,
+      "train_tokens_per_second": 120.906
+    },
+    {
+      "epoch": 0.04404597701149425,
+      "grad_norm": 7.046376705169678,
+      "learning_rate": 1.0456140350877194e-05,
+      "loss": 0.4264727830886841,
+      "num_input_tokens_seen": 305522,
+      "step": 958,
+      "train_runtime": 2526.4038,
+      "train_tokens_per_second": 120.932
+    },
+    {
+      "epoch": 0.0440919540229885,
+      "grad_norm": 6.419041156768799,
+      "learning_rate": 1.0446115288220552e-05,
+      "loss": 0.39340972900390625,
+      "num_input_tokens_seen": 305974,
+      "step": 959,
+      "train_runtime": 2528.4093,
+      "train_tokens_per_second": 121.014
+    },
+    {
+      "epoch": 0.044137931034482755,
+      "grad_norm": 7.852453231811523,
+      "learning_rate": 1.043609022556391e-05,
+      "loss": 0.5217800140380859,
+      "num_input_tokens_seen": 306238,
+      "step": 960,
+      "train_runtime": 2530.3805,
+      "train_tokens_per_second": 121.024
+    },
+    {
+      "epoch": 0.044183908045977015,
+      "grad_norm": 7.855498313903809,
+      "learning_rate": 1.0426065162907268e-05,
+      "loss": 0.41073358058929443,
+      "num_input_tokens_seen": 306538,
+      "step": 961,
+      "train_runtime": 2545.1014,
+      "train_tokens_per_second": 120.442
+    },
+    {
+      "epoch": 0.04422988505747127,
+      "grad_norm": 9.012527465820312,
+      "learning_rate": 1.0416040100250628e-05,
+      "loss": 0.606840968132019,
+      "num_input_tokens_seen": 306788,
+      "step": 962,
+      "train_runtime": 2547.0491,
+      "train_tokens_per_second": 120.448
+    },
+    {
+      "epoch": 0.04427586206896552,
+      "grad_norm": 6.915031433105469,
+      "learning_rate": 1.0406015037593986e-05,
+      "loss": 0.2802923321723938,
+      "num_input_tokens_seen": 307022,
+      "step": 963,
+      "train_runtime": 2549.002,
+      "train_tokens_per_second": 120.448
+    },
+    {
+      "epoch": 0.04432183908045977,
+      "grad_norm": 8.415030479431152,
+      "learning_rate": 1.0395989974937343e-05,
+      "loss": 0.5813425779342651,
+      "num_input_tokens_seen": 307332,
+      "step": 964,
+      "train_runtime": 2551.0138,
+      "train_tokens_per_second": 120.474
+    },
+    {
+      "epoch": 0.044367816091954025,
+      "grad_norm": 6.746933460235596,
+      "learning_rate": 1.0385964912280702e-05,
+      "loss": 0.3976888954639435,
+      "num_input_tokens_seen": 307602,
+      "step": 965,
+      "train_runtime": 2552.9589,
+      "train_tokens_per_second": 120.488
+    },
+    {
+      "epoch": 0.04441379310344828,
+      "grad_norm": 7.740719795227051,
+      "learning_rate": 1.0375939849624062e-05,
+      "loss": 0.4100640118122101,
+      "num_input_tokens_seen": 307882,
+      "step": 966,
+      "train_runtime": 2555.0131,
+      "train_tokens_per_second": 120.501
+    },
+    {
+      "epoch": 0.04445977011494253,
+      "grad_norm": 7.349896430969238,
+      "learning_rate": 1.0365914786967419e-05,
+      "loss": 0.35742422938346863,
+      "num_input_tokens_seen": 308236,
+      "step": 967,
+      "train_runtime": 2557.119,
+      "train_tokens_per_second": 120.54
+    },
+    {
+      "epoch": 0.04450574712643678,
+      "grad_norm": 7.031965732574463,
+      "learning_rate": 1.0355889724310777e-05,
+      "loss": 0.5171423554420471,
+      "num_input_tokens_seen": 308604,
+      "step": 968,
+      "train_runtime": 2559.168,
+      "train_tokens_per_second": 120.588
+    },
+    {
+      "epoch": 0.044551724137931036,
+      "grad_norm": 7.60201358795166,
+      "learning_rate": 1.0345864661654137e-05,
+      "loss": 0.30621081590652466,
+      "num_input_tokens_seen": 308816,
+      "step": 969,
+      "train_runtime": 2561.1545,
+      "train_tokens_per_second": 120.577
+    },
+    {
+      "epoch": 0.04459770114942529,
+      "grad_norm": 6.667523384094238,
+      "learning_rate": 1.0335839598997494e-05,
+      "loss": 0.39199793338775635,
+      "num_input_tokens_seen": 309152,
+      "step": 970,
+      "train_runtime": 2563.1336,
+      "train_tokens_per_second": 120.615
+    },
+    {
+      "epoch": 0.04464367816091954,
+      "grad_norm": 8.55335807800293,
+      "learning_rate": 1.0325814536340853e-05,
+      "loss": 0.4693825840950012,
+      "num_input_tokens_seen": 309474,
+      "step": 971,
+      "train_runtime": 2565.0905,
+      "train_tokens_per_second": 120.648
+    },
+    {
+      "epoch": 0.04468965517241379,
+      "grad_norm": 11.010098457336426,
+      "learning_rate": 1.0315789473684213e-05,
+      "loss": 0.5334758758544922,
+      "num_input_tokens_seen": 309782,
+      "step": 972,
+      "train_runtime": 2567.0642,
+      "train_tokens_per_second": 120.676
+    },
+    {
+      "epoch": 0.044735632183908046,
+      "grad_norm": 7.299499034881592,
+      "learning_rate": 1.030576441102757e-05,
+      "loss": 0.692724883556366,
+      "num_input_tokens_seen": 310288,
+      "step": 973,
+      "train_runtime": 2569.0533,
+      "train_tokens_per_second": 120.779
+    },
+    {
+      "epoch": 0.0447816091954023,
+      "grad_norm": 7.25387716293335,
+      "learning_rate": 1.0295739348370928e-05,
+      "loss": 0.5500563979148865,
+      "num_input_tokens_seen": 310646,
+      "step": 974,
+      "train_runtime": 2571.0177,
+      "train_tokens_per_second": 120.826
+    },
+    {
+      "epoch": 0.04482758620689655,
+      "grad_norm": 7.856181621551514,
+      "learning_rate": 1.0285714285714285e-05,
+      "loss": 0.5135598182678223,
+      "num_input_tokens_seen": 310980,
+      "step": 975,
+      "train_runtime": 2573.1115,
+      "train_tokens_per_second": 120.858
+    },
+    {
+      "epoch": 0.044873563218390804,
+      "grad_norm": 10.145916938781738,
+      "learning_rate": 1.0275689223057645e-05,
+      "loss": 0.5267456769943237,
+      "num_input_tokens_seen": 311268,
+      "step": 976,
+      "train_runtime": 2575.0716,
+      "train_tokens_per_second": 120.877
+    },
+    {
+      "epoch": 0.044919540229885056,
+      "grad_norm": 7.2765913009643555,
+      "learning_rate": 1.0265664160401003e-05,
+      "loss": 0.34480470418930054,
+      "num_input_tokens_seen": 311602,
+      "step": 977,
+      "train_runtime": 2577.067,
+      "train_tokens_per_second": 120.913
+    },
+    {
+      "epoch": 0.04496551724137931,
+      "grad_norm": 6.886968612670898,
+      "learning_rate": 1.025563909774436e-05,
+      "loss": 0.39189067482948303,
+      "num_input_tokens_seen": 311864,
+      "step": 978,
+      "train_runtime": 2579.019,
+      "train_tokens_per_second": 120.923
+    },
+    {
+      "epoch": 0.04501149425287356,
+      "grad_norm": 8.318005561828613,
+      "learning_rate": 1.024561403508772e-05,
+      "loss": 0.44863125681877136,
+      "num_input_tokens_seen": 312140,
+      "step": 979,
+      "train_runtime": 2580.9644,
+      "train_tokens_per_second": 120.939
+    },
+    {
+      "epoch": 0.045057471264367814,
+      "grad_norm": 8.14795207977295,
+      "learning_rate": 1.0235588972431079e-05,
+      "loss": 0.5619063377380371,
+      "num_input_tokens_seen": 312410,
+      "step": 980,
+      "train_runtime": 2582.9212,
+      "train_tokens_per_second": 120.952
+    },
+    {
+      "epoch": 0.04510344827586207,
+      "grad_norm": 7.044286251068115,
+      "learning_rate": 1.0225563909774436e-05,
+      "loss": 0.4981921911239624,
+      "num_input_tokens_seen": 312826,
+      "step": 981,
+      "train_runtime": 2584.9074,
+      "train_tokens_per_second": 121.02
+    },
+    {
+      "epoch": 0.04514942528735632,
+      "grad_norm": 7.485442161560059,
+      "learning_rate": 1.0215538847117796e-05,
+      "loss": 0.513433575630188,
+      "num_input_tokens_seen": 313166,
+      "step": 982,
+      "train_runtime": 2586.9505,
+      "train_tokens_per_second": 121.056
+    },
+    {
+      "epoch": 0.04519540229885057,
+      "grad_norm": 7.272876739501953,
+      "learning_rate": 1.0205513784461154e-05,
+      "loss": 0.4137285053730011,
+      "num_input_tokens_seen": 313438,
+      "step": 983,
+      "train_runtime": 2588.914,
+      "train_tokens_per_second": 121.069
+    },
+    {
+      "epoch": 0.045241379310344824,
+      "grad_norm": 7.248037815093994,
+      "learning_rate": 1.0195488721804511e-05,
+      "loss": 0.4240766167640686,
+      "num_input_tokens_seen": 313780,
+      "step": 984,
+      "train_runtime": 2590.935,
+      "train_tokens_per_second": 121.107
+    },
+    {
+      "epoch": 0.04528735632183908,
+      "grad_norm": 9.385568618774414,
+      "learning_rate": 1.018546365914787e-05,
+      "loss": 0.6178704500198364,
+      "num_input_tokens_seen": 314138,
+      "step": 985,
+      "train_runtime": 2592.9684,
+      "train_tokens_per_second": 121.15
+    },
+    {
+      "epoch": 0.04533333333333334,
+      "grad_norm": 8.061515808105469,
+      "learning_rate": 1.017543859649123e-05,
+      "loss": 0.44137874245643616,
+      "num_input_tokens_seen": 314496,
+      "step": 986,
+      "train_runtime": 2594.9659,
+      "train_tokens_per_second": 121.195
+    },
+    {
+      "epoch": 0.04537931034482759,
+      "grad_norm": 7.891003131866455,
+      "learning_rate": 1.0165413533834587e-05,
+      "loss": 0.45911043882369995,
+      "num_input_tokens_seen": 314790,
+      "step": 987,
+      "train_runtime": 2596.919,
+      "train_tokens_per_second": 121.217
+    },
+    {
+      "epoch": 0.04542528735632184,
+      "grad_norm": 7.65133810043335,
+      "learning_rate": 1.0155388471177945e-05,
+      "loss": 0.39683860540390015,
+      "num_input_tokens_seen": 315008,
+      "step": 988,
+      "train_runtime": 2598.8565,
+      "train_tokens_per_second": 121.21
+    },
+    {
+      "epoch": 0.045471264367816094,
+      "grad_norm": 6.9997076988220215,
+      "learning_rate": 1.0145363408521305e-05,
+      "loss": 0.40674012899398804,
+      "num_input_tokens_seen": 315338,
+      "step": 989,
+      "train_runtime": 2600.8158,
+      "train_tokens_per_second": 121.246
+    },
+    {
+      "epoch": 0.04551724137931035,
+      "grad_norm": 9.910818099975586,
+      "learning_rate": 1.0135338345864662e-05,
+      "loss": 0.5541273355484009,
+      "num_input_tokens_seen": 315702,
+      "step": 990,
+      "train_runtime": 2602.7546,
+      "train_tokens_per_second": 121.295
+    },
+    {
+      "epoch": 0.0455632183908046,
+      "grad_norm": 7.932651519775391,
+      "learning_rate": 1.012531328320802e-05,
+      "loss": 0.44719380140304565,
+      "num_input_tokens_seen": 316008,
+      "step": 991,
+      "train_runtime": 2617.923,
+      "train_tokens_per_second": 120.709
+    },
+    {
+      "epoch": 0.04560919540229885,
+      "grad_norm": 8.502070426940918,
+      "learning_rate": 1.011528822055138e-05,
+      "loss": 0.59912109375,
+      "num_input_tokens_seen": 316354,
+      "step": 992,
+      "train_runtime": 2619.9059,
+      "train_tokens_per_second": 120.75
+    },
+    {
+      "epoch": 0.045655172413793105,
+      "grad_norm": 8.09085464477539,
+      "learning_rate": 1.0105263157894738e-05,
+      "loss": 0.5389885306358337,
+      "num_input_tokens_seen": 316728,
+      "step": 993,
+      "train_runtime": 2621.892,
+      "train_tokens_per_second": 120.801
+    },
+    {
+      "epoch": 0.04570114942528736,
+      "grad_norm": 6.942469596862793,
+      "learning_rate": 1.0095238095238096e-05,
+      "loss": 0.3687466084957123,
+      "num_input_tokens_seen": 316986,
+      "step": 994,
+      "train_runtime": 2623.8331,
+      "train_tokens_per_second": 120.81
+    },
+    {
+      "epoch": 0.04574712643678161,
+      "grad_norm": 7.85964298248291,
+      "learning_rate": 1.0085213032581453e-05,
+      "loss": 0.6065055131912231,
+      "num_input_tokens_seen": 317410,
+      "step": 995,
+      "train_runtime": 2625.9026,
+      "train_tokens_per_second": 120.877
+    },
+    {
+      "epoch": 0.04579310344827586,
+      "grad_norm": 6.808088779449463,
+      "learning_rate": 1.0075187969924813e-05,
+      "loss": 0.42792007327079773,
+      "num_input_tokens_seen": 317712,
+      "step": 996,
+      "train_runtime": 2627.9333,
+      "train_tokens_per_second": 120.898
+    },
+    {
+      "epoch": 0.045839080459770115,
+      "grad_norm": 7.471822261810303,
+      "learning_rate": 1.0065162907268171e-05,
+      "loss": 0.5098698735237122,
+      "num_input_tokens_seen": 318052,
+      "step": 997,
+      "train_runtime": 2629.9432,
+      "train_tokens_per_second": 120.935
+    },
+    {
+      "epoch": 0.04588505747126437,
+      "grad_norm": 8.29421615600586,
+      "learning_rate": 1.0055137844611528e-05,
+      "loss": 0.4087499976158142,
+      "num_input_tokens_seen": 318288,
+      "step": 998,
+      "train_runtime": 2631.9251,
+      "train_tokens_per_second": 120.934
+    },
+    {
+      "epoch": 0.04593103448275862,
+      "grad_norm": 8.272956848144531,
+      "learning_rate": 1.0045112781954888e-05,
+      "loss": 0.640299379825592,
+      "num_input_tokens_seen": 318644,
+      "step": 999,
+      "train_runtime": 2633.9459,
+      "train_tokens_per_second": 120.976
+    },
+    {
+      "epoch": 0.04597701149425287,
+      "grad_norm": 7.522319316864014,
+      "learning_rate": 1.0035087719298247e-05,
+      "loss": 0.47425469756126404,
+      "num_input_tokens_seen": 318948,
+      "step": 1000,
+      "train_runtime": 2635.9461,
+      "train_tokens_per_second": 120.999
+    },
+    {
+      "epoch": 0.04597701149425287,
+      "eval_loss": 1.7173823118209839,
+      "eval_runtime": 55.1446,
+      "eval_samples_per_second": 18.134,
+      "eval_steps_per_second": 9.067,
+      "num_input_tokens_seen": 318948,
+      "step": 1000
+    },
+    {
+      "epoch": 0.046022988505747126,
+      "grad_norm": 6.9273600578308105,
+      "learning_rate": 1.0025062656641604e-05,
+      "loss": 0.46524420380592346,
+      "num_input_tokens_seen": 319260,
+      "step": 1001,
+      "train_runtime": 2693.2393,
+      "train_tokens_per_second": 118.541
+    },
+    {
+      "epoch": 0.04606896551724138,
+      "grad_norm": 13.683260917663574,
+      "learning_rate": 1.0015037593984964e-05,
+      "loss": 0.5733553171157837,
+      "num_input_tokens_seen": 319556,
+      "step": 1002,
+      "train_runtime": 2695.2467,
+      "train_tokens_per_second": 118.563
+    },
+    {
+      "epoch": 0.04611494252873563,
+      "grad_norm": 7.395852088928223,
+      "learning_rate": 1.0005012531328322e-05,
+      "loss": 0.4862205684185028,
+      "num_input_tokens_seen": 319984,
+      "step": 1003,
+      "train_runtime": 2697.3607,
+      "train_tokens_per_second": 118.629
+    },
+    {
+      "epoch": 0.04616091954022988,
+      "grad_norm": 6.674742698669434,
+      "learning_rate": 9.99498746867168e-06,
+      "loss": 0.4223348796367645,
+      "num_input_tokens_seen": 320264,
+      "step": 1004,
+      "train_runtime": 2699.3785,
+      "train_tokens_per_second": 118.644
+    },
+    {
+      "epoch": 0.046206896551724136,
+      "grad_norm": 7.067044734954834,
+      "learning_rate": 9.984962406015038e-06,
+      "loss": 0.36694225668907166,
+      "num_input_tokens_seen": 320532,
+      "step": 1005,
+      "train_runtime": 2701.4024,
+      "train_tokens_per_second": 118.654
+    },
+    {
+      "epoch": 0.04625287356321839,
+      "grad_norm": 7.723959922790527,
+      "learning_rate": 9.974937343358396e-06,
+      "loss": 0.4250967800617218,
+      "num_input_tokens_seen": 320852,
+      "step": 1006,
+      "train_runtime": 2703.3783,
+      "train_tokens_per_second": 118.686
+    },
+    {
+      "epoch": 0.04629885057471264,
+      "grad_norm": 6.667568206787109,
+      "learning_rate": 9.964912280701755e-06,
+      "loss": 0.3907022178173065,
+      "num_input_tokens_seen": 321148,
+      "step": 1007,
+      "train_runtime": 2705.3199,
+      "train_tokens_per_second": 118.71
+    },
+    {
+      "epoch": 0.046344827586206894,
+      "grad_norm": 7.259830951690674,
+      "learning_rate": 9.954887218045113e-06,
+      "loss": 0.37166735529899597,
+      "num_input_tokens_seen": 321460,
+      "step": 1008,
+      "train_runtime": 2707.2957,
+      "train_tokens_per_second": 118.738
+    },
+    {
+      "epoch": 0.046390804597701146,
+      "grad_norm": 8.259964942932129,
+      "learning_rate": 9.944862155388472e-06,
+      "loss": 0.4947519898414612,
+      "num_input_tokens_seen": 321724,
+      "step": 1009,
+      "train_runtime": 2709.2475,
+      "train_tokens_per_second": 118.75
+    },
+    {
+      "epoch": 0.0464367816091954,
+      "grad_norm": 6.72380256652832,
+      "learning_rate": 9.93483709273183e-06,
+      "loss": 0.2749987840652466,
+      "num_input_tokens_seen": 321966,
+      "step": 1010,
+      "train_runtime": 2711.1998,
+      "train_tokens_per_second": 118.754
+    },
+    {
+      "epoch": 0.04648275862068966,
+      "grad_norm": 7.118277072906494,
+      "learning_rate": 9.924812030075189e-06,
+      "loss": 0.3629249036312103,
+      "num_input_tokens_seen": 322232,
+      "step": 1011,
+      "train_runtime": 2713.1547,
+      "train_tokens_per_second": 118.767
+    },
+    {
+      "epoch": 0.04652873563218391,
+      "grad_norm": 6.580169677734375,
+      "learning_rate": 9.914786967418547e-06,
+      "loss": 0.35471388697624207,
+      "num_input_tokens_seen": 322632,
+      "step": 1012,
+      "train_runtime": 2715.1573,
+      "train_tokens_per_second": 118.826
+    },
+    {
+      "epoch": 0.046574712643678164,
+      "grad_norm": 7.097537994384766,
+      "learning_rate": 9.904761904761906e-06,
+      "loss": 0.3747185170650482,
+      "num_input_tokens_seen": 322882,
+      "step": 1013,
+      "train_runtime": 2717.1019,
+      "train_tokens_per_second": 118.833
+    },
+    {
+      "epoch": 0.046620689655172416,
+      "grad_norm": 6.884876251220703,
+      "learning_rate": 9.894736842105264e-06,
+      "loss": 0.49874210357666016,
+      "num_input_tokens_seen": 323314,
+      "step": 1014,
+      "train_runtime": 2719.1045,
+      "train_tokens_per_second": 118.905
+    },
+    {
+      "epoch": 0.04666666666666667,
+      "grad_norm": 8.016444206237793,
+      "learning_rate": 9.884711779448623e-06,
+      "loss": 0.5174450278282166,
+      "num_input_tokens_seen": 323672,
+      "step": 1015,
+      "train_runtime": 2721.096,
+      "train_tokens_per_second": 118.949
+    },
+    {
+      "epoch": 0.04671264367816092,
+      "grad_norm": 9.015275001525879,
+      "learning_rate": 9.87468671679198e-06,
+      "loss": 0.3318605422973633,
+      "num_input_tokens_seen": 323916,
+      "step": 1016,
+      "train_runtime": 2723.0452,
+      "train_tokens_per_second": 118.954
+    },
+    {
+      "epoch": 0.046758620689655174,
+      "grad_norm": 7.451302528381348,
+      "learning_rate": 9.86466165413534e-06,
+      "loss": 0.35202082991600037,
+      "num_input_tokens_seen": 324194,
+      "step": 1017,
+      "train_runtime": 2725.0022,
+      "train_tokens_per_second": 118.97
+    },
+    {
+      "epoch": 0.04680459770114943,
+      "grad_norm": 6.906904697418213,
+      "learning_rate": 9.854636591478698e-06,
+      "loss": 0.5809968709945679,
+      "num_input_tokens_seen": 324726,
+      "step": 1018,
+      "train_runtime": 2727.0309,
+      "train_tokens_per_second": 119.077
+    },
+    {
+      "epoch": 0.04685057471264368,
+      "grad_norm": 7.594064235687256,
+      "learning_rate": 9.844611528822055e-06,
+      "loss": 0.43761134147644043,
+      "num_input_tokens_seen": 325010,
+      "step": 1019,
+      "train_runtime": 2729.0043,
+      "train_tokens_per_second": 119.095
+    },
+    {
+      "epoch": 0.04689655172413793,
+      "grad_norm": 7.633775234222412,
+      "learning_rate": 9.834586466165415e-06,
+      "loss": 0.3381976783275604,
+      "num_input_tokens_seen": 325346,
+      "step": 1020,
+      "train_runtime": 2731.1163,
+      "train_tokens_per_second": 119.126
+    },
+    {
+      "epoch": 0.046942528735632184,
+      "grad_norm": 7.1371002197265625,
+      "learning_rate": 9.824561403508772e-06,
+      "loss": 0.5091744065284729,
+      "num_input_tokens_seen": 326268,
+      "step": 1021,
+      "train_runtime": 2746.5022,
+      "train_tokens_per_second": 118.794
+    },
+    {
+      "epoch": 0.04698850574712644,
+      "grad_norm": 7.4820051193237305,
+      "learning_rate": 9.81453634085213e-06,
+      "loss": 0.45310309529304504,
+      "num_input_tokens_seen": 326540,
+      "step": 1022,
+      "train_runtime": 2748.5398,
+      "train_tokens_per_second": 118.805
+    },
+    {
+      "epoch": 0.04703448275862069,
+      "grad_norm": 7.082658767700195,
+      "learning_rate": 9.80451127819549e-06,
+      "loss": 0.36116427183151245,
+      "num_input_tokens_seen": 326768,
+      "step": 1023,
+      "train_runtime": 2750.5045,
+      "train_tokens_per_second": 118.803
+    },
+    {
+      "epoch": 0.04708045977011494,
+      "grad_norm": 5.537309169769287,
+      "learning_rate": 9.794486215538847e-06,
+      "loss": 0.3114401698112488,
+      "num_input_tokens_seen": 326992,
+      "step": 1024,
+      "train_runtime": 2752.4864,
+      "train_tokens_per_second": 118.799
+    },
+    {
+      "epoch": 0.047126436781609195,
+      "grad_norm": 8.36307144165039,
+      "learning_rate": 9.784461152882206e-06,
+      "loss": 0.355997771024704,
+      "num_input_tokens_seen": 327218,
+      "step": 1025,
+      "train_runtime": 2754.4641,
+      "train_tokens_per_second": 118.796
+    },
+    {
+      "epoch": 0.04717241379310345,
+      "grad_norm": 7.312787055969238,
+      "learning_rate": 9.774436090225564e-06,
+      "loss": 0.3455454707145691,
+      "num_input_tokens_seen": 327508,
+      "step": 1026,
+      "train_runtime": 2756.4876,
+      "train_tokens_per_second": 118.814
+    },
+    {
+      "epoch": 0.0472183908045977,
+      "grad_norm": 7.130361557006836,
+      "learning_rate": 9.764411027568923e-06,
+      "loss": 0.4309677183628082,
+      "num_input_tokens_seen": 327826,
+      "step": 1027,
+      "train_runtime": 2758.4651,
+      "train_tokens_per_second": 118.844
+    },
+    {
+      "epoch": 0.04726436781609195,
+      "grad_norm": 7.263982772827148,
+      "learning_rate": 9.754385964912281e-06,
+      "loss": 0.39605748653411865,
+      "num_input_tokens_seen": 328062,
+      "step": 1028,
+      "train_runtime": 2760.4029,
+      "train_tokens_per_second": 118.846
+    },
+    {
+      "epoch": 0.047310344827586205,
+      "grad_norm": 6.753585338592529,
+      "learning_rate": 9.74436090225564e-06,
+      "loss": 0.3983078896999359,
+      "num_input_tokens_seen": 328386,
+      "step": 1029,
+      "train_runtime": 2762.4619,
+      "train_tokens_per_second": 118.874
+    },
+    {
+      "epoch": 0.04735632183908046,
+      "grad_norm": 7.322591781616211,
+      "learning_rate": 9.734335839598998e-06,
+      "loss": 0.3312835395336151,
+      "num_input_tokens_seen": 328610,
+      "step": 1030,
+      "train_runtime": 2764.4537,
+      "train_tokens_per_second": 118.87
+    },
+    {
+      "epoch": 0.04740229885057471,
+      "grad_norm": 7.22749137878418,
+      "learning_rate": 9.724310776942357e-06,
+      "loss": 0.377351850271225,
+      "num_input_tokens_seen": 328878,
+      "step": 1031,
+      "train_runtime": 2766.4086,
+      "train_tokens_per_second": 118.883
+    },
+    {
+      "epoch": 0.04744827586206896,
+      "grad_norm": 7.166253089904785,
+      "learning_rate": 9.714285714285715e-06,
+      "loss": 0.43618372082710266,
+      "num_input_tokens_seen": 329226,
+      "step": 1032,
+      "train_runtime": 2768.3822,
+      "train_tokens_per_second": 118.924
+    },
+    {
+      "epoch": 0.047494252873563216,
+      "grad_norm": 7.763054847717285,
+      "learning_rate": 9.704260651629074e-06,
+      "loss": 0.48253417015075684,
+      "num_input_tokens_seen": 329488,
+      "step": 1033,
+      "train_runtime": 2770.3245,
+      "train_tokens_per_second": 118.935
+    },
+    {
+      "epoch": 0.04754022988505747,
+      "grad_norm": 7.806169033050537,
+      "learning_rate": 9.694235588972432e-06,
+      "loss": 0.40141862630844116,
+      "num_input_tokens_seen": 329786,
+      "step": 1034,
+      "train_runtime": 2772.3003,
+      "train_tokens_per_second": 118.958
+    },
+    {
+      "epoch": 0.04758620689655172,
+      "grad_norm": 7.727939605712891,
+      "learning_rate": 9.68421052631579e-06,
+      "loss": 0.43414315581321716,
+      "num_input_tokens_seen": 330084,
+      "step": 1035,
+      "train_runtime": 2774.2547,
+      "train_tokens_per_second": 118.981
+    },
+    {
+      "epoch": 0.04763218390804598,
+      "grad_norm": 9.545808792114258,
+      "learning_rate": 9.674185463659147e-06,
+      "loss": 0.6307319402694702,
+      "num_input_tokens_seen": 330398,
+      "step": 1036,
+      "train_runtime": 2776.2079,
+      "train_tokens_per_second": 119.011
+    },
+    {
+      "epoch": 0.04767816091954023,
+      "grad_norm": 11.138605117797852,
+      "learning_rate": 9.664160401002508e-06,
+      "loss": 0.518451452255249,
+      "num_input_tokens_seen": 330694,
+      "step": 1037,
+      "train_runtime": 2778.272,
+      "train_tokens_per_second": 119.029
+    },
+    {
+      "epoch": 0.047724137931034485,
+      "grad_norm": 7.458420276641846,
+      "learning_rate": 9.654135338345866e-06,
+      "loss": 0.5234805941581726,
+      "num_input_tokens_seen": 331048,
+      "step": 1038,
+      "train_runtime": 2780.2557,
+      "train_tokens_per_second": 119.071
+    },
+    {
+      "epoch": 0.04777011494252874,
+      "grad_norm": 7.09022855758667,
+      "learning_rate": 9.644110275689223e-06,
+      "loss": 0.40761256217956543,
+      "num_input_tokens_seen": 331406,
+      "step": 1039,
+      "train_runtime": 2782.298,
+      "train_tokens_per_second": 119.112
+    },
+    {
+      "epoch": 0.04781609195402299,
+      "grad_norm": 11.615792274475098,
+      "learning_rate": 9.634085213032583e-06,
+      "loss": 0.46781405806541443,
+      "num_input_tokens_seen": 331688,
+      "step": 1040,
+      "train_runtime": 2784.2822,
+      "train_tokens_per_second": 119.129
+    },
+    {
+      "epoch": 0.04786206896551724,
+      "grad_norm": 5.815955638885498,
+      "learning_rate": 9.62406015037594e-06,
+      "loss": 0.2744513154029846,
+      "num_input_tokens_seen": 331908,
+      "step": 1041,
+      "train_runtime": 2786.2316,
+      "train_tokens_per_second": 119.124
+    },
+    {
+      "epoch": 0.047908045977011496,
+      "grad_norm": 8.99447250366211,
+      "learning_rate": 9.614035087719298e-06,
+      "loss": 0.3928970694541931,
+      "num_input_tokens_seen": 332148,
+      "step": 1042,
+      "train_runtime": 2788.1834,
+      "train_tokens_per_second": 119.127
+    },
+    {
+      "epoch": 0.04795402298850575,
+      "grad_norm": 8.52641773223877,
+      "learning_rate": 9.604010025062659e-06,
+      "loss": 0.4255361557006836,
+      "num_input_tokens_seen": 332452,
+      "step": 1043,
+      "train_runtime": 2790.2292,
+      "train_tokens_per_second": 119.149
+    },
+    {
+      "epoch": 0.048,
+      "grad_norm": 6.427272319793701,
+      "learning_rate": 9.593984962406015e-06,
+      "loss": 0.28434616327285767,
+      "num_input_tokens_seen": 332712,
+      "step": 1044,
+      "train_runtime": 2792.2759,
+      "train_tokens_per_second": 119.154
+    },
+    {
+      "epoch": 0.048045977011494254,
+      "grad_norm": 8.600190162658691,
+      "learning_rate": 9.583959899749374e-06,
+      "loss": 0.44470202922821045,
+      "num_input_tokens_seen": 332982,
+      "step": 1045,
+      "train_runtime": 2794.3692,
+      "train_tokens_per_second": 119.162
+    },
+    {
+      "epoch": 0.048091954022988506,
+      "grad_norm": 7.377002716064453,
+      "learning_rate": 9.573934837092732e-06,
+      "loss": 0.42285972833633423,
+      "num_input_tokens_seen": 333386,
+      "step": 1046,
+      "train_runtime": 2796.3878,
+      "train_tokens_per_second": 119.22
+    },
+    {
+      "epoch": 0.04813793103448276,
+      "grad_norm": 8.298364639282227,
+      "learning_rate": 9.56390977443609e-06,
+      "loss": 0.5232506990432739,
+      "num_input_tokens_seen": 333702,
+      "step": 1047,
+      "train_runtime": 2798.5052,
+      "train_tokens_per_second": 119.243
+    },
+    {
+      "epoch": 0.04818390804597701,
+      "grad_norm": 9.52237606048584,
+      "learning_rate": 9.55388471177945e-06,
+      "loss": 0.4316413998603821,
+      "num_input_tokens_seen": 334004,
+      "step": 1048,
+      "train_runtime": 2800.5729,
+      "train_tokens_per_second": 119.263
+    },
+    {
+      "epoch": 0.048229885057471264,
+      "grad_norm": 8.141901969909668,
+      "learning_rate": 9.543859649122808e-06,
+      "loss": 0.36646947264671326,
+      "num_input_tokens_seen": 334246,
+      "step": 1049,
+      "train_runtime": 2802.565,
+      "train_tokens_per_second": 119.264
+    },
+    {
+      "epoch": 0.04827586206896552,
+      "grad_norm": 8.07127857208252,
+      "learning_rate": 9.533834586466166e-06,
+      "loss": 0.49001407623291016,
+      "num_input_tokens_seen": 334584,
+      "step": 1050,
+      "train_runtime": 2804.5894,
+      "train_tokens_per_second": 119.299
+    },
+    {
+      "epoch": 0.04832183908045977,
+      "grad_norm": 8.100993156433105,
+      "learning_rate": 9.523809523809525e-06,
+      "loss": 0.45338425040245056,
+      "num_input_tokens_seen": 334854,
+      "step": 1051,
+      "train_runtime": 2820.4966,
+      "train_tokens_per_second": 118.722
+    },
+    {
+      "epoch": 0.04836781609195402,
+      "grad_norm": 8.30936336517334,
+      "learning_rate": 9.513784461152883e-06,
+      "loss": 0.43239009380340576,
+      "num_input_tokens_seen": 335062,
+      "step": 1052,
+      "train_runtime": 2822.5391,
+      "train_tokens_per_second": 118.709
+    },
+    {
+      "epoch": 0.048413793103448274,
+      "grad_norm": 10.822750091552734,
+      "learning_rate": 9.503759398496242e-06,
+      "loss": 0.5045852661132812,
+      "num_input_tokens_seen": 335386,
+      "step": 1053,
+      "train_runtime": 2824.5594,
+      "train_tokens_per_second": 118.739
+    },
+    {
+      "epoch": 0.04845977011494253,
+      "grad_norm": 7.888216972351074,
+      "learning_rate": 9.4937343358396e-06,
+      "loss": 0.47344547510147095,
+      "num_input_tokens_seen": 335774,
+      "step": 1054,
+      "train_runtime": 2826.614,
+      "train_tokens_per_second": 118.79
+    },
+    {
+      "epoch": 0.04850574712643678,
+      "grad_norm": 8.200318336486816,
+      "learning_rate": 9.483709273182959e-06,
+      "loss": 0.590822160243988,
+      "num_input_tokens_seen": 336072,
+      "step": 1055,
+      "train_runtime": 2828.6399,
+      "train_tokens_per_second": 118.81
+    },
+    {
+      "epoch": 0.04855172413793103,
+      "grad_norm": 8.955785751342773,
+      "learning_rate": 9.473684210526315e-06,
+      "loss": 0.5345810055732727,
+      "num_input_tokens_seen": 336352,
+      "step": 1056,
+      "train_runtime": 2830.597,
+      "train_tokens_per_second": 118.827
+    },
+    {
+      "epoch": 0.048597701149425285,
+      "grad_norm": 7.858545303344727,
+      "learning_rate": 9.463659147869676e-06,
+      "loss": 0.4652344584465027,
+      "num_input_tokens_seen": 336634,
+      "step": 1057,
+      "train_runtime": 2832.5933,
+      "train_tokens_per_second": 118.843
+    },
+    {
+      "epoch": 0.04864367816091954,
+      "grad_norm": 6.894465446472168,
+      "learning_rate": 9.453634085213034e-06,
+      "loss": 0.37855982780456543,
+      "num_input_tokens_seen": 336922,
+      "step": 1058,
+      "train_runtime": 2834.5698,
+      "train_tokens_per_second": 118.862
+    },
+    {
+      "epoch": 0.04868965517241379,
+      "grad_norm": 8.946717262268066,
+      "learning_rate": 9.443609022556391e-06,
+      "loss": 0.5390365123748779,
+      "num_input_tokens_seen": 337180,
+      "step": 1059,
+      "train_runtime": 2836.6011,
+      "train_tokens_per_second": 118.868
+    },
+    {
+      "epoch": 0.04873563218390804,
+      "grad_norm": 8.382868766784668,
+      "learning_rate": 9.43358395989975e-06,
+      "loss": 0.5391836762428284,
+      "num_input_tokens_seen": 337460,
+      "step": 1060,
+      "train_runtime": 2838.5884,
+      "train_tokens_per_second": 118.883
+    },
+    {
+      "epoch": 0.0487816091954023,
+      "grad_norm": 7.397761821746826,
+      "learning_rate": 9.423558897243108e-06,
+      "loss": 0.40352505445480347,
+      "num_input_tokens_seen": 337818,
+      "step": 1061,
+      "train_runtime": 2840.6271,
+      "train_tokens_per_second": 118.924
+    },
+    {
+      "epoch": 0.048827586206896555,
+      "grad_norm": 9.155498504638672,
+      "learning_rate": 9.413533834586466e-06,
+      "loss": 0.5041525959968567,
+      "num_input_tokens_seen": 338152,
+      "step": 1062,
+      "train_runtime": 2842.6538,
+      "train_tokens_per_second": 118.956
+    },
+    {
+      "epoch": 0.04887356321839081,
+      "grad_norm": 8.839866638183594,
+      "learning_rate": 9.403508771929825e-06,
+      "loss": 0.5147091150283813,
+      "num_input_tokens_seen": 338490,
+      "step": 1063,
+      "train_runtime": 2844.726,
+      "train_tokens_per_second": 118.989
+    },
+    {
+      "epoch": 0.04891954022988506,
+      "grad_norm": 7.973474502563477,
+      "learning_rate": 9.393483709273183e-06,
+      "loss": 0.47533437609672546,
+      "num_input_tokens_seen": 338778,
+      "step": 1064,
+      "train_runtime": 2846.6857,
+      "train_tokens_per_second": 119.008
+    },
+    {
+      "epoch": 0.04896551724137931,
+      "grad_norm": 7.850538730621338,
+      "learning_rate": 9.383458646616542e-06,
+      "loss": 0.3330311179161072,
+      "num_input_tokens_seen": 339058,
+      "step": 1065,
+      "train_runtime": 2848.6412,
+      "train_tokens_per_second": 119.024
+    },
+    {
+      "epoch": 0.049011494252873565,
+      "grad_norm": 7.935583114624023,
+      "learning_rate": 9.3734335839599e-06,
+      "loss": 0.4993591904640198,
+      "num_input_tokens_seen": 339354,
+      "step": 1066,
+      "train_runtime": 2850.6322,
+      "train_tokens_per_second": 119.045
+    },
+    {
+      "epoch": 0.04905747126436782,
+      "grad_norm": 8.978984832763672,
+      "learning_rate": 9.363408521303259e-06,
+      "loss": 0.5230007171630859,
+      "num_input_tokens_seen": 339706,
+      "step": 1067,
+      "train_runtime": 2852.6395,
+      "train_tokens_per_second": 119.085
+    },
+    {
+      "epoch": 0.04910344827586207,
+      "grad_norm": 9.016287803649902,
+      "learning_rate": 9.353383458646617e-06,
+      "loss": 0.4591520130634308,
+      "num_input_tokens_seen": 339994,
+      "step": 1068,
+      "train_runtime": 2854.609,
+      "train_tokens_per_second": 119.104
+    },
+    {
+      "epoch": 0.04914942528735632,
+      "grad_norm": 7.878653526306152,
+      "learning_rate": 9.343358395989976e-06,
+      "loss": 0.38075506687164307,
+      "num_input_tokens_seen": 340262,
+      "step": 1069,
+      "train_runtime": 2856.6093,
+      "train_tokens_per_second": 119.114
+    },
+    {
+      "epoch": 0.049195402298850575,
+      "grad_norm": 15.90587043762207,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 0.45033928751945496,
+      "num_input_tokens_seen": 340498,
+      "step": 1070,
+      "train_runtime": 2858.5511,
+      "train_tokens_per_second": 119.116
+    },
+    {
+      "epoch": 0.04924137931034483,
+      "grad_norm": 8.365623474121094,
+      "learning_rate": 9.323308270676693e-06,
+      "loss": 0.46389204263687134,
+      "num_input_tokens_seen": 340732,
+      "step": 1071,
+      "train_runtime": 2860.4917,
+      "train_tokens_per_second": 119.117
+    },
+    {
+      "epoch": 0.04928735632183908,
+      "grad_norm": 7.068108558654785,
+      "learning_rate": 9.313283208020051e-06,
+      "loss": 0.40667521953582764,
+      "num_input_tokens_seen": 341012,
+      "step": 1072,
+      "train_runtime": 2862.4885,
+      "train_tokens_per_second": 119.131
+    },
+    {
+      "epoch": 0.04933333333333333,
+      "grad_norm": 6.176692485809326,
+      "learning_rate": 9.30325814536341e-06,
+      "loss": 0.25460129976272583,
+      "num_input_tokens_seen": 341244,
+      "step": 1073,
+      "train_runtime": 2864.4842,
+      "train_tokens_per_second": 119.129
+    },
+    {
+      "epoch": 0.049379310344827586,
+      "grad_norm": 8.334495544433594,
+      "learning_rate": 9.293233082706767e-06,
+      "loss": 0.42830508947372437,
+      "num_input_tokens_seen": 341490,
+      "step": 1074,
+      "train_runtime": 2866.439,
+      "train_tokens_per_second": 119.134
+    },
+    {
+      "epoch": 0.04942528735632184,
+      "grad_norm": 8.23608112335205,
+      "learning_rate": 9.283208020050127e-06,
+      "loss": 0.5677105784416199,
+      "num_input_tokens_seen": 341828,
+      "step": 1075,
+      "train_runtime": 2868.4276,
+      "train_tokens_per_second": 119.169
+    },
+    {
+      "epoch": 0.04947126436781609,
+      "grad_norm": 15.787196159362793,
+      "learning_rate": 9.273182957393484e-06,
+      "loss": 0.5085170269012451,
+      "num_input_tokens_seen": 342084,
+      "step": 1076,
+      "train_runtime": 2870.3745,
+      "train_tokens_per_second": 119.177
+    },
+    {
+      "epoch": 0.049517241379310344,
+      "grad_norm": 7.366590976715088,
+      "learning_rate": 9.263157894736842e-06,
+      "loss": 0.4475935101509094,
+      "num_input_tokens_seen": 342496,
+      "step": 1077,
+      "train_runtime": 2872.4136,
+      "train_tokens_per_second": 119.236
+    },
+    {
+      "epoch": 0.049563218390804596,
+      "grad_norm": 8.59300708770752,
+      "learning_rate": 9.253132832080202e-06,
+      "loss": 0.40707725286483765,
+      "num_input_tokens_seen": 342794,
+      "step": 1078,
+      "train_runtime": 2874.4167,
+      "train_tokens_per_second": 119.257
+    },
+    {
+      "epoch": 0.04960919540229885,
+      "grad_norm": 7.537459850311279,
+      "learning_rate": 9.243107769423559e-06,
+      "loss": 0.2980315089225769,
+      "num_input_tokens_seen": 343024,
+      "step": 1079,
+      "train_runtime": 2876.3932,
+      "train_tokens_per_second": 119.255
+    },
+    {
+      "epoch": 0.0496551724137931,
+      "grad_norm": 8.121476173400879,
+      "learning_rate": 9.233082706766918e-06,
+      "loss": 0.37251514196395874,
+      "num_input_tokens_seen": 343362,
+      "step": 1080,
+      "train_runtime": 2878.349,
+      "train_tokens_per_second": 119.291
+    },
+    {
+      "epoch": 0.049701149425287354,
+      "grad_norm": 6.9334940910339355,
+      "learning_rate": 9.223057644110276e-06,
+      "loss": 0.3058217763900757,
+      "num_input_tokens_seen": 343578,
+      "step": 1081,
+      "train_runtime": 2893.554,
+      "train_tokens_per_second": 118.739
+    },
+    {
+      "epoch": 0.04974712643678161,
+      "grad_norm": 7.499663352966309,
+      "learning_rate": 9.213032581453634e-06,
+      "loss": 0.43159013986587524,
+      "num_input_tokens_seen": 343852,
+      "step": 1082,
+      "train_runtime": 2895.5143,
+      "train_tokens_per_second": 118.753
+    },
+    {
+      "epoch": 0.04979310344827586,
+      "grad_norm": 7.372053146362305,
+      "learning_rate": 9.203007518796993e-06,
+      "loss": 0.3495287597179413,
+      "num_input_tokens_seen": 344100,
+      "step": 1083,
+      "train_runtime": 2897.5735,
+      "train_tokens_per_second": 118.755
+    },
+    {
+      "epoch": 0.04983908045977011,
+      "grad_norm": 6.331181049346924,
+      "learning_rate": 9.192982456140351e-06,
+      "loss": 0.42055273056030273,
+      "num_input_tokens_seen": 344528,
+      "step": 1084,
+      "train_runtime": 2899.5819,
+      "train_tokens_per_second": 118.82
+    },
+    {
+      "epoch": 0.049885057471264364,
+      "grad_norm": 7.463900566101074,
+      "learning_rate": 9.18295739348371e-06,
+      "loss": 0.510176956653595,
+      "num_input_tokens_seen": 344788,
+      "step": 1085,
+      "train_runtime": 2901.5421,
+      "train_tokens_per_second": 118.829
+    },
+    {
+      "epoch": 0.049931034482758624,
+      "grad_norm": 9.023658752441406,
+      "learning_rate": 9.172932330827068e-06,
+      "loss": 0.45956721901893616,
+      "num_input_tokens_seen": 345018,
+      "step": 1086,
+      "train_runtime": 2903.489,
+      "train_tokens_per_second": 118.829
+    },
+    {
+      "epoch": 0.049977011494252876,
+      "grad_norm": 7.309625148773193,
+      "learning_rate": 9.162907268170427e-06,
+      "loss": 0.47155243158340454,
+      "num_input_tokens_seen": 345366,
+      "step": 1087,
+      "train_runtime": 2905.5344,
+      "train_tokens_per_second": 118.865
+    },
+    {
+      "epoch": 0.05002298850574713,
+      "grad_norm": 7.602726936340332,
+      "learning_rate": 9.152882205513785e-06,
+      "loss": 0.3295368254184723,
+      "num_input_tokens_seen": 345568,
+      "step": 1088,
+      "train_runtime": 2907.5286,
+      "train_tokens_per_second": 118.853
+    },
+    {
+      "epoch": 0.05006896551724138,
+      "grad_norm": 10.183154106140137,
+      "learning_rate": 9.142857142857144e-06,
+      "loss": 0.4740368127822876,
+      "num_input_tokens_seen": 345860,
+      "step": 1089,
+      "train_runtime": 2909.4781,
+      "train_tokens_per_second": 118.874
+    },
+    {
+      "epoch": 0.050114942528735634,
+      "grad_norm": 8.111772537231445,
+      "learning_rate": 9.132832080200502e-06,
+      "loss": 0.48555850982666016,
+      "num_input_tokens_seen": 346158,
+      "step": 1090,
+      "train_runtime": 2911.4362,
+      "train_tokens_per_second": 118.896
+    },
+    {
+      "epoch": 0.05016091954022989,
+      "grad_norm": 8.932589530944824,
+      "learning_rate": 9.12280701754386e-06,
+      "loss": 0.5541291832923889,
+      "num_input_tokens_seen": 346458,
+      "step": 1091,
+      "train_runtime": 2913.395,
+      "train_tokens_per_second": 118.919
+    },
+    {
+      "epoch": 0.05020689655172414,
+      "grad_norm": 6.192288875579834,
+      "learning_rate": 9.11278195488722e-06,
+      "loss": 0.3381243348121643,
+      "num_input_tokens_seen": 346682,
+      "step": 1092,
+      "train_runtime": 2915.3275,
+      "train_tokens_per_second": 118.917
+    },
+    {
+      "epoch": 0.05025287356321839,
+      "grad_norm": 7.6365485191345215,
+      "learning_rate": 9.102756892230578e-06,
+      "loss": 0.3119629919528961,
+      "num_input_tokens_seen": 346906,
+      "step": 1093,
+      "train_runtime": 2917.3278,
+      "train_tokens_per_second": 118.912
+    },
+    {
+      "epoch": 0.050298850574712645,
+      "grad_norm": 8.041963577270508,
+      "learning_rate": 9.092731829573935e-06,
+      "loss": 0.3600906729698181,
+      "num_input_tokens_seen": 347214,
+      "step": 1094,
+      "train_runtime": 2919.3592,
+      "train_tokens_per_second": 118.935
+    },
+    {
+      "epoch": 0.0503448275862069,
+      "grad_norm": 7.429046630859375,
+      "learning_rate": 9.082706766917295e-06,
+      "loss": 0.44096723198890686,
+      "num_input_tokens_seen": 347524,
+      "step": 1095,
+      "train_runtime": 2921.3277,
+      "train_tokens_per_second": 118.961
+    },
+    {
+      "epoch": 0.05039080459770115,
+      "grad_norm": 8.05813980102539,
+      "learning_rate": 9.072681704260652e-06,
+      "loss": 0.6170783042907715,
+      "num_input_tokens_seen": 347912,
+      "step": 1096,
+      "train_runtime": 2923.3265,
+      "train_tokens_per_second": 119.012
+    },
+    {
+      "epoch": 0.0504367816091954,
+      "grad_norm": 9.138258934020996,
+      "learning_rate": 9.06265664160401e-06,
+      "loss": 0.6430239677429199,
+      "num_input_tokens_seen": 348210,
+      "step": 1097,
+      "train_runtime": 2925.2762,
+      "train_tokens_per_second": 119.035
+    },
+    {
+      "epoch": 0.050482758620689655,
+      "grad_norm": 7.271088123321533,
+      "learning_rate": 9.05263157894737e-06,
+      "loss": 0.3724372088909149,
+      "num_input_tokens_seen": 348472,
+      "step": 1098,
+      "train_runtime": 2927.2245,
+      "train_tokens_per_second": 119.045
+    },
+    {
+      "epoch": 0.05052873563218391,
+      "grad_norm": 9.22109317779541,
+      "learning_rate": 9.042606516290727e-06,
+      "loss": 0.5635727643966675,
+      "num_input_tokens_seen": 348792,
+      "step": 1099,
+      "train_runtime": 2929.1772,
+      "train_tokens_per_second": 119.075
+    },
+    {
+      "epoch": 0.05057471264367816,
+      "grad_norm": 7.555117130279541,
+      "learning_rate": 9.032581453634086e-06,
+      "loss": 0.46877145767211914,
+      "num_input_tokens_seen": 349152,
+      "step": 1100,
+      "train_runtime": 2931.1636,
+      "train_tokens_per_second": 119.117
+    },
+    {
+      "epoch": 0.05062068965517241,
+      "grad_norm": 7.465952396392822,
+      "learning_rate": 9.022556390977444e-06,
+      "loss": 0.414106547832489,
+      "num_input_tokens_seen": 349442,
+      "step": 1101,
+      "train_runtime": 2933.1308,
+      "train_tokens_per_second": 119.136
+    },
+    {
+      "epoch": 0.050666666666666665,
+      "grad_norm": 8.254798889160156,
+      "learning_rate": 9.012531328320803e-06,
+      "loss": 0.5444568395614624,
+      "num_input_tokens_seen": 349754,
+      "step": 1102,
+      "train_runtime": 2935.0995,
+      "train_tokens_per_second": 119.163
+    },
+    {
+      "epoch": 0.05071264367816092,
+      "grad_norm": 7.881249904632568,
+      "learning_rate": 9.002506265664161e-06,
+      "loss": 0.44626545906066895,
+      "num_input_tokens_seen": 350048,
+      "step": 1103,
+      "train_runtime": 2937.0817,
+      "train_tokens_per_second": 119.182
+    },
+    {
+      "epoch": 0.05075862068965517,
+      "grad_norm": 7.48320198059082,
+      "learning_rate": 8.99248120300752e-06,
+      "loss": 0.3517981469631195,
+      "num_input_tokens_seen": 350280,
+      "step": 1104,
+      "train_runtime": 2939.0352,
+      "train_tokens_per_second": 119.182
+    },
+    {
+      "epoch": 0.05080459770114942,
+      "grad_norm": 8.440960884094238,
+      "learning_rate": 8.982456140350878e-06,
+      "loss": 0.4097956418991089,
+      "num_input_tokens_seen": 350572,
+      "step": 1105,
+      "train_runtime": 2940.995,
+      "train_tokens_per_second": 119.202
+    },
+    {
+      "epoch": 0.050850574712643676,
+      "grad_norm": 8.25269603729248,
+      "learning_rate": 8.972431077694236e-06,
+      "loss": 0.48889172077178955,
+      "num_input_tokens_seen": 350902,
+      "step": 1106,
+      "train_runtime": 2942.9575,
+      "train_tokens_per_second": 119.234
+    },
+    {
+      "epoch": 0.05089655172413793,
+      "grad_norm": 7.786253452301025,
+      "learning_rate": 8.962406015037595e-06,
+      "loss": 0.395815908908844,
+      "num_input_tokens_seen": 351170,
+      "step": 1107,
+      "train_runtime": 2944.9076,
+      "train_tokens_per_second": 119.247
+    },
+    {
+      "epoch": 0.05094252873563218,
+      "grad_norm": 9.107592582702637,
+      "learning_rate": 8.952380952380953e-06,
+      "loss": 0.4389643967151642,
+      "num_input_tokens_seen": 351470,
+      "step": 1108,
+      "train_runtime": 2946.8827,
+      "train_tokens_per_second": 119.268
+    },
+    {
+      "epoch": 0.050988505747126434,
+      "grad_norm": 7.5205769538879395,
+      "learning_rate": 8.942355889724312e-06,
+      "loss": 0.4152507781982422,
+      "num_input_tokens_seen": 351762,
+      "step": 1109,
+      "train_runtime": 2948.8932,
+      "train_tokens_per_second": 119.286
+    },
+    {
+      "epoch": 0.05103448275862069,
+      "grad_norm": 7.165195465087891,
+      "learning_rate": 8.93233082706767e-06,
+      "loss": 0.30179163813591003,
+      "num_input_tokens_seen": 352052,
+      "step": 1110,
+      "train_runtime": 2950.8661,
+      "train_tokens_per_second": 119.305
+    },
+    {
+      "epoch": 0.051080459770114946,
+      "grad_norm": 6.335844993591309,
+      "learning_rate": 8.922305764411027e-06,
+      "loss": 0.3719693422317505,
+      "num_input_tokens_seen": 352372,
+      "step": 1111,
+      "train_runtime": 2965.7246,
+      "train_tokens_per_second": 118.815
+    },
+    {
+      "epoch": 0.0511264367816092,
+      "grad_norm": 7.419812202453613,
+      "learning_rate": 8.912280701754387e-06,
+      "loss": 0.4624994397163391,
+      "num_input_tokens_seen": 352686,
+      "step": 1112,
+      "train_runtime": 2967.9535,
+      "train_tokens_per_second": 118.831
+    },
+    {
+      "epoch": 0.05117241379310345,
+      "grad_norm": 6.95173454284668,
+      "learning_rate": 8.902255639097746e-06,
+      "loss": 0.4526553153991699,
+      "num_input_tokens_seen": 353018,
+      "step": 1113,
+      "train_runtime": 2970.0208,
+      "train_tokens_per_second": 118.86
+    },
+    {
+      "epoch": 0.051218390804597703,
+      "grad_norm": 8.418750762939453,
+      "learning_rate": 8.892230576441103e-06,
+      "loss": 0.511723518371582,
+      "num_input_tokens_seen": 353334,
+      "step": 1114,
+      "train_runtime": 2971.9865,
+      "train_tokens_per_second": 118.888
+    },
+    {
+      "epoch": 0.051264367816091956,
+      "grad_norm": 10.712956428527832,
+      "learning_rate": 8.882205513784463e-06,
+      "loss": 0.5056782960891724,
+      "num_input_tokens_seen": 353550,
+      "step": 1115,
+      "train_runtime": 2973.9452,
+      "train_tokens_per_second": 118.882
+    },
+    {
+      "epoch": 0.05131034482758621,
+      "grad_norm": 8.346527099609375,
+      "learning_rate": 8.87218045112782e-06,
+      "loss": 0.46319133043289185,
+      "num_input_tokens_seen": 353906,
+      "step": 1116,
+      "train_runtime": 2975.9605,
+      "train_tokens_per_second": 118.922
+    },
+    {
+      "epoch": 0.05135632183908046,
+      "grad_norm": 8.855955123901367,
+      "learning_rate": 8.862155388471178e-06,
+      "loss": 0.5140814185142517,
+      "num_input_tokens_seen": 354200,
+      "step": 1117,
+      "train_runtime": 2977.9194,
+      "train_tokens_per_second": 118.942
+    },
+    {
+      "epoch": 0.051402298850574714,
+      "grad_norm": 8.370678901672363,
+      "learning_rate": 8.852130325814537e-06,
+      "loss": 0.4319264888763428,
+      "num_input_tokens_seen": 354460,
+      "step": 1118,
+      "train_runtime": 2979.8993,
+      "train_tokens_per_second": 118.95
+    },
+    {
+      "epoch": 0.051448275862068966,
+      "grad_norm": 7.829343795776367,
+      "learning_rate": 8.842105263157895e-06,
+      "loss": 0.4126385450363159,
+      "num_input_tokens_seen": 354706,
+      "step": 1119,
+      "train_runtime": 2981.8373,
+      "train_tokens_per_second": 118.956
+    },
+    {
+      "epoch": 0.05149425287356322,
+      "grad_norm": 7.9071269035339355,
+      "learning_rate": 8.832080200501254e-06,
+      "loss": 0.5065313577651978,
+      "num_input_tokens_seen": 355032,
+      "step": 1120,
+      "train_runtime": 2983.8669,
+      "train_tokens_per_second": 118.984
+    },
+    {
+      "epoch": 0.05154022988505747,
+      "grad_norm": 9.05221176147461,
+      "learning_rate": 8.822055137844612e-06,
+      "loss": 0.4188013970851898,
+      "num_input_tokens_seen": 355404,
+      "step": 1121,
+      "train_runtime": 2985.9206,
+      "train_tokens_per_second": 119.027
+    },
+    {
+      "epoch": 0.051586206896551724,
+      "grad_norm": 7.693556308746338,
+      "learning_rate": 8.81203007518797e-06,
+      "loss": 0.4153413474559784,
+      "num_input_tokens_seen": 355678,
+      "step": 1122,
+      "train_runtime": 2987.9234,
+      "train_tokens_per_second": 119.039
+    },
+    {
+      "epoch": 0.05163218390804598,
+      "grad_norm": 6.513828754425049,
+      "learning_rate": 8.802005012531329e-06,
+      "loss": 0.5046350955963135,
+      "num_input_tokens_seen": 356220,
+      "step": 1123,
+      "train_runtime": 2989.9668,
+      "train_tokens_per_second": 119.138
+    },
+    {
+      "epoch": 0.05167816091954023,
+      "grad_norm": 8.505133628845215,
+      "learning_rate": 8.791979949874688e-06,
+      "loss": 0.5596924424171448,
+      "num_input_tokens_seen": 356826,
+      "step": 1124,
+      "train_runtime": 2992.0902,
+      "train_tokens_per_second": 119.256
+    },
+    {
+      "epoch": 0.05172413793103448,
+      "grad_norm": 9.080416679382324,
+      "learning_rate": 8.781954887218046e-06,
+      "loss": 0.4769168794155121,
+      "num_input_tokens_seen": 357114,
+      "step": 1125,
+      "train_runtime": 2994.0474,
+      "train_tokens_per_second": 119.275
+    },
+    {
+      "epoch": 0.051770114942528735,
+      "grad_norm": 7.264612197875977,
+      "learning_rate": 8.771929824561405e-06,
+      "loss": 0.3889122009277344,
+      "num_input_tokens_seen": 357362,
+      "step": 1126,
+      "train_runtime": 2995.9843,
+      "train_tokens_per_second": 119.28
+    },
+    {
+      "epoch": 0.05181609195402299,
+      "grad_norm": 7.490901470184326,
+      "learning_rate": 8.761904761904763e-06,
+      "loss": 0.42194709181785583,
+      "num_input_tokens_seen": 357730,
+      "step": 1127,
+      "train_runtime": 2997.9807,
+      "train_tokens_per_second": 119.324
+    },
+    {
+      "epoch": 0.05186206896551724,
+      "grad_norm": 7.1191205978393555,
+      "learning_rate": 8.751879699248122e-06,
+      "loss": 0.37631088495254517,
+      "num_input_tokens_seen": 358026,
+      "step": 1128,
+      "train_runtime": 2999.9522,
+      "train_tokens_per_second": 119.344
+    },
+    {
+      "epoch": 0.05190804597701149,
+      "grad_norm": 7.080322265625,
+      "learning_rate": 8.741854636591478e-06,
+      "loss": 0.36222198605537415,
+      "num_input_tokens_seen": 358358,
+      "step": 1129,
+      "train_runtime": 3001.8958,
+      "train_tokens_per_second": 119.377
+    },
+    {
+      "epoch": 0.051954022988505745,
+      "grad_norm": 7.011504173278809,
+      "learning_rate": 8.731829573934839e-06,
+      "loss": 0.445604145526886,
+      "num_input_tokens_seen": 358802,
+      "step": 1130,
+      "train_runtime": 3003.8914,
+      "train_tokens_per_second": 119.446
+    },
+    {
+      "epoch": 0.052,
+      "grad_norm": 6.998114585876465,
+      "learning_rate": 8.721804511278195e-06,
+      "loss": 0.34580928087234497,
+      "num_input_tokens_seen": 359086,
+      "step": 1131,
+      "train_runtime": 3005.8366,
+      "train_tokens_per_second": 119.463
+    },
+    {
+      "epoch": 0.05204597701149425,
+      "grad_norm": 6.779391765594482,
+      "learning_rate": 8.711779448621554e-06,
+      "loss": 0.3365050256252289,
+      "num_input_tokens_seen": 359336,
+      "step": 1132,
+      "train_runtime": 3007.7775,
+      "train_tokens_per_second": 119.469
+    },
+    {
+      "epoch": 0.0520919540229885,
+      "grad_norm": 7.664047718048096,
+      "learning_rate": 8.701754385964914e-06,
+      "loss": 0.4866427779197693,
+      "num_input_tokens_seen": 359644,
+      "step": 1133,
+      "train_runtime": 3009.7348,
+      "train_tokens_per_second": 119.494
+    },
+    {
+      "epoch": 0.052137931034482755,
+      "grad_norm": 6.811913967132568,
+      "learning_rate": 8.69172932330827e-06,
+      "loss": 0.46658676862716675,
+      "num_input_tokens_seen": 359962,
+      "step": 1134,
+      "train_runtime": 3011.6919,
+      "train_tokens_per_second": 119.522
+    },
+    {
+      "epoch": 0.052183908045977015,
+      "grad_norm": 6.3152360916137695,
+      "learning_rate": 8.68170426065163e-06,
+      "loss": 0.4892335832118988,
+      "num_input_tokens_seen": 360424,
+      "step": 1135,
+      "train_runtime": 3013.6881,
+      "train_tokens_per_second": 119.596
+    },
+    {
+      "epoch": 0.05222988505747127,
+      "grad_norm": 7.800539016723633,
+      "learning_rate": 8.671679197994988e-06,
+      "loss": 0.426530659198761,
+      "num_input_tokens_seen": 360686,
+      "step": 1136,
+      "train_runtime": 3015.6362,
+      "train_tokens_per_second": 119.605
+    },
+    {
+      "epoch": 0.05227586206896552,
+      "grad_norm": 9.333327293395996,
+      "learning_rate": 8.661654135338346e-06,
+      "loss": 0.5790461301803589,
+      "num_input_tokens_seen": 361040,
+      "step": 1137,
+      "train_runtime": 3017.5823,
+      "train_tokens_per_second": 119.645
+    },
+    {
+      "epoch": 0.05232183908045977,
+      "grad_norm": 8.419378280639648,
+      "learning_rate": 8.651629072681705e-06,
+      "loss": 0.3921374976634979,
+      "num_input_tokens_seen": 361310,
+      "step": 1138,
+      "train_runtime": 3019.53,
+      "train_tokens_per_second": 119.658
+    },
+    {
+      "epoch": 0.052367816091954025,
+      "grad_norm": 7.641000747680664,
+      "learning_rate": 8.641604010025063e-06,
+      "loss": 0.32036256790161133,
+      "num_input_tokens_seen": 361546,
+      "step": 1139,
+      "train_runtime": 3021.4694,
+      "train_tokens_per_second": 119.659
+    },
+    {
+      "epoch": 0.05241379310344828,
+      "grad_norm": 8.194195747375488,
+      "learning_rate": 8.631578947368422e-06,
+      "loss": 0.3756909966468811,
+      "num_input_tokens_seen": 361798,
+      "step": 1140,
+      "train_runtime": 3023.4163,
+      "train_tokens_per_second": 119.665
+    },
+    {
+      "epoch": 0.05245977011494253,
+      "grad_norm": 9.875768661499023,
+      "learning_rate": 8.62155388471178e-06,
+      "loss": 0.4790947437286377,
+      "num_input_tokens_seen": 362022,
+      "step": 1141,
+      "train_runtime": 3038.7437,
+      "train_tokens_per_second": 119.135
+    },
+    {
+      "epoch": 0.05250574712643678,
+      "grad_norm": 8.130186080932617,
+      "learning_rate": 8.611528822055139e-06,
+      "loss": 0.3870607614517212,
+      "num_input_tokens_seen": 362326,
+      "step": 1142,
+      "train_runtime": 3040.6917,
+      "train_tokens_per_second": 119.159
+    },
+    {
+      "epoch": 0.052551724137931036,
+      "grad_norm": 6.7159013748168945,
+      "learning_rate": 8.601503759398497e-06,
+      "loss": 0.3064056634902954,
+      "num_input_tokens_seen": 362580,
+      "step": 1143,
+      "train_runtime": 3042.6334,
+      "train_tokens_per_second": 119.167
+    },
+    {
+      "epoch": 0.05259770114942529,
+      "grad_norm": 9.547625541687012,
+      "learning_rate": 8.591478696741856e-06,
+      "loss": 0.5110470652580261,
+      "num_input_tokens_seen": 362902,
+      "step": 1144,
+      "train_runtime": 3044.58,
+      "train_tokens_per_second": 119.196
+    },
+    {
+      "epoch": 0.05264367816091954,
+      "grad_norm": 7.9411468505859375,
+      "learning_rate": 8.581453634085214e-06,
+      "loss": 0.37716397643089294,
+      "num_input_tokens_seen": 363178,
+      "step": 1145,
+      "train_runtime": 3046.5253,
+      "train_tokens_per_second": 119.211
+    },
+    {
+      "epoch": 0.052689655172413793,
+      "grad_norm": 7.182112693786621,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 0.5067034363746643,
+      "num_input_tokens_seen": 363608,
+      "step": 1146,
+      "train_runtime": 3048.5144,
+      "train_tokens_per_second": 119.274
+    },
+    {
+      "epoch": 0.052735632183908046,
+      "grad_norm": 7.81766414642334,
+      "learning_rate": 8.561403508771931e-06,
+      "loss": 0.515543520450592,
+      "num_input_tokens_seen": 363984,
+      "step": 1147,
+      "train_runtime": 3050.4733,
+      "train_tokens_per_second": 119.321
+    },
+    {
+      "epoch": 0.0527816091954023,
+      "grad_norm": 6.620061874389648,
+      "learning_rate": 8.55137844611529e-06,
+      "loss": 0.3774106204509735,
+      "num_input_tokens_seen": 364260,
+      "step": 1148,
+      "train_runtime": 3052.436,
+      "train_tokens_per_second": 119.334
+    },
+    {
+      "epoch": 0.05282758620689655,
+      "grad_norm": 8.190850257873535,
+      "learning_rate": 8.541353383458646e-06,
+      "loss": 0.404315710067749,
+      "num_input_tokens_seen": 364486,
+      "step": 1149,
+      "train_runtime": 3054.3736,
+      "train_tokens_per_second": 119.332
+    },
+    {
+      "epoch": 0.052873563218390804,
+      "grad_norm": 6.6342010498046875,
+      "learning_rate": 8.531328320802007e-06,
+      "loss": 0.3467126786708832,
+      "num_input_tokens_seen": 364750,
+      "step": 1150,
+      "train_runtime": 3056.3456,
+      "train_tokens_per_second": 119.342
+    },
+    {
+      "epoch": 0.052919540229885056,
+      "grad_norm": 6.534978866577148,
+      "learning_rate": 8.521303258145363e-06,
+      "loss": 0.4126523733139038,
+      "num_input_tokens_seen": 365032,
+      "step": 1151,
+      "train_runtime": 3058.4116,
+      "train_tokens_per_second": 119.353
+    },
+    {
+      "epoch": 0.05296551724137931,
+      "grad_norm": 10.278910636901855,
+      "learning_rate": 8.511278195488722e-06,
+      "loss": 0.5738921761512756,
+      "num_input_tokens_seen": 365358,
+      "step": 1152,
+      "train_runtime": 3060.487,
+      "train_tokens_per_second": 119.379
+    },
+    {
+      "epoch": 0.05301149425287356,
+      "grad_norm": 7.18212366104126,
+      "learning_rate": 8.501253132832082e-06,
+      "loss": 0.4024466276168823,
+      "num_input_tokens_seen": 365600,
+      "step": 1153,
+      "train_runtime": 3062.4906,
+      "train_tokens_per_second": 119.38
+    },
+    {
+      "epoch": 0.053057471264367814,
+      "grad_norm": 8.520891189575195,
+      "learning_rate": 8.491228070175439e-06,
+      "loss": 0.6133989095687866,
+      "num_input_tokens_seen": 366018,
+      "step": 1154,
+      "train_runtime": 3064.5452,
+      "train_tokens_per_second": 119.436
+    },
+    {
+      "epoch": 0.05310344827586207,
+      "grad_norm": 7.127443313598633,
+      "learning_rate": 8.481203007518797e-06,
+      "loss": 0.36948350071907043,
+      "num_input_tokens_seen": 366268,
+      "step": 1155,
+      "train_runtime": 3066.6832,
+      "train_tokens_per_second": 119.435
+    },
+    {
+      "epoch": 0.05314942528735632,
+      "grad_norm": 6.519278049468994,
+      "learning_rate": 8.471177944862156e-06,
+      "loss": 0.33019205927848816,
+      "num_input_tokens_seen": 366556,
+      "step": 1156,
+      "train_runtime": 3068.7022,
+      "train_tokens_per_second": 119.45
+    },
+    {
+      "epoch": 0.05319540229885057,
+      "grad_norm": 7.162330150604248,
+      "learning_rate": 8.461152882205514e-06,
+      "loss": 0.485843300819397,
+      "num_input_tokens_seen": 367108,
+      "step": 1157,
+      "train_runtime": 3070.7748,
+      "train_tokens_per_second": 119.549
+    },
+    {
+      "epoch": 0.053241379310344825,
+      "grad_norm": 8.591348648071289,
+      "learning_rate": 8.451127819548873e-06,
+      "loss": 0.44043344259262085,
+      "num_input_tokens_seen": 367422,
+      "step": 1158,
+      "train_runtime": 3072.7318,
+      "train_tokens_per_second": 119.575
+    },
+    {
+      "epoch": 0.05328735632183908,
+      "grad_norm": 7.478002071380615,
+      "learning_rate": 8.441102756892231e-06,
+      "loss": 0.49568963050842285,
+      "num_input_tokens_seen": 367746,
+      "step": 1159,
+      "train_runtime": 3074.6878,
+      "train_tokens_per_second": 119.604
+    },
+    {
+      "epoch": 0.05333333333333334,
+      "grad_norm": 9.927587509155273,
+      "learning_rate": 8.43107769423559e-06,
+      "loss": 0.6322132349014282,
+      "num_input_tokens_seen": 368220,
+      "step": 1160,
+      "train_runtime": 3076.6903,
+      "train_tokens_per_second": 119.681
+    },
+    {
+      "epoch": 0.05337931034482759,
+      "grad_norm": 8.052186012268066,
+      "learning_rate": 8.421052631578948e-06,
+      "loss": 0.48395630717277527,
+      "num_input_tokens_seen": 368488,
+      "step": 1161,
+      "train_runtime": 3078.6659,
+      "train_tokens_per_second": 119.691
+    },
+    {
+      "epoch": 0.05342528735632184,
+      "grad_norm": 8.77470874786377,
+      "learning_rate": 8.411027568922307e-06,
+      "loss": 0.43330156803131104,
+      "num_input_tokens_seen": 368712,
+      "step": 1162,
+      "train_runtime": 3080.7208,
+      "train_tokens_per_second": 119.684
+    },
+    {
+      "epoch": 0.053471264367816095,
+      "grad_norm": 8.309282302856445,
+      "learning_rate": 8.401002506265665e-06,
+      "loss": 0.4578382074832916,
+      "num_input_tokens_seen": 369070,
+      "step": 1163,
+      "train_runtime": 3082.8172,
+      "train_tokens_per_second": 119.718
+    },
+    {
+      "epoch": 0.05351724137931035,
+      "grad_norm": 7.133829593658447,
+      "learning_rate": 8.390977443609024e-06,
+      "loss": 0.39338231086730957,
+      "num_input_tokens_seen": 369306,
+      "step": 1164,
+      "train_runtime": 3084.7634,
+      "train_tokens_per_second": 119.719
+    },
+    {
+      "epoch": 0.0535632183908046,
+      "grad_norm": 7.313564300537109,
+      "learning_rate": 8.380952380952382e-06,
+      "loss": 0.3970176577568054,
+      "num_input_tokens_seen": 369636,
+      "step": 1165,
+      "train_runtime": 3086.7139,
+      "train_tokens_per_second": 119.751
+    },
+    {
+      "epoch": 0.05360919540229885,
+      "grad_norm": 8.829938888549805,
+      "learning_rate": 8.370927318295739e-06,
+      "loss": 0.7930089831352234,
+      "num_input_tokens_seen": 370028,
+      "step": 1166,
+      "train_runtime": 3088.7001,
+      "train_tokens_per_second": 119.801
+    },
+    {
+      "epoch": 0.053655172413793105,
+      "grad_norm": 7.2194061279296875,
+      "learning_rate": 8.3609022556391e-06,
+      "loss": 0.4449452757835388,
+      "num_input_tokens_seen": 370412,
+      "step": 1167,
+      "train_runtime": 3090.8028,
+      "train_tokens_per_second": 119.843
+    },
+    {
+      "epoch": 0.05370114942528736,
+      "grad_norm": 8.99480152130127,
+      "learning_rate": 8.350877192982458e-06,
+      "loss": 0.32509714365005493,
+      "num_input_tokens_seen": 370656,
+      "step": 1168,
+      "train_runtime": 3092.8992,
+      "train_tokens_per_second": 119.841
+    },
+    {
+      "epoch": 0.05374712643678161,
+      "grad_norm": 7.847662448883057,
+      "learning_rate": 8.340852130325814e-06,
+      "loss": 0.5404742956161499,
+      "num_input_tokens_seen": 370992,
+      "step": 1169,
+      "train_runtime": 3094.8838,
+      "train_tokens_per_second": 119.873
+    },
+    {
+      "epoch": 0.05379310344827586,
+      "grad_norm": 7.609344005584717,
+      "learning_rate": 8.330827067669175e-06,
+      "loss": 0.48472529649734497,
+      "num_input_tokens_seen": 371248,
+      "step": 1170,
+      "train_runtime": 3096.8275,
+      "train_tokens_per_second": 119.88
+    },
+    {
+      "epoch": 0.053839080459770115,
+      "grad_norm": 7.672946453094482,
+      "learning_rate": 8.320802005012531e-06,
+      "loss": 0.5179246664047241,
+      "num_input_tokens_seen": 371696,
+      "step": 1171,
+      "train_runtime": 3110.6778,
+      "train_tokens_per_second": 119.49
+    },
+    {
+      "epoch": 0.05388505747126437,
+      "grad_norm": 7.6113128662109375,
+      "learning_rate": 8.31077694235589e-06,
+      "loss": 0.5013529062271118,
+      "num_input_tokens_seen": 371990,
+      "step": 1172,
+      "train_runtime": 3112.6332,
+      "train_tokens_per_second": 119.51
+    },
+    {
+      "epoch": 0.05393103448275862,
+      "grad_norm": 7.152410984039307,
+      "learning_rate": 8.300751879699248e-06,
+      "loss": 0.4111625552177429,
+      "num_input_tokens_seen": 372308,
+      "step": 1173,
+      "train_runtime": 3114.6851,
+      "train_tokens_per_second": 119.533
+    },
+    {
+      "epoch": 0.05397701149425287,
+      "grad_norm": 6.6777825355529785,
+      "learning_rate": 8.290726817042607e-06,
+      "loss": 0.3815338611602783,
+      "num_input_tokens_seen": 372556,
+      "step": 1174,
+      "train_runtime": 3116.7387,
+      "train_tokens_per_second": 119.534
+    },
+    {
+      "epoch": 0.054022988505747126,
+      "grad_norm": 6.369187831878662,
+      "learning_rate": 8.280701754385965e-06,
+      "loss": 0.3696746528148651,
+      "num_input_tokens_seen": 372790,
+      "step": 1175,
+      "train_runtime": 3118.7573,
+      "train_tokens_per_second": 119.532
+    },
+    {
+      "epoch": 0.05406896551724138,
+      "grad_norm": 8.142765998840332,
+      "learning_rate": 8.270676691729324e-06,
+      "loss": 0.5129543542861938,
+      "num_input_tokens_seen": 373122,
+      "step": 1176,
+      "train_runtime": 3120.7294,
+      "train_tokens_per_second": 119.562
+    },
+    {
+      "epoch": 0.05411494252873563,
+      "grad_norm": 7.096915245056152,
+      "learning_rate": 8.260651629072682e-06,
+      "loss": 0.419854074716568,
+      "num_input_tokens_seen": 373386,
+      "step": 1177,
+      "train_runtime": 3122.699,
+      "train_tokens_per_second": 119.572
+    },
+    {
+      "epoch": 0.054160919540229883,
+      "grad_norm": 7.837054252624512,
+      "learning_rate": 8.25062656641604e-06,
+      "loss": 0.4381014406681061,
+      "num_input_tokens_seen": 373642,
+      "step": 1178,
+      "train_runtime": 3124.6539,
+      "train_tokens_per_second": 119.579
+    },
+    {
+      "epoch": 0.054206896551724136,
+      "grad_norm": 8.150774002075195,
+      "learning_rate": 8.2406015037594e-06,
+      "loss": 0.4414655864238739,
+      "num_input_tokens_seen": 373980,
+      "step": 1179,
+      "train_runtime": 3126.6275,
+      "train_tokens_per_second": 119.611
+    },
+    {
+      "epoch": 0.05425287356321839,
+      "grad_norm": 7.266263484954834,
+      "learning_rate": 8.230576441102758e-06,
+      "loss": 0.3821980059146881,
+      "num_input_tokens_seen": 374272,
+      "step": 1180,
+      "train_runtime": 3128.5834,
+      "train_tokens_per_second": 119.63
+    },
+    {
+      "epoch": 0.05429885057471264,
+      "grad_norm": 9.496729850769043,
+      "learning_rate": 8.220551378446116e-06,
+      "loss": 0.52981036901474,
+      "num_input_tokens_seen": 374554,
+      "step": 1181,
+      "train_runtime": 3130.5408,
+      "train_tokens_per_second": 119.645
+    },
+    {
+      "epoch": 0.054344827586206894,
+      "grad_norm": 7.736052989959717,
+      "learning_rate": 8.210526315789475e-06,
+      "loss": 0.5340881943702698,
+      "num_input_tokens_seen": 374940,
+      "step": 1182,
+      "train_runtime": 3132.5336,
+      "train_tokens_per_second": 119.692
+    },
+    {
+      "epoch": 0.054390804597701146,
+      "grad_norm": 6.161796569824219,
+      "learning_rate": 8.200501253132833e-06,
+      "loss": 0.39382392168045044,
+      "num_input_tokens_seen": 375326,
+      "step": 1183,
+      "train_runtime": 3134.5148,
+      "train_tokens_per_second": 119.74
+    },
+    {
+      "epoch": 0.0544367816091954,
+      "grad_norm": 7.190888404846191,
+      "learning_rate": 8.190476190476192e-06,
+      "loss": 0.3610159456729889,
+      "num_input_tokens_seen": 375574,
+      "step": 1184,
+      "train_runtime": 3136.4564,
+      "train_tokens_per_second": 119.745
+    },
+    {
+      "epoch": 0.05448275862068966,
+      "grad_norm": 7.161884784698486,
+      "learning_rate": 8.18045112781955e-06,
+      "loss": 0.48947563767433167,
+      "num_input_tokens_seen": 375918,
+      "step": 1185,
+      "train_runtime": 3138.4192,
+      "train_tokens_per_second": 119.779
+    },
+    {
+      "epoch": 0.05452873563218391,
+      "grad_norm": 7.510371685028076,
+      "learning_rate": 8.170426065162907e-06,
+      "loss": 0.4124743342399597,
+      "num_input_tokens_seen": 376164,
+      "step": 1186,
+      "train_runtime": 3140.3846,
+      "train_tokens_per_second": 119.783
+    },
+    {
+      "epoch": 0.054574712643678164,
+      "grad_norm": 6.725326061248779,
+      "learning_rate": 8.160401002506266e-06,
+      "loss": 0.36423835158348083,
+      "num_input_tokens_seen": 376406,
+      "step": 1187,
+      "train_runtime": 3142.3379,
+      "train_tokens_per_second": 119.785
+    },
+    {
+      "epoch": 0.054620689655172416,
+      "grad_norm": 6.782083034515381,
+      "learning_rate": 8.150375939849626e-06,
+      "loss": 0.36468422412872314,
+      "num_input_tokens_seen": 376692,
+      "step": 1188,
+      "train_runtime": 3144.3041,
+      "train_tokens_per_second": 119.801
+    },
+    {
+      "epoch": 0.05466666666666667,
+      "grad_norm": 7.643582344055176,
+      "learning_rate": 8.140350877192983e-06,
+      "loss": 0.4475553631782532,
+      "num_input_tokens_seen": 377160,
+      "step": 1189,
+      "train_runtime": 3146.3364,
+      "train_tokens_per_second": 119.873
+    },
+    {
+      "epoch": 0.05471264367816092,
+      "grad_norm": 8.300542831420898,
+      "learning_rate": 8.130325814536341e-06,
+      "loss": 0.4346177577972412,
+      "num_input_tokens_seen": 377426,
+      "step": 1190,
+      "train_runtime": 3148.2823,
+      "train_tokens_per_second": 119.883
+    },
+    {
+      "epoch": 0.054758620689655174,
+      "grad_norm": 14.15939998626709,
+      "learning_rate": 8.1203007518797e-06,
+      "loss": 0.47130388021469116,
+      "num_input_tokens_seen": 377728,
+      "step": 1191,
+      "train_runtime": 3150.2485,
+      "train_tokens_per_second": 119.904
+    },
+    {
+      "epoch": 0.05480459770114943,
+      "grad_norm": 8.699316024780273,
+      "learning_rate": 8.110275689223058e-06,
+      "loss": 0.5361881256103516,
+      "num_input_tokens_seen": 378066,
+      "step": 1192,
+      "train_runtime": 3152.2371,
+      "train_tokens_per_second": 119.936
+    },
+    {
+      "epoch": 0.05485057471264368,
+      "grad_norm": 6.897034645080566,
+      "learning_rate": 8.100250626566416e-06,
+      "loss": 0.3854222595691681,
+      "num_input_tokens_seen": 378412,
+      "step": 1193,
+      "train_runtime": 3154.2237,
+      "train_tokens_per_second": 119.97
+    },
+    {
+      "epoch": 0.05489655172413793,
+      "grad_norm": 7.7607269287109375,
+      "learning_rate": 8.090225563909775e-06,
+      "loss": 0.3880585730075836,
+      "num_input_tokens_seen": 378658,
+      "step": 1194,
+      "train_runtime": 3156.2065,
+      "train_tokens_per_second": 119.973
+    },
+    {
+      "epoch": 0.054942528735632185,
+      "grad_norm": 6.789251327514648,
+      "learning_rate": 8.080200501253133e-06,
+      "loss": 0.3877938389778137,
+      "num_input_tokens_seen": 378950,
+      "step": 1195,
+      "train_runtime": 3158.2332,
+      "train_tokens_per_second": 119.988
+    },
+    {
+      "epoch": 0.05498850574712644,
+      "grad_norm": 7.403102397918701,
+      "learning_rate": 8.070175438596492e-06,
+      "loss": 0.507320761680603,
+      "num_input_tokens_seen": 379220,
+      "step": 1196,
+      "train_runtime": 3160.2494,
+      "train_tokens_per_second": 119.997
+    },
+    {
+      "epoch": 0.05503448275862069,
+      "grad_norm": 7.399575233459473,
+      "learning_rate": 8.06015037593985e-06,
+      "loss": 0.6930140852928162,
+      "num_input_tokens_seen": 379848,
+      "step": 1197,
+      "train_runtime": 3162.448,
+      "train_tokens_per_second": 120.112
+    },
+    {
+      "epoch": 0.05508045977011494,
+      "grad_norm": 6.599159240722656,
+      "learning_rate": 8.050125313283209e-06,
+      "loss": 0.46932291984558105,
+      "num_input_tokens_seen": 380290,
+      "step": 1198,
+      "train_runtime": 3164.518,
+      "train_tokens_per_second": 120.173
+    },
+    {
+      "epoch": 0.055126436781609195,
+      "grad_norm": 7.153855323791504,
+      "learning_rate": 8.040100250626567e-06,
+      "loss": 0.2693643569946289,
+      "num_input_tokens_seen": 380506,
+      "step": 1199,
+      "train_runtime": 3166.4775,
+      "train_tokens_per_second": 120.167
+    },
+    {
+      "epoch": 0.05517241379310345,
+      "grad_norm": 7.928609371185303,
+      "learning_rate": 8.030075187969926e-06,
+      "loss": 0.4285462498664856,
+      "num_input_tokens_seen": 380884,
+      "step": 1200,
+      "train_runtime": 3168.5553,
+      "train_tokens_per_second": 120.207
+    },
+    {
+      "epoch": 0.05517241379310345,
+      "eval_loss": 1.692144513130188,
+      "eval_runtime": 54.5988,
+      "eval_samples_per_second": 18.315,
+      "eval_steps_per_second": 9.158,
+      "num_input_tokens_seen": 380884,
+      "step": 1200
+    },
+    {
+      "epoch": 0.0552183908045977,
+      "grad_norm": 7.031571388244629,
+      "learning_rate": 8.020050125313283e-06,
+      "loss": 0.3508245646953583,
+      "num_input_tokens_seen": 381152,
+      "step": 1201,
+      "train_runtime": 3238.3257,
+      "train_tokens_per_second": 117.7
+    },
+    {
+      "epoch": 0.05526436781609195,
+      "grad_norm": 7.969499111175537,
+      "learning_rate": 8.010025062656643e-06,
+      "loss": 0.5498538017272949,
+      "num_input_tokens_seen": 381434,
+      "step": 1202,
+      "train_runtime": 3240.2813,
+      "train_tokens_per_second": 117.716
+    },
+    {
+      "epoch": 0.055310344827586205,
+      "grad_norm": 8.85057258605957,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.5166440606117249,
+      "num_input_tokens_seen": 381748,
+      "step": 1203,
+      "train_runtime": 3242.284,
+      "train_tokens_per_second": 117.74
+    },
+    {
+      "epoch": 0.05535632183908046,
+      "grad_norm": 6.515604019165039,
+      "learning_rate": 7.989974937343358e-06,
+      "loss": 0.38619551062583923,
+      "num_input_tokens_seen": 382004,
+      "step": 1204,
+      "train_runtime": 3244.2439,
+      "train_tokens_per_second": 117.748
+    },
+    {
+      "epoch": 0.05540229885057471,
+      "grad_norm": 8.197484970092773,
+      "learning_rate": 7.979949874686718e-06,
+      "loss": 0.5534597039222717,
+      "num_input_tokens_seen": 382364,
+      "step": 1205,
+      "train_runtime": 3246.2164,
+      "train_tokens_per_second": 117.788
+    },
+    {
+      "epoch": 0.05544827586206896,
+      "grad_norm": 6.80437707901001,
+      "learning_rate": 7.969924812030075e-06,
+      "loss": 0.5063154697418213,
+      "num_input_tokens_seen": 382746,
+      "step": 1206,
+      "train_runtime": 3248.2386,
+      "train_tokens_per_second": 117.832
+    },
+    {
+      "epoch": 0.055494252873563216,
+      "grad_norm": 7.77208137512207,
+      "learning_rate": 7.959899749373434e-06,
+      "loss": 0.4124975800514221,
+      "num_input_tokens_seen": 383006,
+      "step": 1207,
+      "train_runtime": 3250.3198,
+      "train_tokens_per_second": 117.836
+    },
+    {
+      "epoch": 0.05554022988505747,
+      "grad_norm": 5.83624267578125,
+      "learning_rate": 7.949874686716794e-06,
+      "loss": 0.2796672582626343,
+      "num_input_tokens_seen": 383234,
+      "step": 1208,
+      "train_runtime": 3252.3877,
+      "train_tokens_per_second": 117.832
+    },
+    {
+      "epoch": 0.05558620689655172,
+      "grad_norm": 7.766402244567871,
+      "learning_rate": 7.93984962406015e-06,
+      "loss": 0.46150755882263184,
+      "num_input_tokens_seen": 383550,
+      "step": 1209,
+      "train_runtime": 3254.4166,
+      "train_tokens_per_second": 117.855
+    },
+    {
+      "epoch": 0.05563218390804598,
+      "grad_norm": 6.165184497833252,
+      "learning_rate": 7.929824561403509e-06,
+      "loss": 0.3527635335922241,
+      "num_input_tokens_seen": 383886,
+      "step": 1210,
+      "train_runtime": 3256.5657,
+      "train_tokens_per_second": 117.881
+    },
+    {
+      "epoch": 0.05567816091954023,
+      "grad_norm": 6.623737335205078,
+      "learning_rate": 7.919799498746868e-06,
+      "loss": 0.30056440830230713,
+      "num_input_tokens_seen": 384140,
+      "step": 1211,
+      "train_runtime": 3258.5339,
+      "train_tokens_per_second": 117.887
+    },
+    {
+      "epoch": 0.055724137931034486,
+      "grad_norm": 7.612054347991943,
+      "learning_rate": 7.909774436090226e-06,
+      "loss": 0.44327080249786377,
+      "num_input_tokens_seen": 384442,
+      "step": 1212,
+      "train_runtime": 3260.504,
+      "train_tokens_per_second": 117.909
+    },
+    {
+      "epoch": 0.05577011494252874,
+      "grad_norm": 8.566631317138672,
+      "learning_rate": 7.899749373433585e-06,
+      "loss": 0.49499887228012085,
+      "num_input_tokens_seen": 384758,
+      "step": 1213,
+      "train_runtime": 3262.5097,
+      "train_tokens_per_second": 117.933
+    },
+    {
+      "epoch": 0.05581609195402299,
+      "grad_norm": 9.735397338867188,
+      "learning_rate": 7.889724310776943e-06,
+      "loss": 0.47953516244888306,
+      "num_input_tokens_seen": 385036,
+      "step": 1214,
+      "train_runtime": 3264.6896,
+      "train_tokens_per_second": 117.94
+    },
+    {
+      "epoch": 0.05586206896551724,
+      "grad_norm": 13.594995498657227,
+      "learning_rate": 7.879699248120301e-06,
+      "loss": 0.36916059255599976,
+      "num_input_tokens_seen": 385294,
+      "step": 1215,
+      "train_runtime": 3266.7717,
+      "train_tokens_per_second": 117.943
+    },
+    {
+      "epoch": 0.055908045977011496,
+      "grad_norm": 7.692044258117676,
+      "learning_rate": 7.86967418546366e-06,
+      "loss": 0.49710947275161743,
+      "num_input_tokens_seen": 385606,
+      "step": 1216,
+      "train_runtime": 3268.7645,
+      "train_tokens_per_second": 117.967
+    },
+    {
+      "epoch": 0.05595402298850575,
+      "grad_norm": 7.8550615310668945,
+      "learning_rate": 7.859649122807018e-06,
+      "loss": 0.420956015586853,
+      "num_input_tokens_seen": 385916,
+      "step": 1217,
+      "train_runtime": 3270.8602,
+      "train_tokens_per_second": 117.986
+    },
+    {
+      "epoch": 0.056,
+      "grad_norm": 6.607285499572754,
+      "learning_rate": 7.849624060150377e-06,
+      "loss": 0.35990452766418457,
+      "num_input_tokens_seen": 386266,
+      "step": 1218,
+      "train_runtime": 3272.8397,
+      "train_tokens_per_second": 118.022
+    },
+    {
+      "epoch": 0.056045977011494254,
+      "grad_norm": 8.219738960266113,
+      "learning_rate": 7.839598997493735e-06,
+      "loss": 0.4552280306816101,
+      "num_input_tokens_seen": 386618,
+      "step": 1219,
+      "train_runtime": 3274.8135,
+      "train_tokens_per_second": 118.058
+    },
+    {
+      "epoch": 0.056091954022988506,
+      "grad_norm": 9.075862884521484,
+      "learning_rate": 7.829573934837094e-06,
+      "loss": 0.38085708022117615,
+      "num_input_tokens_seen": 386878,
+      "step": 1220,
+      "train_runtime": 3276.779,
+      "train_tokens_per_second": 118.067
+    },
+    {
+      "epoch": 0.05613793103448276,
+      "grad_norm": 7.036026954650879,
+      "learning_rate": 7.81954887218045e-06,
+      "loss": 0.44221633672714233,
+      "num_input_tokens_seen": 387248,
+      "step": 1221,
+      "train_runtime": 3278.7603,
+      "train_tokens_per_second": 118.108
+    },
+    {
+      "epoch": 0.05618390804597701,
+      "grad_norm": 7.284511089324951,
+      "learning_rate": 7.809523809523811e-06,
+      "loss": 0.47441187500953674,
+      "num_input_tokens_seen": 387486,
+      "step": 1222,
+      "train_runtime": 3280.7017,
+      "train_tokens_per_second": 118.111
+    },
+    {
+      "epoch": 0.056229885057471264,
+      "grad_norm": 7.1411824226379395,
+      "learning_rate": 7.79949874686717e-06,
+      "loss": 0.3490273356437683,
+      "num_input_tokens_seen": 387746,
+      "step": 1223,
+      "train_runtime": 3282.6805,
+      "train_tokens_per_second": 118.119
+    },
+    {
+      "epoch": 0.05627586206896552,
+      "grad_norm": 8.54399585723877,
+      "learning_rate": 7.789473684210526e-06,
+      "loss": 0.4305766522884369,
+      "num_input_tokens_seen": 387976,
+      "step": 1224,
+      "train_runtime": 3284.6258,
+      "train_tokens_per_second": 118.119
+    },
+    {
+      "epoch": 0.05632183908045977,
+      "grad_norm": 7.717691898345947,
+      "learning_rate": 7.779448621553886e-06,
+      "loss": 0.45238345861434937,
+      "num_input_tokens_seen": 388232,
+      "step": 1225,
+      "train_runtime": 3286.5665,
+      "train_tokens_per_second": 118.127
+    },
+    {
+      "epoch": 0.05636781609195402,
+      "grad_norm": 7.620163917541504,
+      "learning_rate": 7.769423558897243e-06,
+      "loss": 0.5245876908302307,
+      "num_input_tokens_seen": 388628,
+      "step": 1226,
+      "train_runtime": 3288.5618,
+      "train_tokens_per_second": 118.176
+    },
+    {
+      "epoch": 0.056413793103448275,
+      "grad_norm": 7.724833965301514,
+      "learning_rate": 7.759398496240602e-06,
+      "loss": 0.3621765673160553,
+      "num_input_tokens_seen": 388890,
+      "step": 1227,
+      "train_runtime": 3290.5009,
+      "train_tokens_per_second": 118.186
+    },
+    {
+      "epoch": 0.05645977011494253,
+      "grad_norm": 7.4883222579956055,
+      "learning_rate": 7.749373433583962e-06,
+      "loss": 0.5157286524772644,
+      "num_input_tokens_seen": 389214,
+      "step": 1228,
+      "train_runtime": 3292.6802,
+      "train_tokens_per_second": 118.206
+    },
+    {
+      "epoch": 0.05650574712643678,
+      "grad_norm": 7.155076026916504,
+      "learning_rate": 7.739348370927319e-06,
+      "loss": 0.4434982240200043,
+      "num_input_tokens_seen": 389698,
+      "step": 1229,
+      "train_runtime": 3294.7445,
+      "train_tokens_per_second": 118.279
+    },
+    {
+      "epoch": 0.05655172413793103,
+      "grad_norm": 8.831174850463867,
+      "learning_rate": 7.729323308270677e-06,
+      "loss": 0.39858779311180115,
+      "num_input_tokens_seen": 389950,
+      "step": 1230,
+      "train_runtime": 3296.6909,
+      "train_tokens_per_second": 118.285
+    },
+    {
+      "epoch": 0.056597701149425285,
+      "grad_norm": 9.107735633850098,
+      "learning_rate": 7.719298245614036e-06,
+      "loss": 0.4810582399368286,
+      "num_input_tokens_seen": 390264,
+      "step": 1231,
+      "train_runtime": 3311.3898,
+      "train_tokens_per_second": 117.855
+    },
+    {
+      "epoch": 0.05664367816091954,
+      "grad_norm": 7.23051643371582,
+      "learning_rate": 7.709273182957394e-06,
+      "loss": 0.3264503479003906,
+      "num_input_tokens_seen": 390496,
+      "step": 1232,
+      "train_runtime": 3313.3228,
+      "train_tokens_per_second": 117.856
+    },
+    {
+      "epoch": 0.05668965517241379,
+      "grad_norm": 7.617827892303467,
+      "learning_rate": 7.699248120300753e-06,
+      "loss": 0.3266094923019409,
+      "num_input_tokens_seen": 390736,
+      "step": 1233,
+      "train_runtime": 3315.2535,
+      "train_tokens_per_second": 117.86
+    },
+    {
+      "epoch": 0.05673563218390804,
+      "grad_norm": 6.893436431884766,
+      "learning_rate": 7.689223057644111e-06,
+      "loss": 0.3524739742279053,
+      "num_input_tokens_seen": 391048,
+      "step": 1234,
+      "train_runtime": 3317.2784,
+      "train_tokens_per_second": 117.882
+    },
+    {
+      "epoch": 0.0567816091954023,
+      "grad_norm": 8.767608642578125,
+      "learning_rate": 7.67919799498747e-06,
+      "loss": 0.6118502616882324,
+      "num_input_tokens_seen": 391298,
+      "step": 1235,
+      "train_runtime": 3319.352,
+      "train_tokens_per_second": 117.884
+    },
+    {
+      "epoch": 0.056827586206896555,
+      "grad_norm": 6.283548355102539,
+      "learning_rate": 7.669172932330828e-06,
+      "loss": 0.2690774202346802,
+      "num_input_tokens_seen": 391556,
+      "step": 1236,
+      "train_runtime": 3321.4105,
+      "train_tokens_per_second": 117.888
+    },
+    {
+      "epoch": 0.05687356321839081,
+      "grad_norm": 6.369739532470703,
+      "learning_rate": 7.659147869674187e-06,
+      "loss": 0.4399859309196472,
+      "num_input_tokens_seen": 391930,
+      "step": 1237,
+      "train_runtime": 3323.3931,
+      "train_tokens_per_second": 117.931
+    },
+    {
+      "epoch": 0.05691954022988506,
+      "grad_norm": 8.632977485656738,
+      "learning_rate": 7.649122807017545e-06,
+      "loss": 0.4601448178291321,
+      "num_input_tokens_seen": 392238,
+      "step": 1238,
+      "train_runtime": 3325.3327,
+      "train_tokens_per_second": 117.955
+    },
+    {
+      "epoch": 0.05696551724137931,
+      "grad_norm": 7.602078914642334,
+      "learning_rate": 7.639097744360904e-06,
+      "loss": 0.3859010934829712,
+      "num_input_tokens_seen": 392522,
+      "step": 1239,
+      "train_runtime": 3327.2671,
+      "train_tokens_per_second": 117.971
+    },
+    {
+      "epoch": 0.057011494252873565,
+      "grad_norm": 7.507519245147705,
+      "learning_rate": 7.629072681704261e-06,
+      "loss": 0.3987613916397095,
+      "num_input_tokens_seen": 392774,
+      "step": 1240,
+      "train_runtime": 3329.2005,
+      "train_tokens_per_second": 117.978
+    },
+    {
+      "epoch": 0.05705747126436782,
+      "grad_norm": 6.93795919418335,
+      "learning_rate": 7.61904761904762e-06,
+      "loss": 0.32270294427871704,
+      "num_input_tokens_seen": 393050,
+      "step": 1241,
+      "train_runtime": 3331.1414,
+      "train_tokens_per_second": 117.993
+    },
+    {
+      "epoch": 0.05710344827586207,
+      "grad_norm": 8.514482498168945,
+      "learning_rate": 7.609022556390978e-06,
+      "loss": 0.40500426292419434,
+      "num_input_tokens_seen": 393290,
+      "step": 1242,
+      "train_runtime": 3333.0875,
+      "train_tokens_per_second": 117.996
+    },
+    {
+      "epoch": 0.05714942528735632,
+      "grad_norm": 6.902530193328857,
+      "learning_rate": 7.598997493734337e-06,
+      "loss": 0.3956249952316284,
+      "num_input_tokens_seen": 393654,
+      "step": 1243,
+      "train_runtime": 3335.0571,
+      "train_tokens_per_second": 118.035
+    },
+    {
+      "epoch": 0.057195402298850576,
+      "grad_norm": 8.342933654785156,
+      "learning_rate": 7.588972431077694e-06,
+      "loss": 0.49615880846977234,
+      "num_input_tokens_seen": 393926,
+      "step": 1244,
+      "train_runtime": 3336.9995,
+      "train_tokens_per_second": 118.048
+    },
+    {
+      "epoch": 0.05724137931034483,
+      "grad_norm": 8.174407005310059,
+      "learning_rate": 7.578947368421054e-06,
+      "loss": 0.3771926760673523,
+      "num_input_tokens_seen": 394208,
+      "step": 1245,
+      "train_runtime": 3338.9395,
+      "train_tokens_per_second": 118.064
+    },
+    {
+      "epoch": 0.05728735632183908,
+      "grad_norm": 8.724871635437012,
+      "learning_rate": 7.568922305764411e-06,
+      "loss": 0.42995113134384155,
+      "num_input_tokens_seen": 394508,
+      "step": 1246,
+      "train_runtime": 3340.8876,
+      "train_tokens_per_second": 118.085
+    },
+    {
+      "epoch": 0.05733333333333333,
+      "grad_norm": 9.025880813598633,
+      "learning_rate": 7.55889724310777e-06,
+      "loss": 0.4564989507198334,
+      "num_input_tokens_seen": 394840,
+      "step": 1247,
+      "train_runtime": 3342.8568,
+      "train_tokens_per_second": 118.115
+    },
+    {
+      "epoch": 0.057379310344827586,
+      "grad_norm": 17.948413848876953,
+      "learning_rate": 7.548872180451129e-06,
+      "loss": 0.3961571455001831,
+      "num_input_tokens_seen": 395110,
+      "step": 1248,
+      "train_runtime": 3344.8059,
+      "train_tokens_per_second": 118.126
+    },
+    {
+      "epoch": 0.05742528735632184,
+      "grad_norm": 7.265836715698242,
+      "learning_rate": 7.538847117794487e-06,
+      "loss": 0.39476943016052246,
+      "num_input_tokens_seen": 395366,
+      "step": 1249,
+      "train_runtime": 3346.7549,
+      "train_tokens_per_second": 118.134
+    },
+    {
+      "epoch": 0.05747126436781609,
+      "grad_norm": 7.201603889465332,
+      "learning_rate": 7.528822055137845e-06,
+      "loss": 0.3717278838157654,
+      "num_input_tokens_seen": 395590,
+      "step": 1250,
+      "train_runtime": 3348.6935,
+      "train_tokens_per_second": 118.133
+    },
+    {
+      "epoch": 0.057517241379310344,
+      "grad_norm": 7.79808235168457,
+      "learning_rate": 7.518796992481203e-06,
+      "loss": 0.47616317868232727,
+      "num_input_tokens_seen": 395896,
+      "step": 1251,
+      "train_runtime": 3350.7426,
+      "train_tokens_per_second": 118.152
+    },
+    {
+      "epoch": 0.057563218390804596,
+      "grad_norm": 10.870890617370605,
+      "learning_rate": 7.508771929824562e-06,
+      "loss": 0.5085861682891846,
+      "num_input_tokens_seen": 396272,
+      "step": 1252,
+      "train_runtime": 3352.7317,
+      "train_tokens_per_second": 118.194
+    },
+    {
+      "epoch": 0.05760919540229885,
+      "grad_norm": 9.563769340515137,
+      "learning_rate": 7.498746867167921e-06,
+      "loss": 0.47975170612335205,
+      "num_input_tokens_seen": 396506,
+      "step": 1253,
+      "train_runtime": 3354.6775,
+      "train_tokens_per_second": 118.195
+    },
+    {
+      "epoch": 0.0576551724137931,
+      "grad_norm": 9.062041282653809,
+      "learning_rate": 7.488721804511278e-06,
+      "loss": 0.6291418671607971,
+      "num_input_tokens_seen": 396770,
+      "step": 1254,
+      "train_runtime": 3356.6407,
+      "train_tokens_per_second": 118.204
+    },
+    {
+      "epoch": 0.057701149425287354,
+      "grad_norm": 8.90355396270752,
+      "learning_rate": 7.478696741854638e-06,
+      "loss": 0.42996400594711304,
+      "num_input_tokens_seen": 397032,
+      "step": 1255,
+      "train_runtime": 3358.5904,
+      "train_tokens_per_second": 118.214
+    },
+    {
+      "epoch": 0.05774712643678161,
+      "grad_norm": 8.653199195861816,
+      "learning_rate": 7.468671679197995e-06,
+      "loss": 0.4824579954147339,
+      "num_input_tokens_seen": 397330,
+      "step": 1256,
+      "train_runtime": 3360.5679,
+      "train_tokens_per_second": 118.233
+    },
+    {
+      "epoch": 0.05779310344827586,
+      "grad_norm": 10.245847702026367,
+      "learning_rate": 7.458646616541354e-06,
+      "loss": 0.4312460422515869,
+      "num_input_tokens_seen": 397664,
+      "step": 1257,
+      "train_runtime": 3362.5724,
+      "train_tokens_per_second": 118.262
+    },
+    {
+      "epoch": 0.05783908045977011,
+      "grad_norm": 8.25611400604248,
+      "learning_rate": 7.448621553884713e-06,
+      "loss": 0.4436429738998413,
+      "num_input_tokens_seen": 397976,
+      "step": 1258,
+      "train_runtime": 3364.5213,
+      "train_tokens_per_second": 118.286
+    },
+    {
+      "epoch": 0.057885057471264365,
+      "grad_norm": 6.373208045959473,
+      "learning_rate": 7.438596491228071e-06,
+      "loss": 0.32305020093917847,
+      "num_input_tokens_seen": 398216,
+      "step": 1259,
+      "train_runtime": 3366.458,
+      "train_tokens_per_second": 118.289
+    },
+    {
+      "epoch": 0.057931034482758624,
+      "grad_norm": 8.443294525146484,
+      "learning_rate": 7.428571428571429e-06,
+      "loss": 0.38860228657722473,
+      "num_input_tokens_seen": 398534,
+      "step": 1260,
+      "train_runtime": 3368.456,
+      "train_tokens_per_second": 118.314
+    },
+    {
+      "epoch": 0.05797701149425288,
+      "grad_norm": 6.778960227966309,
+      "learning_rate": 7.418546365914787e-06,
+      "loss": 0.33598434925079346,
+      "num_input_tokens_seen": 398812,
+      "step": 1261,
+      "train_runtime": 3384.1842,
+      "train_tokens_per_second": 117.846
+    },
+    {
+      "epoch": 0.05802298850574713,
+      "grad_norm": 8.121769905090332,
+      "learning_rate": 7.408521303258146e-06,
+      "loss": 0.3616170883178711,
+      "num_input_tokens_seen": 399044,
+      "step": 1262,
+      "train_runtime": 3386.2081,
+      "train_tokens_per_second": 117.844
+    },
+    {
+      "epoch": 0.05806896551724138,
+      "grad_norm": 6.803779602050781,
+      "learning_rate": 7.398496240601505e-06,
+      "loss": 0.3750128149986267,
+      "num_input_tokens_seen": 399300,
+      "step": 1263,
+      "train_runtime": 3388.2191,
+      "train_tokens_per_second": 117.85
+    },
+    {
+      "epoch": 0.058114942528735634,
+      "grad_norm": 7.254236698150635,
+      "learning_rate": 7.388471177944862e-06,
+      "loss": 0.380276620388031,
+      "num_input_tokens_seen": 399558,
+      "step": 1264,
+      "train_runtime": 3390.2946,
+      "train_tokens_per_second": 117.853
+    },
+    {
+      "epoch": 0.05816091954022989,
+      "grad_norm": 7.822441101074219,
+      "learning_rate": 7.378446115288222e-06,
+      "loss": 0.5770272016525269,
+      "num_input_tokens_seen": 399954,
+      "step": 1265,
+      "train_runtime": 3392.2616,
+      "train_tokens_per_second": 117.902
+    },
+    {
+      "epoch": 0.05820689655172414,
+      "grad_norm": 8.614603042602539,
+      "learning_rate": 7.368421052631579e-06,
+      "loss": 0.3941013216972351,
+      "num_input_tokens_seen": 400194,
+      "step": 1266,
+      "train_runtime": 3394.2646,
+      "train_tokens_per_second": 117.903
+    },
+    {
+      "epoch": 0.05825287356321839,
+      "grad_norm": 9.3148775100708,
+      "learning_rate": 7.358395989974938e-06,
+      "loss": 0.5557112693786621,
+      "num_input_tokens_seen": 400504,
+      "step": 1267,
+      "train_runtime": 3396.2218,
+      "train_tokens_per_second": 117.926
+    },
+    {
+      "epoch": 0.058298850574712645,
+      "grad_norm": 7.287705898284912,
+      "learning_rate": 7.348370927318297e-06,
+      "loss": 0.37800031900405884,
+      "num_input_tokens_seen": 400802,
+      "step": 1268,
+      "train_runtime": 3398.2005,
+      "train_tokens_per_second": 117.945
+    },
+    {
+      "epoch": 0.0583448275862069,
+      "grad_norm": 7.846877574920654,
+      "learning_rate": 7.338345864661655e-06,
+      "loss": 0.46503913402557373,
+      "num_input_tokens_seen": 401122,
+      "step": 1269,
+      "train_runtime": 3400.1976,
+      "train_tokens_per_second": 117.97
+    },
+    {
+      "epoch": 0.05839080459770115,
+      "grad_norm": 6.571041584014893,
+      "learning_rate": 7.328320802005013e-06,
+      "loss": 0.27229586243629456,
+      "num_input_tokens_seen": 401384,
+      "step": 1270,
+      "train_runtime": 3402.1572,
+      "train_tokens_per_second": 117.979
+    },
+    {
+      "epoch": 0.0584367816091954,
+      "grad_norm": 8.546449661254883,
+      "learning_rate": 7.318295739348371e-06,
+      "loss": 0.4096136689186096,
+      "num_input_tokens_seen": 401656,
+      "step": 1271,
+      "train_runtime": 3404.1099,
+      "train_tokens_per_second": 117.991
+    },
+    {
+      "epoch": 0.058482758620689655,
+      "grad_norm": 8.082525253295898,
+      "learning_rate": 7.30827067669173e-06,
+      "loss": 0.4620562195777893,
+      "num_input_tokens_seen": 401990,
+      "step": 1272,
+      "train_runtime": 3406.104,
+      "train_tokens_per_second": 118.02
+    },
+    {
+      "epoch": 0.05852873563218391,
+      "grad_norm": 7.5269999504089355,
+      "learning_rate": 7.298245614035089e-06,
+      "loss": 0.37954992055892944,
+      "num_input_tokens_seen": 402232,
+      "step": 1273,
+      "train_runtime": 3408.0767,
+      "train_tokens_per_second": 118.023
+    },
+    {
+      "epoch": 0.05857471264367816,
+      "grad_norm": 8.238028526306152,
+      "learning_rate": 7.288220551378446e-06,
+      "loss": 0.4700665771961212,
+      "num_input_tokens_seen": 402592,
+      "step": 1274,
+      "train_runtime": 3410.0529,
+      "train_tokens_per_second": 118.06
+    },
+    {
+      "epoch": 0.05862068965517241,
+      "grad_norm": 7.085027694702148,
+      "learning_rate": 7.278195488721806e-06,
+      "loss": 0.2926819324493408,
+      "num_input_tokens_seen": 402808,
+      "step": 1275,
+      "train_runtime": 3412.0022,
+      "train_tokens_per_second": 118.056
+    },
+    {
+      "epoch": 0.058666666666666666,
+      "grad_norm": 7.41765832901001,
+      "learning_rate": 7.268170426065163e-06,
+      "loss": 0.43641120195388794,
+      "num_input_tokens_seen": 403322,
+      "step": 1276,
+      "train_runtime": 3414.0155,
+      "train_tokens_per_second": 118.137
+    },
+    {
+      "epoch": 0.05871264367816092,
+      "grad_norm": 7.371816635131836,
+      "learning_rate": 7.258145363408522e-06,
+      "loss": 0.34357935190200806,
+      "num_input_tokens_seen": 403548,
+      "step": 1277,
+      "train_runtime": 3415.9485,
+      "train_tokens_per_second": 118.136
+    },
+    {
+      "epoch": 0.05875862068965517,
+      "grad_norm": 6.794807434082031,
+      "learning_rate": 7.248120300751881e-06,
+      "loss": 0.39415302872657776,
+      "num_input_tokens_seen": 403866,
+      "step": 1278,
+      "train_runtime": 3417.8982,
+      "train_tokens_per_second": 118.162
+    },
+    {
+      "epoch": 0.05880459770114942,
+      "grad_norm": 14.13974380493164,
+      "learning_rate": 7.238095238095239e-06,
+      "loss": 0.5558474063873291,
+      "num_input_tokens_seen": 404204,
+      "step": 1279,
+      "train_runtime": 3419.8535,
+      "train_tokens_per_second": 118.193
+    },
+    {
+      "epoch": 0.058850574712643676,
+      "grad_norm": 9.345540046691895,
+      "learning_rate": 7.228070175438597e-06,
+      "loss": 0.4721265435218811,
+      "num_input_tokens_seen": 404492,
+      "step": 1280,
+      "train_runtime": 3421.806,
+      "train_tokens_per_second": 118.21
+    },
+    {
+      "epoch": 0.05889655172413793,
+      "grad_norm": 7.707251071929932,
+      "learning_rate": 7.218045112781955e-06,
+      "loss": 0.4041139781475067,
+      "num_input_tokens_seen": 404760,
+      "step": 1281,
+      "train_runtime": 3423.7916,
+      "train_tokens_per_second": 118.22
+    },
+    {
+      "epoch": 0.05894252873563218,
+      "grad_norm": 7.0474395751953125,
+      "learning_rate": 7.208020050125314e-06,
+      "loss": 0.44104093313217163,
+      "num_input_tokens_seen": 405044,
+      "step": 1282,
+      "train_runtime": 3425.7389,
+      "train_tokens_per_second": 118.236
+    },
+    {
+      "epoch": 0.058988505747126434,
+      "grad_norm": 8.901880264282227,
+      "learning_rate": 7.197994987468673e-06,
+      "loss": 0.526037871837616,
+      "num_input_tokens_seen": 405300,
+      "step": 1283,
+      "train_runtime": 3427.6805,
+      "train_tokens_per_second": 118.243
+    },
+    {
+      "epoch": 0.059034482758620686,
+      "grad_norm": 7.946100234985352,
+      "learning_rate": 7.18796992481203e-06,
+      "loss": 0.47331130504608154,
+      "num_input_tokens_seen": 405610,
+      "step": 1284,
+      "train_runtime": 3429.6572,
+      "train_tokens_per_second": 118.265
+    },
+    {
+      "epoch": 0.059080459770114946,
+      "grad_norm": 7.580421447753906,
+      "learning_rate": 7.17794486215539e-06,
+      "loss": 0.4234299957752228,
+      "num_input_tokens_seen": 405962,
+      "step": 1285,
+      "train_runtime": 3431.6612,
+      "train_tokens_per_second": 118.299
+    },
+    {
+      "epoch": 0.0591264367816092,
+      "grad_norm": 7.021231651306152,
+      "learning_rate": 7.167919799498747e-06,
+      "loss": 0.3952452838420868,
+      "num_input_tokens_seen": 406498,
+      "step": 1286,
+      "train_runtime": 3433.7179,
+      "train_tokens_per_second": 118.384
+    },
+    {
+      "epoch": 0.05917241379310345,
+      "grad_norm": 7.601665019989014,
+      "learning_rate": 7.157894736842106e-06,
+      "loss": 0.49580442905426025,
+      "num_input_tokens_seen": 406828,
+      "step": 1287,
+      "train_runtime": 3435.7154,
+      "train_tokens_per_second": 118.411
+    },
+    {
+      "epoch": 0.059218390804597704,
+      "grad_norm": 6.956839561462402,
+      "learning_rate": 7.147869674185464e-06,
+      "loss": 0.34600839018821716,
+      "num_input_tokens_seen": 407066,
+      "step": 1288,
+      "train_runtime": 3437.6834,
+      "train_tokens_per_second": 118.413
+    },
+    {
+      "epoch": 0.059264367816091956,
+      "grad_norm": 7.3096184730529785,
+      "learning_rate": 7.137844611528823e-06,
+      "loss": 0.489816278219223,
+      "num_input_tokens_seen": 407428,
+      "step": 1289,
+      "train_runtime": 3439.6811,
+      "train_tokens_per_second": 118.449
+    },
+    {
+      "epoch": 0.05931034482758621,
+      "grad_norm": 6.580732345581055,
+      "learning_rate": 7.127819548872181e-06,
+      "loss": 0.3271704316139221,
+      "num_input_tokens_seen": 407730,
+      "step": 1290,
+      "train_runtime": 3441.6365,
+      "train_tokens_per_second": 118.47
+    },
+    {
+      "epoch": 0.05935632183908046,
+      "grad_norm": 7.424761772155762,
+      "learning_rate": 7.117794486215539e-06,
+      "loss": 0.5078544616699219,
+      "num_input_tokens_seen": 408040,
+      "step": 1291,
+      "train_runtime": 3456.7098,
+      "train_tokens_per_second": 118.043
+    },
+    {
+      "epoch": 0.059402298850574714,
+      "grad_norm": 6.063399314880371,
+      "learning_rate": 7.107769423558898e-06,
+      "loss": 0.3531324565410614,
+      "num_input_tokens_seen": 408322,
+      "step": 1292,
+      "train_runtime": 3458.8274,
+      "train_tokens_per_second": 118.052
+    },
+    {
+      "epoch": 0.05944827586206897,
+      "grad_norm": 5.98874044418335,
+      "learning_rate": 7.097744360902257e-06,
+      "loss": 0.2616596519947052,
+      "num_input_tokens_seen": 408570,
+      "step": 1293,
+      "train_runtime": 3460.8696,
+      "train_tokens_per_second": 118.054
+    },
+    {
+      "epoch": 0.05949425287356322,
+      "grad_norm": 8.002218246459961,
+      "learning_rate": 7.087719298245614e-06,
+      "loss": 0.5409700870513916,
+      "num_input_tokens_seen": 408980,
+      "step": 1294,
+      "train_runtime": 3462.8598,
+      "train_tokens_per_second": 118.105
+    },
+    {
+      "epoch": 0.05954022988505747,
+      "grad_norm": 8.38762378692627,
+      "learning_rate": 7.077694235588973e-06,
+      "loss": 0.5406583547592163,
+      "num_input_tokens_seen": 409380,
+      "step": 1295,
+      "train_runtime": 3464.8512,
+      "train_tokens_per_second": 118.152
+    },
+    {
+      "epoch": 0.059586206896551724,
+      "grad_norm": 6.846781253814697,
+      "learning_rate": 7.067669172932331e-06,
+      "loss": 0.4995633065700531,
+      "num_input_tokens_seen": 409840,
+      "step": 1296,
+      "train_runtime": 3466.8324,
+      "train_tokens_per_second": 118.217
+    },
+    {
+      "epoch": 0.05963218390804598,
+      "grad_norm": 7.333960056304932,
+      "learning_rate": 7.05764411027569e-06,
+      "loss": 0.39683273434638977,
+      "num_input_tokens_seen": 410066,
+      "step": 1297,
+      "train_runtime": 3468.7775,
+      "train_tokens_per_second": 118.216
+    },
+    {
+      "epoch": 0.05967816091954023,
+      "grad_norm": 8.614952087402344,
+      "learning_rate": 7.047619047619048e-06,
+      "loss": 0.48535892367362976,
+      "num_input_tokens_seen": 410390,
+      "step": 1298,
+      "train_runtime": 3470.7503,
+      "train_tokens_per_second": 118.242
+    },
+    {
+      "epoch": 0.05972413793103448,
+      "grad_norm": 7.655506134033203,
+      "learning_rate": 7.037593984962407e-06,
+      "loss": 0.3905705511569977,
+      "num_input_tokens_seen": 410762,
+      "step": 1299,
+      "train_runtime": 3472.7062,
+      "train_tokens_per_second": 118.283
+    },
+    {
+      "epoch": 0.059770114942528735,
+      "grad_norm": 6.981317043304443,
+      "learning_rate": 7.027568922305765e-06,
+      "loss": 0.3227069079875946,
+      "num_input_tokens_seen": 411000,
+      "step": 1300,
+      "train_runtime": 3474.6614,
+      "train_tokens_per_second": 118.285
+    },
+    {
+      "epoch": 0.05981609195402299,
+      "grad_norm": 7.703155517578125,
+      "learning_rate": 7.017543859649123e-06,
+      "loss": 0.4682047963142395,
+      "num_input_tokens_seen": 411256,
+      "step": 1301,
+      "train_runtime": 3476.606,
+      "train_tokens_per_second": 118.292
+    },
+    {
+      "epoch": 0.05986206896551724,
+      "grad_norm": 7.655060768127441,
+      "learning_rate": 7.0075187969924815e-06,
+      "loss": 0.4079965353012085,
+      "num_input_tokens_seen": 411528,
+      "step": 1302,
+      "train_runtime": 3478.5598,
+      "train_tokens_per_second": 118.304
+    },
+    {
+      "epoch": 0.05990804597701149,
+      "grad_norm": 10.578437805175781,
+      "learning_rate": 6.997493734335841e-06,
+      "loss": 0.5338407754898071,
+      "num_input_tokens_seen": 411888,
+      "step": 1303,
+      "train_runtime": 3480.5568,
+      "train_tokens_per_second": 118.34
+    },
+    {
+      "epoch": 0.059954022988505745,
+      "grad_norm": 9.146214485168457,
+      "learning_rate": 6.9874686716791984e-06,
+      "loss": 0.5038683414459229,
+      "num_input_tokens_seen": 412268,
+      "step": 1304,
+      "train_runtime": 3482.5921,
+      "train_tokens_per_second": 118.38
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 8.496904373168945,
+      "learning_rate": 6.977443609022557e-06,
+      "loss": 0.5401176810264587,
+      "num_input_tokens_seen": 412628,
+      "step": 1305,
+      "train_runtime": 3484.591,
+      "train_tokens_per_second": 118.415
+    },
+    {
+      "epoch": 0.06004597701149425,
+      "grad_norm": 7.559762477874756,
+      "learning_rate": 6.9674185463659146e-06,
+      "loss": 0.6347652673721313,
+      "num_input_tokens_seen": 412996,
+      "step": 1306,
+      "train_runtime": 3486.5571,
+      "train_tokens_per_second": 118.454
+    },
+    {
+      "epoch": 0.0600919540229885,
+      "grad_norm": 7.396735191345215,
+      "learning_rate": 6.957393483709274e-06,
+      "loss": 0.483487606048584,
+      "num_input_tokens_seen": 413252,
+      "step": 1307,
+      "train_runtime": 3488.6458,
+      "train_tokens_per_second": 118.456
+    },
+    {
+      "epoch": 0.060137931034482756,
+      "grad_norm": 7.182208061218262,
+      "learning_rate": 6.947368421052632e-06,
+      "loss": 0.3920372426509857,
+      "num_input_tokens_seen": 413554,
+      "step": 1308,
+      "train_runtime": 3490.5904,
+      "train_tokens_per_second": 118.477
+    },
+    {
+      "epoch": 0.06018390804597701,
+      "grad_norm": 10.241389274597168,
+      "learning_rate": 6.93734335839599e-06,
+      "loss": 0.42586126923561096,
+      "num_input_tokens_seen": 413836,
+      "step": 1309,
+      "train_runtime": 3492.5411,
+      "train_tokens_per_second": 118.491
+    },
+    {
+      "epoch": 0.06022988505747127,
+      "grad_norm": 7.382781028747559,
+      "learning_rate": 6.927318295739349e-06,
+      "loss": 0.4327068030834198,
+      "num_input_tokens_seen": 414068,
+      "step": 1310,
+      "train_runtime": 3494.4816,
+      "train_tokens_per_second": 118.492
+    },
+    {
+      "epoch": 0.06027586206896552,
+      "grad_norm": 7.766989231109619,
+      "learning_rate": 6.917293233082707e-06,
+      "loss": 0.43942388892173767,
+      "num_input_tokens_seen": 414360,
+      "step": 1311,
+      "train_runtime": 3496.4585,
+      "train_tokens_per_second": 118.508
+    },
+    {
+      "epoch": 0.06032183908045977,
+      "grad_norm": 8.081896781921387,
+      "learning_rate": 6.9072681704260655e-06,
+      "loss": 0.5534281134605408,
+      "num_input_tokens_seen": 414716,
+      "step": 1312,
+      "train_runtime": 3498.4216,
+      "train_tokens_per_second": 118.544
+    },
+    {
+      "epoch": 0.060367816091954025,
+      "grad_norm": 8.45753002166748,
+      "learning_rate": 6.897243107769425e-06,
+      "loss": 0.4607003331184387,
+      "num_input_tokens_seen": 414934,
+      "step": 1313,
+      "train_runtime": 3500.3525,
+      "train_tokens_per_second": 118.541
+    },
+    {
+      "epoch": 0.06041379310344828,
+      "grad_norm": 7.786255836486816,
+      "learning_rate": 6.8872180451127825e-06,
+      "loss": 0.4713561534881592,
+      "num_input_tokens_seen": 415214,
+      "step": 1314,
+      "train_runtime": 3502.3008,
+      "train_tokens_per_second": 118.555
+    },
+    {
+      "epoch": 0.06045977011494253,
+      "grad_norm": 7.114749908447266,
+      "learning_rate": 6.877192982456141e-06,
+      "loss": 0.4207516014575958,
+      "num_input_tokens_seen": 415576,
+      "step": 1315,
+      "train_runtime": 3504.2469,
+      "train_tokens_per_second": 118.592
+    },
+    {
+      "epoch": 0.06050574712643678,
+      "grad_norm": 6.800426959991455,
+      "learning_rate": 6.867167919799499e-06,
+      "loss": 0.3828384578227997,
+      "num_input_tokens_seen": 415894,
+      "step": 1316,
+      "train_runtime": 3506.1857,
+      "train_tokens_per_second": 118.617
+    },
+    {
+      "epoch": 0.060551724137931036,
+      "grad_norm": 5.2941060066223145,
+      "learning_rate": 6.857142857142858e-06,
+      "loss": 0.3798636794090271,
+      "num_input_tokens_seen": 416944,
+      "step": 1317,
+      "train_runtime": 3508.3295,
+      "train_tokens_per_second": 118.844
+    },
+    {
+      "epoch": 0.06059770114942529,
+      "grad_norm": 7.210157871246338,
+      "learning_rate": 6.847117794486216e-06,
+      "loss": 0.428189754486084,
+      "num_input_tokens_seen": 417384,
+      "step": 1318,
+      "train_runtime": 3510.3183,
+      "train_tokens_per_second": 118.902
+    },
+    {
+      "epoch": 0.06064367816091954,
+      "grad_norm": 7.294408798217773,
+      "learning_rate": 6.837092731829574e-06,
+      "loss": 0.42618033289909363,
+      "num_input_tokens_seen": 417738,
+      "step": 1319,
+      "train_runtime": 3512.2878,
+      "train_tokens_per_second": 118.936
+    },
+    {
+      "epoch": 0.060689655172413794,
+      "grad_norm": 7.582489013671875,
+      "learning_rate": 6.827067669172933e-06,
+      "loss": 0.34816285967826843,
+      "num_input_tokens_seen": 417996,
+      "step": 1320,
+      "train_runtime": 3514.2285,
+      "train_tokens_per_second": 118.944
+    },
+    {
+      "epoch": 0.060735632183908046,
+      "grad_norm": 6.12922477722168,
+      "learning_rate": 6.817042606516291e-06,
+      "loss": 0.3263629674911499,
+      "num_input_tokens_seen": 418246,
+      "step": 1321,
+      "train_runtime": 3529.3064,
+      "train_tokens_per_second": 118.507
+    },
+    {
+      "epoch": 0.0607816091954023,
+      "grad_norm": 10.189580917358398,
+      "learning_rate": 6.8070175438596495e-06,
+      "loss": 0.6183764338493347,
+      "num_input_tokens_seen": 418528,
+      "step": 1322,
+      "train_runtime": 3531.4512,
+      "train_tokens_per_second": 118.514
+    },
+    {
+      "epoch": 0.06082758620689655,
+      "grad_norm": 7.070530414581299,
+      "learning_rate": 6.796992481203009e-06,
+      "loss": 0.3519631624221802,
+      "num_input_tokens_seen": 418774,
+      "step": 1323,
+      "train_runtime": 3533.4149,
+      "train_tokens_per_second": 118.518
+    },
+    {
+      "epoch": 0.060873563218390804,
+      "grad_norm": 7.724711894989014,
+      "learning_rate": 6.7869674185463665e-06,
+      "loss": 0.4881995618343353,
+      "num_input_tokens_seen": 419070,
+      "step": 1324,
+      "train_runtime": 3535.3752,
+      "train_tokens_per_second": 118.536
+    },
+    {
+      "epoch": 0.06091954022988506,
+      "grad_norm": 5.729980945587158,
+      "learning_rate": 6.776942355889725e-06,
+      "loss": 0.29812783002853394,
+      "num_input_tokens_seen": 419334,
+      "step": 1325,
+      "train_runtime": 3537.3324,
+      "train_tokens_per_second": 118.545
+    },
+    {
+      "epoch": 0.06096551724137931,
+      "grad_norm": 8.302903175354004,
+      "learning_rate": 6.766917293233083e-06,
+      "loss": 0.48672789335250854,
+      "num_input_tokens_seen": 419648,
+      "step": 1326,
+      "train_runtime": 3539.3059,
+      "train_tokens_per_second": 118.568
+    },
+    {
+      "epoch": 0.06101149425287356,
+      "grad_norm": 7.919209003448486,
+      "learning_rate": 6.756892230576442e-06,
+      "loss": 0.6153263449668884,
+      "num_input_tokens_seen": 419968,
+      "step": 1327,
+      "train_runtime": 3541.2637,
+      "train_tokens_per_second": 118.593
+    },
+    {
+      "epoch": 0.061057471264367814,
+      "grad_norm": 7.529419422149658,
+      "learning_rate": 6.7468671679198004e-06,
+      "loss": 0.48105958104133606,
+      "num_input_tokens_seen": 420248,
+      "step": 1328,
+      "train_runtime": 3543.2024,
+      "train_tokens_per_second": 118.607
+    },
+    {
+      "epoch": 0.06110344827586207,
+      "grad_norm": 7.931469917297363,
+      "learning_rate": 6.736842105263158e-06,
+      "loss": 0.49645185470581055,
+      "num_input_tokens_seen": 420516,
+      "step": 1329,
+      "train_runtime": 3545.1423,
+      "train_tokens_per_second": 118.618
+    },
+    {
+      "epoch": 0.06114942528735632,
+      "grad_norm": 7.305312156677246,
+      "learning_rate": 6.726817042606517e-06,
+      "loss": 0.434931218624115,
+      "num_input_tokens_seen": 420802,
+      "step": 1330,
+      "train_runtime": 3547.092,
+      "train_tokens_per_second": 118.633
+    },
+    {
+      "epoch": 0.06119540229885057,
+      "grad_norm": 8.194070816040039,
+      "learning_rate": 6.716791979949875e-06,
+      "loss": 0.5417882204055786,
+      "num_input_tokens_seen": 421060,
+      "step": 1331,
+      "train_runtime": 3549.0343,
+      "train_tokens_per_second": 118.641
+    },
+    {
+      "epoch": 0.061241379310344825,
+      "grad_norm": 11.66019058227539,
+      "learning_rate": 6.7067669172932335e-06,
+      "loss": 0.35559171438217163,
+      "num_input_tokens_seen": 421304,
+      "step": 1332,
+      "train_runtime": 3550.9941,
+      "train_tokens_per_second": 118.644
+    },
+    {
+      "epoch": 0.06128735632183908,
+      "grad_norm": 6.619271755218506,
+      "learning_rate": 6.696741854636593e-06,
+      "loss": 0.40130195021629333,
+      "num_input_tokens_seen": 421586,
+      "step": 1333,
+      "train_runtime": 3552.9545,
+      "train_tokens_per_second": 118.658
+    },
+    {
+      "epoch": 0.06133333333333333,
+      "grad_norm": 8.271845817565918,
+      "learning_rate": 6.6867167919799505e-06,
+      "loss": 0.5244696736335754,
+      "num_input_tokens_seen": 421912,
+      "step": 1334,
+      "train_runtime": 3554.9245,
+      "train_tokens_per_second": 118.684
+    },
+    {
+      "epoch": 0.06137931034482759,
+      "grad_norm": 8.006850242614746,
+      "learning_rate": 6.676691729323309e-06,
+      "loss": 0.48754143714904785,
+      "num_input_tokens_seen": 422202,
+      "step": 1335,
+      "train_runtime": 3556.8743,
+      "train_tokens_per_second": 118.7
+    },
+    {
+      "epoch": 0.06142528735632184,
+      "grad_norm": 5.9052276611328125,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.334730863571167,
+      "num_input_tokens_seen": 422486,
+      "step": 1336,
+      "train_runtime": 3558.8205,
+      "train_tokens_per_second": 118.715
+    },
+    {
+      "epoch": 0.061471264367816095,
+      "grad_norm": 7.1550068855285645,
+      "learning_rate": 6.656641604010026e-06,
+      "loss": 0.357038676738739,
+      "num_input_tokens_seen": 422766,
+      "step": 1337,
+      "train_runtime": 3560.7693,
+      "train_tokens_per_second": 118.729
+    },
+    {
+      "epoch": 0.06151724137931035,
+      "grad_norm": 7.906791687011719,
+      "learning_rate": 6.6466165413533845e-06,
+      "loss": 0.4327539801597595,
+      "num_input_tokens_seen": 423062,
+      "step": 1338,
+      "train_runtime": 3562.756,
+      "train_tokens_per_second": 118.746
+    },
+    {
+      "epoch": 0.0615632183908046,
+      "grad_norm": 8.896774291992188,
+      "learning_rate": 6.636591478696742e-06,
+      "loss": 0.5419115424156189,
+      "num_input_tokens_seen": 423436,
+      "step": 1339,
+      "train_runtime": 3564.7422,
+      "train_tokens_per_second": 118.784
+    },
+    {
+      "epoch": 0.06160919540229885,
+      "grad_norm": 8.44174861907959,
+      "learning_rate": 6.6265664160401014e-06,
+      "loss": 0.49105167388916016,
+      "num_input_tokens_seen": 423758,
+      "step": 1340,
+      "train_runtime": 3566.7401,
+      "train_tokens_per_second": 118.808
+    },
+    {
+      "epoch": 0.061655172413793105,
+      "grad_norm": 8.32571792602539,
+      "learning_rate": 6.616541353383459e-06,
+      "loss": 0.44149884581565857,
+      "num_input_tokens_seen": 424096,
+      "step": 1341,
+      "train_runtime": 3568.7361,
+      "train_tokens_per_second": 118.836
+    },
+    {
+      "epoch": 0.06170114942528736,
+      "grad_norm": 8.409771919250488,
+      "learning_rate": 6.6065162907268176e-06,
+      "loss": 0.3973810076713562,
+      "num_input_tokens_seen": 424368,
+      "step": 1342,
+      "train_runtime": 3570.73,
+      "train_tokens_per_second": 118.846
+    },
+    {
+      "epoch": 0.06174712643678161,
+      "grad_norm": 8.931021690368652,
+      "learning_rate": 6.596491228070177e-06,
+      "loss": 0.468519926071167,
+      "num_input_tokens_seen": 424676,
+      "step": 1343,
+      "train_runtime": 3572.7194,
+      "train_tokens_per_second": 118.866
+    },
+    {
+      "epoch": 0.06179310344827586,
+      "grad_norm": 7.6605401039123535,
+      "learning_rate": 6.5864661654135345e-06,
+      "loss": 0.5063818693161011,
+      "num_input_tokens_seen": 425054,
+      "step": 1344,
+      "train_runtime": 3574.7079,
+      "train_tokens_per_second": 118.906
+    },
+    {
+      "epoch": 0.061839080459770115,
+      "grad_norm": 6.713517665863037,
+      "learning_rate": 6.576441102756893e-06,
+      "loss": 0.500751256942749,
+      "num_input_tokens_seen": 425386,
+      "step": 1345,
+      "train_runtime": 3576.6605,
+      "train_tokens_per_second": 118.934
+    },
+    {
+      "epoch": 0.06188505747126437,
+      "grad_norm": 7.185019493103027,
+      "learning_rate": 6.566416040100251e-06,
+      "loss": 0.47278982400894165,
+      "num_input_tokens_seen": 425668,
+      "step": 1346,
+      "train_runtime": 3578.6073,
+      "train_tokens_per_second": 118.948
+    },
+    {
+      "epoch": 0.06193103448275862,
+      "grad_norm": 7.3665771484375,
+      "learning_rate": 6.55639097744361e-06,
+      "loss": 0.47877997159957886,
+      "num_input_tokens_seen": 426030,
+      "step": 1347,
+      "train_runtime": 3580.6381,
+      "train_tokens_per_second": 118.982
+    },
+    {
+      "epoch": 0.06197701149425287,
+      "grad_norm": 6.61741828918457,
+      "learning_rate": 6.5463659147869685e-06,
+      "loss": 0.4474172294139862,
+      "num_input_tokens_seen": 426364,
+      "step": 1348,
+      "train_runtime": 3582.669,
+      "train_tokens_per_second": 119.007
+    },
+    {
+      "epoch": 0.062022988505747126,
+      "grad_norm": 8.149283409118652,
+      "learning_rate": 6.536340852130326e-06,
+      "loss": 0.4722183644771576,
+      "num_input_tokens_seen": 426662,
+      "step": 1349,
+      "train_runtime": 3584.6178,
+      "train_tokens_per_second": 119.026
+    },
+    {
+      "epoch": 0.06206896551724138,
+      "grad_norm": 8.140104293823242,
+      "learning_rate": 6.526315789473685e-06,
+      "loss": 0.4514812231063843,
+      "num_input_tokens_seen": 426914,
+      "step": 1350,
+      "train_runtime": 3586.5856,
+      "train_tokens_per_second": 119.031
+    },
+    {
+      "epoch": 0.06211494252873563,
+      "grad_norm": 8.069449424743652,
+      "learning_rate": 6.516290726817043e-06,
+      "loss": 0.49403613805770874,
+      "num_input_tokens_seen": 427176,
+      "step": 1351,
+      "train_runtime": 3600.8129,
+      "train_tokens_per_second": 118.633
+    },
+    {
+      "epoch": 0.062160919540229884,
+      "grad_norm": 8.522804260253906,
+      "learning_rate": 6.506265664160402e-06,
+      "loss": 0.5751454830169678,
+      "num_input_tokens_seen": 427578,
+      "step": 1352,
+      "train_runtime": 3602.8636,
+      "train_tokens_per_second": 118.677
+    },
+    {
+      "epoch": 0.062206896551724136,
+      "grad_norm": 7.504092216491699,
+      "learning_rate": 6.49624060150376e-06,
+      "loss": 0.47312673926353455,
+      "num_input_tokens_seen": 427856,
+      "step": 1353,
+      "train_runtime": 3604.8132,
+      "train_tokens_per_second": 118.69
+    },
+    {
+      "epoch": 0.06225287356321839,
+      "grad_norm": 5.708718299865723,
+      "learning_rate": 6.486215538847119e-06,
+      "loss": 0.4190950393676758,
+      "num_input_tokens_seen": 428516,
+      "step": 1354,
+      "train_runtime": 3606.855,
+      "train_tokens_per_second": 118.806
+    },
+    {
+      "epoch": 0.06229885057471264,
+      "grad_norm": 8.364093780517578,
+      "learning_rate": 6.476190476190477e-06,
+      "loss": 0.4656526446342468,
+      "num_input_tokens_seen": 428836,
+      "step": 1355,
+      "train_runtime": 3608.8256,
+      "train_tokens_per_second": 118.83
+    },
+    {
+      "epoch": 0.062344827586206894,
+      "grad_norm": 7.833532333374023,
+      "learning_rate": 6.466165413533835e-06,
+      "loss": 0.5170052647590637,
+      "num_input_tokens_seen": 429170,
+      "step": 1356,
+      "train_runtime": 3610.8219,
+      "train_tokens_per_second": 118.857
+    },
+    {
+      "epoch": 0.06239080459770115,
+      "grad_norm": 7.987048625946045,
+      "learning_rate": 6.456140350877193e-06,
+      "loss": 0.4124986529350281,
+      "num_input_tokens_seen": 429470,
+      "step": 1357,
+      "train_runtime": 3612.8281,
+      "train_tokens_per_second": 118.874
+    },
+    {
+      "epoch": 0.0624367816091954,
+      "grad_norm": 6.0981831550598145,
+      "learning_rate": 6.4461152882205525e-06,
+      "loss": 0.300201952457428,
+      "num_input_tokens_seen": 429750,
+      "step": 1358,
+      "train_runtime": 3614.7847,
+      "train_tokens_per_second": 118.887
+    },
+    {
+      "epoch": 0.06248275862068965,
+      "grad_norm": 7.8474507331848145,
+      "learning_rate": 6.43609022556391e-06,
+      "loss": 0.31665804982185364,
+      "num_input_tokens_seen": 429990,
+      "step": 1359,
+      "train_runtime": 3616.7413,
+      "train_tokens_per_second": 118.889
+    },
+    {
+      "epoch": 0.0625287356321839,
+      "grad_norm": 8.21008586883545,
+      "learning_rate": 6.426065162907269e-06,
+      "loss": 0.5030484795570374,
+      "num_input_tokens_seen": 430508,
+      "step": 1360,
+      "train_runtime": 3618.7621,
+      "train_tokens_per_second": 118.966
+    },
+    {
+      "epoch": 0.06257471264367816,
+      "grad_norm": 11.589370727539062,
+      "learning_rate": 6.416040100250627e-06,
+      "loss": 0.4762667119503021,
+      "num_input_tokens_seen": 430838,
+      "step": 1361,
+      "train_runtime": 3620.7453,
+      "train_tokens_per_second": 118.992
+    },
+    {
+      "epoch": 0.06262068965517241,
+      "grad_norm": 6.966498851776123,
+      "learning_rate": 6.406015037593986e-06,
+      "loss": 0.3605503737926483,
+      "num_input_tokens_seen": 431164,
+      "step": 1362,
+      "train_runtime": 3622.7009,
+      "train_tokens_per_second": 119.017
+    },
+    {
+      "epoch": 0.06266666666666666,
+      "grad_norm": 8.372251510620117,
+      "learning_rate": 6.395989974937344e-06,
+      "loss": 0.5541883111000061,
+      "num_input_tokens_seen": 431522,
+      "step": 1363,
+      "train_runtime": 3624.7763,
+      "train_tokens_per_second": 119.048
+    },
+    {
+      "epoch": 0.06271264367816091,
+      "grad_norm": 8.219268798828125,
+      "learning_rate": 6.385964912280702e-06,
+      "loss": 0.4730372130870819,
+      "num_input_tokens_seen": 431852,
+      "step": 1364,
+      "train_runtime": 3626.7527,
+      "train_tokens_per_second": 119.074
+    },
+    {
+      "epoch": 0.06275862068965517,
+      "grad_norm": 7.887789726257324,
+      "learning_rate": 6.375939849624061e-06,
+      "loss": 0.5083958506584167,
+      "num_input_tokens_seen": 432178,
+      "step": 1365,
+      "train_runtime": 3628.7334,
+      "train_tokens_per_second": 119.099
+    },
+    {
+      "epoch": 0.06280459770114942,
+      "grad_norm": 8.433984756469727,
+      "learning_rate": 6.365914786967419e-06,
+      "loss": 0.4906807243824005,
+      "num_input_tokens_seen": 432400,
+      "step": 1366,
+      "train_runtime": 3630.6867,
+      "train_tokens_per_second": 119.096
+    },
+    {
+      "epoch": 0.06285057471264367,
+      "grad_norm": 6.564528465270996,
+      "learning_rate": 6.355889724310777e-06,
+      "loss": 0.42632395029067993,
+      "num_input_tokens_seen": 432766,
+      "step": 1367,
+      "train_runtime": 3632.6607,
+      "train_tokens_per_second": 119.132
+    },
+    {
+      "epoch": 0.06289655172413793,
+      "grad_norm": 7.940859317779541,
+      "learning_rate": 6.3458646616541366e-06,
+      "loss": 0.38434866070747375,
+      "num_input_tokens_seen": 433018,
+      "step": 1368,
+      "train_runtime": 3634.6848,
+      "train_tokens_per_second": 119.135
+    },
+    {
+      "epoch": 0.06294252873563218,
+      "grad_norm": 15.380034446716309,
+      "learning_rate": 6.335839598997494e-06,
+      "loss": 0.43325796723365784,
+      "num_input_tokens_seen": 433280,
+      "step": 1369,
+      "train_runtime": 3636.682,
+      "train_tokens_per_second": 119.142
+    },
+    {
+      "epoch": 0.06298850574712643,
+      "grad_norm": 8.542527198791504,
+      "learning_rate": 6.325814536340853e-06,
+      "loss": 0.5187516212463379,
+      "num_input_tokens_seen": 433534,
+      "step": 1370,
+      "train_runtime": 3638.6375,
+      "train_tokens_per_second": 119.147
+    },
+    {
+      "epoch": 0.06303448275862068,
+      "grad_norm": 7.916285514831543,
+      "learning_rate": 6.31578947368421e-06,
+      "loss": 0.4286421537399292,
+      "num_input_tokens_seen": 433842,
+      "step": 1371,
+      "train_runtime": 3640.6189,
+      "train_tokens_per_second": 119.167
+    },
+    {
+      "epoch": 0.06308045977011495,
+      "grad_norm": 6.741034984588623,
+      "learning_rate": 6.30576441102757e-06,
+      "loss": 0.360577255487442,
+      "num_input_tokens_seen": 434152,
+      "step": 1372,
+      "train_runtime": 3642.5855,
+      "train_tokens_per_second": 119.188
+    },
+    {
+      "epoch": 0.0631264367816092,
+      "grad_norm": 6.093510150909424,
+      "learning_rate": 6.295739348370928e-06,
+      "loss": 0.3625005781650543,
+      "num_input_tokens_seen": 434436,
+      "step": 1373,
+      "train_runtime": 3644.5764,
+      "train_tokens_per_second": 119.201
+    },
+    {
+      "epoch": 0.06317241379310345,
+      "grad_norm": 7.795077323913574,
+      "learning_rate": 6.285714285714286e-06,
+      "loss": 0.5406826734542847,
+      "num_input_tokens_seen": 434800,
+      "step": 1374,
+      "train_runtime": 3646.5783,
+      "train_tokens_per_second": 119.235
+    },
+    {
+      "epoch": 0.06321839080459771,
+      "grad_norm": 8.354170799255371,
+      "learning_rate": 6.275689223057645e-06,
+      "loss": 0.47981852293014526,
+      "num_input_tokens_seen": 435068,
+      "step": 1375,
+      "train_runtime": 3648.5481,
+      "train_tokens_per_second": 119.244
+    },
+    {
+      "epoch": 0.06326436781609196,
+      "grad_norm": 7.820892333984375,
+      "learning_rate": 6.265664160401003e-06,
+      "loss": 0.5885209441184998,
+      "num_input_tokens_seen": 435390,
+      "step": 1376,
+      "train_runtime": 3650.5505,
+      "train_tokens_per_second": 119.267
+    },
+    {
+      "epoch": 0.06331034482758621,
+      "grad_norm": 8.482800483703613,
+      "learning_rate": 6.255639097744361e-06,
+      "loss": 0.49485084414482117,
+      "num_input_tokens_seen": 435874,
+      "step": 1377,
+      "train_runtime": 3652.5762,
+      "train_tokens_per_second": 119.333
+    },
+    {
+      "epoch": 0.06335632183908046,
+      "grad_norm": 9.001823425292969,
+      "learning_rate": 6.245614035087721e-06,
+      "loss": 0.41019052267074585,
+      "num_input_tokens_seen": 436094,
+      "step": 1378,
+      "train_runtime": 3654.541,
+      "train_tokens_per_second": 119.329
+    },
+    {
+      "epoch": 0.06340229885057472,
+      "grad_norm": 8.457934379577637,
+      "learning_rate": 6.235588972431078e-06,
+      "loss": 0.5790400505065918,
+      "num_input_tokens_seen": 436438,
+      "step": 1379,
+      "train_runtime": 3656.5834,
+      "train_tokens_per_second": 119.357
+    },
+    {
+      "epoch": 0.06344827586206897,
+      "grad_norm": 6.0701518058776855,
+      "learning_rate": 6.225563909774437e-06,
+      "loss": 0.4767570197582245,
+      "num_input_tokens_seen": 437090,
+      "step": 1380,
+      "train_runtime": 3658.6915,
+      "train_tokens_per_second": 119.466
+    },
+    {
+      "epoch": 0.06349425287356322,
+      "grad_norm": 7.297056198120117,
+      "learning_rate": 6.215538847117794e-06,
+      "loss": 0.5049152374267578,
+      "num_input_tokens_seen": 437380,
+      "step": 1381,
+      "train_runtime": 3673.5632,
+      "train_tokens_per_second": 119.062
+    },
+    {
+      "epoch": 0.06354022988505748,
+      "grad_norm": 7.5144171714782715,
+      "learning_rate": 6.205513784461154e-06,
+      "loss": 0.4231078624725342,
+      "num_input_tokens_seen": 437678,
+      "step": 1382,
+      "train_runtime": 3675.5234,
+      "train_tokens_per_second": 119.079
+    },
+    {
+      "epoch": 0.06358620689655173,
+      "grad_norm": 7.24461030960083,
+      "learning_rate": 6.195488721804512e-06,
+      "loss": 0.41516321897506714,
+      "num_input_tokens_seen": 437950,
+      "step": 1383,
+      "train_runtime": 3677.4903,
+      "train_tokens_per_second": 119.089
+    },
+    {
+      "epoch": 0.06363218390804598,
+      "grad_norm": 7.727677345275879,
+      "learning_rate": 6.18546365914787e-06,
+      "loss": 0.40505528450012207,
+      "num_input_tokens_seen": 438206,
+      "step": 1384,
+      "train_runtime": 3679.4483,
+      "train_tokens_per_second": 119.096
+    },
+    {
+      "epoch": 0.06367816091954023,
+      "grad_norm": 7.370773792266846,
+      "learning_rate": 6.175438596491229e-06,
+      "loss": 0.43799248337745667,
+      "num_input_tokens_seen": 438468,
+      "step": 1385,
+      "train_runtime": 3681.4073,
+      "train_tokens_per_second": 119.103
+    },
+    {
+      "epoch": 0.06372413793103449,
+      "grad_norm": 8.690412521362305,
+      "learning_rate": 6.165413533834587e-06,
+      "loss": 0.41677355766296387,
+      "num_input_tokens_seen": 438742,
+      "step": 1386,
+      "train_runtime": 3683.3517,
+      "train_tokens_per_second": 119.115
+    },
+    {
+      "epoch": 0.06377011494252874,
+      "grad_norm": 8.734532356262207,
+      "learning_rate": 6.155388471177945e-06,
+      "loss": 0.5436221361160278,
+      "num_input_tokens_seen": 438994,
+      "step": 1387,
+      "train_runtime": 3685.285,
+      "train_tokens_per_second": 119.121
+    },
+    {
+      "epoch": 0.06381609195402299,
+      "grad_norm": 9.047920227050781,
+      "learning_rate": 6.145363408521305e-06,
+      "loss": 0.3691633641719818,
+      "num_input_tokens_seen": 439276,
+      "step": 1388,
+      "train_runtime": 3687.2795,
+      "train_tokens_per_second": 119.133
+    },
+    {
+      "epoch": 0.06386206896551724,
+      "grad_norm": 7.246007442474365,
+      "learning_rate": 6.135338345864662e-06,
+      "loss": 0.4282595217227936,
+      "num_input_tokens_seen": 439522,
+      "step": 1389,
+      "train_runtime": 3689.2213,
+      "train_tokens_per_second": 119.137
+    },
+    {
+      "epoch": 0.0639080459770115,
+      "grad_norm": 7.2857441902160645,
+      "learning_rate": 6.125313283208021e-06,
+      "loss": 0.3658398687839508,
+      "num_input_tokens_seen": 439784,
+      "step": 1390,
+      "train_runtime": 3691.1706,
+      "train_tokens_per_second": 119.145
+    },
+    {
+      "epoch": 0.06395402298850575,
+      "grad_norm": 7.951290130615234,
+      "learning_rate": 6.115288220551378e-06,
+      "loss": 0.45302778482437134,
+      "num_input_tokens_seen": 440146,
+      "step": 1391,
+      "train_runtime": 3693.1312,
+      "train_tokens_per_second": 119.18
+    },
+    {
+      "epoch": 0.064,
+      "grad_norm": 9.196980476379395,
+      "learning_rate": 6.105263157894738e-06,
+      "loss": 0.5447526574134827,
+      "num_input_tokens_seen": 440538,
+      "step": 1392,
+      "train_runtime": 3695.1157,
+      "train_tokens_per_second": 119.222
+    },
+    {
+      "epoch": 0.06404597701149425,
+      "grad_norm": 8.422050476074219,
+      "learning_rate": 6.095238095238096e-06,
+      "loss": 0.5552542209625244,
+      "num_input_tokens_seen": 440994,
+      "step": 1393,
+      "train_runtime": 3697.2557,
+      "train_tokens_per_second": 119.276
+    },
+    {
+      "epoch": 0.0640919540229885,
+      "grad_norm": 6.602896213531494,
+      "learning_rate": 6.085213032581454e-06,
+      "loss": 0.3959818482398987,
+      "num_input_tokens_seen": 441340,
+      "step": 1394,
+      "train_runtime": 3699.2815,
+      "train_tokens_per_second": 119.304
+    },
+    {
+      "epoch": 0.06413793103448276,
+      "grad_norm": 7.989767074584961,
+      "learning_rate": 6.075187969924813e-06,
+      "loss": 0.5790979266166687,
+      "num_input_tokens_seen": 441730,
+      "step": 1395,
+      "train_runtime": 3701.3001,
+      "train_tokens_per_second": 119.345
+    },
+    {
+      "epoch": 0.06418390804597701,
+      "grad_norm": 7.779969692230225,
+      "learning_rate": 6.065162907268171e-06,
+      "loss": 0.543576717376709,
+      "num_input_tokens_seen": 441974,
+      "step": 1396,
+      "train_runtime": 3703.2411,
+      "train_tokens_per_second": 119.348
+    },
+    {
+      "epoch": 0.06422988505747126,
+      "grad_norm": 6.791033744812012,
+      "learning_rate": 6.055137844611529e-06,
+      "loss": 0.48290449380874634,
+      "num_input_tokens_seen": 442536,
+      "step": 1397,
+      "train_runtime": 3705.2614,
+      "train_tokens_per_second": 119.434
+    },
+    {
+      "epoch": 0.06427586206896552,
+      "grad_norm": 6.259047508239746,
+      "learning_rate": 6.045112781954889e-06,
+      "loss": 0.330578476190567,
+      "num_input_tokens_seen": 442760,
+      "step": 1398,
+      "train_runtime": 3707.2204,
+      "train_tokens_per_second": 119.432
+    },
+    {
+      "epoch": 0.06432183908045977,
+      "grad_norm": 9.04414176940918,
+      "learning_rate": 6.035087719298246e-06,
+      "loss": 0.40303677320480347,
+      "num_input_tokens_seen": 443066,
+      "step": 1399,
+      "train_runtime": 3709.1908,
+      "train_tokens_per_second": 119.451
+    },
+    {
+      "epoch": 0.06436781609195402,
+      "grad_norm": 6.998606204986572,
+      "learning_rate": 6.025062656641605e-06,
+      "loss": 0.4123372435569763,
+      "num_input_tokens_seen": 443440,
+      "step": 1400,
+      "train_runtime": 3711.1837,
+      "train_tokens_per_second": 119.487
+    },
+    {
+      "epoch": 0.06436781609195402,
+      "eval_loss": 1.6728938817977905,
+      "eval_runtime": 53.9714,
+      "eval_samples_per_second": 18.528,
+      "eval_steps_per_second": 9.264,
+      "num_input_tokens_seen": 443440,
+      "step": 1400
+    },
+    {
+      "epoch": 0.06441379310344827,
+      "grad_norm": 9.046266555786133,
+      "learning_rate": 6.015037593984962e-06,
+      "loss": 0.5548663139343262,
+      "num_input_tokens_seen": 443750,
+      "step": 1401,
+      "train_runtime": 3767.1391,
+      "train_tokens_per_second": 117.795
+    },
+    {
+      "epoch": 0.06445977011494253,
+      "grad_norm": 6.469735145568848,
+      "learning_rate": 6.005012531328322e-06,
+      "loss": 0.5285234451293945,
+      "num_input_tokens_seen": 444274,
+      "step": 1402,
+      "train_runtime": 3769.2767,
+      "train_tokens_per_second": 117.867
+    },
+    {
+      "epoch": 0.06450574712643678,
+      "grad_norm": 7.658900737762451,
+      "learning_rate": 5.994987468671679e-06,
+      "loss": 0.5017580986022949,
+      "num_input_tokens_seen": 444728,
+      "step": 1403,
+      "train_runtime": 3771.2683,
+      "train_tokens_per_second": 117.925
+    },
+    {
+      "epoch": 0.06455172413793103,
+      "grad_norm": 7.396538734436035,
+      "learning_rate": 5.984962406015038e-06,
+      "loss": 0.450796902179718,
+      "num_input_tokens_seen": 445066,
+      "step": 1404,
+      "train_runtime": 3773.2684,
+      "train_tokens_per_second": 117.952
+    },
+    {
+      "epoch": 0.06459770114942529,
+      "grad_norm": 7.2905731201171875,
+      "learning_rate": 5.974937343358397e-06,
+      "loss": 0.4379667043685913,
+      "num_input_tokens_seen": 445458,
+      "step": 1405,
+      "train_runtime": 3775.3061,
+      "train_tokens_per_second": 117.993
+    },
+    {
+      "epoch": 0.06464367816091954,
+      "grad_norm": 7.959745407104492,
+      "learning_rate": 5.964912280701755e-06,
+      "loss": 0.5967686772346497,
+      "num_input_tokens_seen": 445828,
+      "step": 1406,
+      "train_runtime": 3777.4003,
+      "train_tokens_per_second": 118.025
+    },
+    {
+      "epoch": 0.06468965517241379,
+      "grad_norm": 7.209251880645752,
+      "learning_rate": 5.954887218045113e-06,
+      "loss": 0.5046592950820923,
+      "num_input_tokens_seen": 446200,
+      "step": 1407,
+      "train_runtime": 3779.4699,
+      "train_tokens_per_second": 118.059
+    },
+    {
+      "epoch": 0.06473563218390804,
+      "grad_norm": 7.277596473693848,
+      "learning_rate": 5.944862155388471e-06,
+      "loss": 0.4014708995819092,
+      "num_input_tokens_seen": 446412,
+      "step": 1408,
+      "train_runtime": 3781.5794,
+      "train_tokens_per_second": 118.049
+    },
+    {
+      "epoch": 0.0647816091954023,
+      "grad_norm": 8.122957229614258,
+      "learning_rate": 5.93483709273183e-06,
+      "loss": 0.6335501074790955,
+      "num_input_tokens_seen": 446732,
+      "step": 1409,
+      "train_runtime": 3783.5476,
+      "train_tokens_per_second": 118.072
+    },
+    {
+      "epoch": 0.06482758620689655,
+      "grad_norm": 7.2711501121521,
+      "learning_rate": 5.924812030075189e-06,
+      "loss": 0.496785044670105,
+      "num_input_tokens_seen": 447034,
+      "step": 1410,
+      "train_runtime": 3785.6334,
+      "train_tokens_per_second": 118.087
+    },
+    {
+      "epoch": 0.0648735632183908,
+      "grad_norm": 8.947657585144043,
+      "learning_rate": 5.9147869674185465e-06,
+      "loss": 0.4516250193119049,
+      "num_input_tokens_seen": 447356,
+      "step": 1411,
+      "train_runtime": 3800.4686,
+      "train_tokens_per_second": 117.711
+    },
+    {
+      "epoch": 0.06491954022988505,
+      "grad_norm": 7.818059921264648,
+      "learning_rate": 5.904761904761905e-06,
+      "loss": 0.44249919056892395,
+      "num_input_tokens_seen": 447744,
+      "step": 1412,
+      "train_runtime": 3802.5779,
+      "train_tokens_per_second": 117.747
+    },
+    {
+      "epoch": 0.0649655172413793,
+      "grad_norm": 8.84902286529541,
+      "learning_rate": 5.8947368421052634e-06,
+      "loss": 0.4483208954334259,
+      "num_input_tokens_seen": 448048,
+      "step": 1413,
+      "train_runtime": 3804.5455,
+      "train_tokens_per_second": 117.766
+    },
+    {
+      "epoch": 0.06501149425287356,
+      "grad_norm": 6.511317253112793,
+      "learning_rate": 5.884711779448622e-06,
+      "loss": 0.3982500433921814,
+      "num_input_tokens_seen": 448366,
+      "step": 1414,
+      "train_runtime": 3806.5661,
+      "train_tokens_per_second": 117.788
+    },
+    {
+      "epoch": 0.06505747126436781,
+      "grad_norm": 7.8486104011535645,
+      "learning_rate": 5.87468671679198e-06,
+      "loss": 0.44934552907943726,
+      "num_input_tokens_seen": 448632,
+      "step": 1415,
+      "train_runtime": 3808.5896,
+      "train_tokens_per_second": 117.795
+    },
+    {
+      "epoch": 0.06510344827586206,
+      "grad_norm": 6.668032169342041,
+      "learning_rate": 5.864661654135339e-06,
+      "loss": 0.37988215684890747,
+      "num_input_tokens_seen": 448862,
+      "step": 1416,
+      "train_runtime": 3810.5591,
+      "train_tokens_per_second": 117.794
+    },
+    {
+      "epoch": 0.06514942528735632,
+      "grad_norm": 7.805704116821289,
+      "learning_rate": 5.854636591478697e-06,
+      "loss": 0.4597431719303131,
+      "num_input_tokens_seen": 449268,
+      "step": 1417,
+      "train_runtime": 3812.568,
+      "train_tokens_per_second": 117.839
+    },
+    {
+      "epoch": 0.06519540229885057,
+      "grad_norm": 8.326112747192383,
+      "learning_rate": 5.844611528822055e-06,
+      "loss": 0.4880395829677582,
+      "num_input_tokens_seen": 449592,
+      "step": 1418,
+      "train_runtime": 3814.5293,
+      "train_tokens_per_second": 117.863
+    },
+    {
+      "epoch": 0.06524137931034482,
+      "grad_norm": 7.665692329406738,
+      "learning_rate": 5.8345864661654135e-06,
+      "loss": 0.4802709221839905,
+      "num_input_tokens_seen": 449910,
+      "step": 1419,
+      "train_runtime": 3816.5481,
+      "train_tokens_per_second": 117.884
+    },
+    {
+      "epoch": 0.06528735632183907,
+      "grad_norm": 6.744765281677246,
+      "learning_rate": 5.824561403508773e-06,
+      "loss": 0.2984175384044647,
+      "num_input_tokens_seen": 450120,
+      "step": 1420,
+      "train_runtime": 3818.633,
+      "train_tokens_per_second": 117.875
+    },
+    {
+      "epoch": 0.06533333333333333,
+      "grad_norm": 7.256897926330566,
+      "learning_rate": 5.8145363408521305e-06,
+      "loss": 0.44814056158065796,
+      "num_input_tokens_seen": 450428,
+      "step": 1421,
+      "train_runtime": 3820.6895,
+      "train_tokens_per_second": 117.892
+    },
+    {
+      "epoch": 0.0653793103448276,
+      "grad_norm": 8.124982833862305,
+      "learning_rate": 5.804511278195489e-06,
+      "loss": 0.49199938774108887,
+      "num_input_tokens_seen": 450886,
+      "step": 1422,
+      "train_runtime": 3822.7805,
+      "train_tokens_per_second": 117.947
+    },
+    {
+      "epoch": 0.06542528735632185,
+      "grad_norm": 8.14277458190918,
+      "learning_rate": 5.7944862155388475e-06,
+      "loss": 0.5476075410842896,
+      "num_input_tokens_seen": 451228,
+      "step": 1423,
+      "train_runtime": 3824.7357,
+      "train_tokens_per_second": 117.976
+    },
+    {
+      "epoch": 0.0654712643678161,
+      "grad_norm": 6.699192523956299,
+      "learning_rate": 5.784461152882206e-06,
+      "loss": 0.34197741746902466,
+      "num_input_tokens_seen": 451444,
+      "step": 1424,
+      "train_runtime": 3826.7233,
+      "train_tokens_per_second": 117.971
+    },
+    {
+      "epoch": 0.06551724137931035,
+      "grad_norm": 6.593231678009033,
+      "learning_rate": 5.7744360902255644e-06,
+      "loss": 0.3856680989265442,
+      "num_input_tokens_seen": 451924,
+      "step": 1425,
+      "train_runtime": 3828.8333,
+      "train_tokens_per_second": 118.032
+    },
+    {
+      "epoch": 0.0655632183908046,
+      "grad_norm": 7.20468282699585,
+      "learning_rate": 5.764411027568922e-06,
+      "loss": 0.3096913695335388,
+      "num_input_tokens_seen": 452190,
+      "step": 1426,
+      "train_runtime": 3830.7791,
+      "train_tokens_per_second": 118.041
+    },
+    {
+      "epoch": 0.06560919540229886,
+      "grad_norm": 7.313076972961426,
+      "learning_rate": 5.754385964912281e-06,
+      "loss": 0.5206544995307922,
+      "num_input_tokens_seen": 452902,
+      "step": 1427,
+      "train_runtime": 3832.8397,
+      "train_tokens_per_second": 118.164
+    },
+    {
+      "epoch": 0.06565517241379311,
+      "grad_norm": 8.288788795471191,
+      "learning_rate": 5.744360902255639e-06,
+      "loss": 0.42484205961227417,
+      "num_input_tokens_seen": 453130,
+      "step": 1428,
+      "train_runtime": 3834.8345,
+      "train_tokens_per_second": 118.162
+    },
+    {
+      "epoch": 0.06570114942528736,
+      "grad_norm": 7.632509231567383,
+      "learning_rate": 5.7343358395989975e-06,
+      "loss": 0.39699095487594604,
+      "num_input_tokens_seen": 453416,
+      "step": 1429,
+      "train_runtime": 3836.7866,
+      "train_tokens_per_second": 118.176
+    },
+    {
+      "epoch": 0.06574712643678161,
+      "grad_norm": 10.508520126342773,
+      "learning_rate": 5.724310776942357e-06,
+      "loss": 0.4499485194683075,
+      "num_input_tokens_seen": 453692,
+      "step": 1430,
+      "train_runtime": 3838.7276,
+      "train_tokens_per_second": 118.188
+    },
+    {
+      "epoch": 0.06579310344827587,
+      "grad_norm": 7.253719329833984,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 0.43608686327934265,
+      "num_input_tokens_seen": 454016,
+      "step": 1431,
+      "train_runtime": 3840.6975,
+      "train_tokens_per_second": 118.212
+    },
+    {
+      "epoch": 0.06583908045977012,
+      "grad_norm": 7.344719886779785,
+      "learning_rate": 5.704260651629073e-06,
+      "loss": 0.4349340796470642,
+      "num_input_tokens_seen": 454282,
+      "step": 1432,
+      "train_runtime": 3842.6286,
+      "train_tokens_per_second": 118.222
+    },
+    {
+      "epoch": 0.06588505747126437,
+      "grad_norm": 11.343481063842773,
+      "learning_rate": 5.694235588972431e-06,
+      "loss": 0.45035284757614136,
+      "num_input_tokens_seen": 454612,
+      "step": 1433,
+      "train_runtime": 3844.5751,
+      "train_tokens_per_second": 118.248
+    },
+    {
+      "epoch": 0.06593103448275862,
+      "grad_norm": 8.23226261138916,
+      "learning_rate": 5.68421052631579e-06,
+      "loss": 0.4740956425666809,
+      "num_input_tokens_seen": 454944,
+      "step": 1434,
+      "train_runtime": 3846.6047,
+      "train_tokens_per_second": 118.272
+    },
+    {
+      "epoch": 0.06597701149425288,
+      "grad_norm": 9.180990219116211,
+      "learning_rate": 5.6741854636591485e-06,
+      "loss": 0.40841978788375854,
+      "num_input_tokens_seen": 455174,
+      "step": 1435,
+      "train_runtime": 3848.5455,
+      "train_tokens_per_second": 118.272
+    },
+    {
+      "epoch": 0.06602298850574713,
+      "grad_norm": 6.975395679473877,
+      "learning_rate": 5.664160401002506e-06,
+      "loss": 0.46920278668403625,
+      "num_input_tokens_seen": 455482,
+      "step": 1436,
+      "train_runtime": 3850.5704,
+      "train_tokens_per_second": 118.289
+    },
+    {
+      "epoch": 0.06606896551724138,
+      "grad_norm": 7.356271743774414,
+      "learning_rate": 5.6541353383458654e-06,
+      "loss": 0.4397551715373993,
+      "num_input_tokens_seen": 455810,
+      "step": 1437,
+      "train_runtime": 3852.5505,
+      "train_tokens_per_second": 118.314
+    },
+    {
+      "epoch": 0.06611494252873563,
+      "grad_norm": 8.738754272460938,
+      "learning_rate": 5.644110275689223e-06,
+      "loss": 0.4972565770149231,
+      "num_input_tokens_seen": 456084,
+      "step": 1438,
+      "train_runtime": 3854.4924,
+      "train_tokens_per_second": 118.325
+    },
+    {
+      "epoch": 0.06616091954022989,
+      "grad_norm": 7.6549577713012695,
+      "learning_rate": 5.6340852130325816e-06,
+      "loss": 0.5112255811691284,
+      "num_input_tokens_seen": 456390,
+      "step": 1439,
+      "train_runtime": 3856.4504,
+      "train_tokens_per_second": 118.345
+    },
+    {
+      "epoch": 0.06620689655172414,
+      "grad_norm": 8.52723503112793,
+      "learning_rate": 5.624060150375941e-06,
+      "loss": 0.479225218296051,
+      "num_input_tokens_seen": 456660,
+      "step": 1440,
+      "train_runtime": 3858.4263,
+      "train_tokens_per_second": 118.354
+    },
+    {
+      "epoch": 0.06625287356321839,
+      "grad_norm": 7.388975620269775,
+      "learning_rate": 5.6140350877192985e-06,
+      "loss": 0.2918142080307007,
+      "num_input_tokens_seen": 456996,
+      "step": 1441,
+      "train_runtime": 3873.3897,
+      "train_tokens_per_second": 117.983
+    },
+    {
+      "epoch": 0.06629885057471264,
+      "grad_norm": 7.529322147369385,
+      "learning_rate": 5.604010025062657e-06,
+      "loss": 0.46961405873298645,
+      "num_input_tokens_seen": 457312,
+      "step": 1442,
+      "train_runtime": 3875.4377,
+      "train_tokens_per_second": 118.003
+    },
+    {
+      "epoch": 0.0663448275862069,
+      "grad_norm": 9.43989372253418,
+      "learning_rate": 5.593984962406015e-06,
+      "loss": 0.5963732004165649,
+      "num_input_tokens_seen": 457634,
+      "step": 1443,
+      "train_runtime": 3877.3831,
+      "train_tokens_per_second": 118.027
+    },
+    {
+      "epoch": 0.06639080459770115,
+      "grad_norm": 12.360222816467285,
+      "learning_rate": 5.583959899749374e-06,
+      "loss": 0.3917125463485718,
+      "num_input_tokens_seen": 457894,
+      "step": 1444,
+      "train_runtime": 3879.3161,
+      "train_tokens_per_second": 118.035
+    },
+    {
+      "epoch": 0.0664367816091954,
+      "grad_norm": 8.409878730773926,
+      "learning_rate": 5.5739348370927325e-06,
+      "loss": 0.47330906987190247,
+      "num_input_tokens_seen": 458194,
+      "step": 1445,
+      "train_runtime": 3881.2778,
+      "train_tokens_per_second": 118.052
+    },
+    {
+      "epoch": 0.06648275862068966,
+      "grad_norm": 6.086939334869385,
+      "learning_rate": 5.56390977443609e-06,
+      "loss": 0.2741166651248932,
+      "num_input_tokens_seen": 458408,
+      "step": 1446,
+      "train_runtime": 3883.2243,
+      "train_tokens_per_second": 118.048
+    },
+    {
+      "epoch": 0.06652873563218391,
+      "grad_norm": 9.503534317016602,
+      "learning_rate": 5.5538847117794495e-06,
+      "loss": 0.4382116496562958,
+      "num_input_tokens_seen": 458702,
+      "step": 1447,
+      "train_runtime": 3885.204,
+      "train_tokens_per_second": 118.064
+    },
+    {
+      "epoch": 0.06657471264367816,
+      "grad_norm": 5.2927045822143555,
+      "learning_rate": 5.543859649122807e-06,
+      "loss": 0.3405531644821167,
+      "num_input_tokens_seen": 459594,
+      "step": 1448,
+      "train_runtime": 3887.3512,
+      "train_tokens_per_second": 118.228
+    },
+    {
+      "epoch": 0.06662068965517241,
+      "grad_norm": 7.682872295379639,
+      "learning_rate": 5.533834586466166e-06,
+      "loss": 0.44900932908058167,
+      "num_input_tokens_seen": 459898,
+      "step": 1449,
+      "train_runtime": 3889.3072,
+      "train_tokens_per_second": 118.247
+    },
+    {
+      "epoch": 0.06666666666666667,
+      "grad_norm": 6.594202995300293,
+      "learning_rate": 5.523809523809525e-06,
+      "loss": 0.2792794406414032,
+      "num_input_tokens_seen": 460168,
+      "step": 1450,
+      "train_runtime": 3891.2575,
+      "train_tokens_per_second": 118.257
+    },
+    {
+      "epoch": 0.06671264367816092,
+      "grad_norm": 6.011887073516846,
+      "learning_rate": 5.5137844611528826e-06,
+      "loss": 0.4249735474586487,
+      "num_input_tokens_seen": 460784,
+      "step": 1451,
+      "train_runtime": 3893.3149,
+      "train_tokens_per_second": 118.353
+    },
+    {
+      "epoch": 0.06675862068965517,
+      "grad_norm": 8.257855415344238,
+      "learning_rate": 5.503759398496241e-06,
+      "loss": 0.4575742185115814,
+      "num_input_tokens_seen": 461102,
+      "step": 1452,
+      "train_runtime": 3895.2778,
+      "train_tokens_per_second": 118.375
+    },
+    {
+      "epoch": 0.06680459770114942,
+      "grad_norm": 9.305832862854004,
+      "learning_rate": 5.493734335839599e-06,
+      "loss": 0.3868868350982666,
+      "num_input_tokens_seen": 461374,
+      "step": 1453,
+      "train_runtime": 3897.367,
+      "train_tokens_per_second": 118.381
+    },
+    {
+      "epoch": 0.06685057471264368,
+      "grad_norm": 7.448078155517578,
+      "learning_rate": 5.483709273182958e-06,
+      "loss": 0.4348413348197937,
+      "num_input_tokens_seen": 461686,
+      "step": 1454,
+      "train_runtime": 3899.3302,
+      "train_tokens_per_second": 118.401
+    },
+    {
+      "epoch": 0.06689655172413793,
+      "grad_norm": 8.324820518493652,
+      "learning_rate": 5.4736842105263165e-06,
+      "loss": 0.48474711179733276,
+      "num_input_tokens_seen": 461928,
+      "step": 1455,
+      "train_runtime": 3901.2856,
+      "train_tokens_per_second": 118.404
+    },
+    {
+      "epoch": 0.06694252873563218,
+      "grad_norm": 7.631561279296875,
+      "learning_rate": 5.463659147869674e-06,
+      "loss": 0.3665958344936371,
+      "num_input_tokens_seen": 462180,
+      "step": 1456,
+      "train_runtime": 3903.2286,
+      "train_tokens_per_second": 118.41
+    },
+    {
+      "epoch": 0.06698850574712643,
+      "grad_norm": 7.605569362640381,
+      "learning_rate": 5.4536340852130335e-06,
+      "loss": 0.3843117952346802,
+      "num_input_tokens_seen": 462412,
+      "step": 1457,
+      "train_runtime": 3905.1744,
+      "train_tokens_per_second": 118.41
+    },
+    {
+      "epoch": 0.06703448275862069,
+      "grad_norm": 7.717450141906738,
+      "learning_rate": 5.443609022556391e-06,
+      "loss": 0.42220956087112427,
+      "num_input_tokens_seen": 462804,
+      "step": 1458,
+      "train_runtime": 3907.1775,
+      "train_tokens_per_second": 118.45
+    },
+    {
+      "epoch": 0.06708045977011494,
+      "grad_norm": 8.893595695495605,
+      "learning_rate": 5.43358395989975e-06,
+      "loss": 0.4857786297798157,
+      "num_input_tokens_seen": 463082,
+      "step": 1459,
+      "train_runtime": 3909.1257,
+      "train_tokens_per_second": 118.462
+    },
+    {
+      "epoch": 0.06712643678160919,
+      "grad_norm": 8.317085266113281,
+      "learning_rate": 5.423558897243109e-06,
+      "loss": 0.40825778245925903,
+      "num_input_tokens_seen": 463398,
+      "step": 1460,
+      "train_runtime": 3911.0823,
+      "train_tokens_per_second": 118.483
+    },
+    {
+      "epoch": 0.06717241379310344,
+      "grad_norm": 7.586710453033447,
+      "learning_rate": 5.413533834586467e-06,
+      "loss": 0.49957364797592163,
+      "num_input_tokens_seen": 463676,
+      "step": 1461,
+      "train_runtime": 3913.0305,
+      "train_tokens_per_second": 118.495
+    },
+    {
+      "epoch": 0.0672183908045977,
+      "grad_norm": 7.044920444488525,
+      "learning_rate": 5.403508771929825e-06,
+      "loss": 0.39399638772010803,
+      "num_input_tokens_seen": 463938,
+      "step": 1462,
+      "train_runtime": 3914.965,
+      "train_tokens_per_second": 118.504
+    },
+    {
+      "epoch": 0.06726436781609195,
+      "grad_norm": 8.023091316223145,
+      "learning_rate": 5.393483709273183e-06,
+      "loss": 0.4946120083332062,
+      "num_input_tokens_seen": 464396,
+      "step": 1463,
+      "train_runtime": 3917.1293,
+      "train_tokens_per_second": 118.555
+    },
+    {
+      "epoch": 0.0673103448275862,
+      "grad_norm": 8.678879737854004,
+      "learning_rate": 5.383458646616542e-06,
+      "loss": 0.6020040512084961,
+      "num_input_tokens_seen": 464842,
+      "step": 1464,
+      "train_runtime": 3919.1546,
+      "train_tokens_per_second": 118.608
+    },
+    {
+      "epoch": 0.06735632183908045,
+      "grad_norm": 7.435107707977295,
+      "learning_rate": 5.3734335839599006e-06,
+      "loss": 0.5132777094841003,
+      "num_input_tokens_seen": 465214,
+      "step": 1465,
+      "train_runtime": 3921.168,
+      "train_tokens_per_second": 118.642
+    },
+    {
+      "epoch": 0.06740229885057471,
+      "grad_norm": 6.2644195556640625,
+      "learning_rate": 5.363408521303258e-06,
+      "loss": 0.3035534620285034,
+      "num_input_tokens_seen": 465474,
+      "step": 1466,
+      "train_runtime": 3923.1102,
+      "train_tokens_per_second": 118.649
+    },
+    {
+      "epoch": 0.06744827586206896,
+      "grad_norm": 9.018959045410156,
+      "learning_rate": 5.3533834586466175e-06,
+      "loss": 0.5224138498306274,
+      "num_input_tokens_seen": 465714,
+      "step": 1467,
+      "train_runtime": 3925.0802,
+      "train_tokens_per_second": 118.651
+    },
+    {
+      "epoch": 0.06749425287356321,
+      "grad_norm": 6.754763603210449,
+      "learning_rate": 5.343358395989975e-06,
+      "loss": 0.4327360987663269,
+      "num_input_tokens_seen": 465968,
+      "step": 1468,
+      "train_runtime": 3927.0267,
+      "train_tokens_per_second": 118.657
+    },
+    {
+      "epoch": 0.06754022988505747,
+      "grad_norm": 7.545434951782227,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 0.5524900555610657,
+      "num_input_tokens_seen": 466276,
+      "step": 1469,
+      "train_runtime": 3928.9765,
+      "train_tokens_per_second": 118.676
+    },
+    {
+      "epoch": 0.06758620689655172,
+      "grad_norm": 7.580416202545166,
+      "learning_rate": 5.323308270676692e-06,
+      "loss": 0.5965360403060913,
+      "num_input_tokens_seen": 466738,
+      "step": 1470,
+      "train_runtime": 3930.9482,
+      "train_tokens_per_second": 118.734
+    },
+    {
+      "epoch": 0.06763218390804597,
+      "grad_norm": 6.887581825256348,
+      "learning_rate": 5.313283208020051e-06,
+      "loss": 0.33703264594078064,
+      "num_input_tokens_seen": 467014,
+      "step": 1471,
+      "train_runtime": 3946.3969,
+      "train_tokens_per_second": 118.339
+    },
+    {
+      "epoch": 0.06767816091954024,
+      "grad_norm": 8.271141052246094,
+      "learning_rate": 5.303258145363409e-06,
+      "loss": 0.4208733141422272,
+      "num_input_tokens_seen": 467310,
+      "step": 1472,
+      "train_runtime": 3948.3652,
+      "train_tokens_per_second": 118.355
+    },
+    {
+      "epoch": 0.06772413793103449,
+      "grad_norm": 7.072446823120117,
+      "learning_rate": 5.293233082706767e-06,
+      "loss": 0.3233204483985901,
+      "num_input_tokens_seen": 467548,
+      "step": 1473,
+      "train_runtime": 3950.2992,
+      "train_tokens_per_second": 118.358
+    },
+    {
+      "epoch": 0.06777011494252874,
+      "grad_norm": 8.18366527557373,
+      "learning_rate": 5.283208020050126e-06,
+      "loss": 0.4296606779098511,
+      "num_input_tokens_seen": 467876,
+      "step": 1474,
+      "train_runtime": 3952.2591,
+      "train_tokens_per_second": 118.382
+    },
+    {
+      "epoch": 0.067816091954023,
+      "grad_norm": 8.32473087310791,
+      "learning_rate": 5.273182957393485e-06,
+      "loss": 0.38647186756134033,
+      "num_input_tokens_seen": 468136,
+      "step": 1475,
+      "train_runtime": 3954.2163,
+      "train_tokens_per_second": 118.389
+    },
+    {
+      "epoch": 0.06786206896551725,
+      "grad_norm": 7.680408000946045,
+      "learning_rate": 5.263157894736842e-06,
+      "loss": 0.4326077401638031,
+      "num_input_tokens_seen": 468392,
+      "step": 1476,
+      "train_runtime": 3956.1621,
+      "train_tokens_per_second": 118.396
+    },
+    {
+      "epoch": 0.0679080459770115,
+      "grad_norm": 6.291998386383057,
+      "learning_rate": 5.253132832080201e-06,
+      "loss": 0.379532128572464,
+      "num_input_tokens_seen": 468698,
+      "step": 1477,
+      "train_runtime": 3958.1197,
+      "train_tokens_per_second": 118.414
+    },
+    {
+      "epoch": 0.06795402298850575,
+      "grad_norm": 8.708852767944336,
+      "learning_rate": 5.243107769423559e-06,
+      "loss": 0.4702181816101074,
+      "num_input_tokens_seen": 469026,
+      "step": 1478,
+      "train_runtime": 3960.0793,
+      "train_tokens_per_second": 118.439
+    },
+    {
+      "epoch": 0.068,
+      "grad_norm": 7.804077625274658,
+      "learning_rate": 5.233082706766918e-06,
+      "loss": 0.406497061252594,
+      "num_input_tokens_seen": 469274,
+      "step": 1479,
+      "train_runtime": 3962.0364,
+      "train_tokens_per_second": 118.443
+    },
+    {
+      "epoch": 0.06804597701149426,
+      "grad_norm": 10.114439010620117,
+      "learning_rate": 5.223057644110276e-06,
+      "loss": 0.5118438601493835,
+      "num_input_tokens_seen": 469602,
+      "step": 1480,
+      "train_runtime": 3963.9901,
+      "train_tokens_per_second": 118.467
+    },
+    {
+      "epoch": 0.06809195402298851,
+      "grad_norm": 9.554535865783691,
+      "learning_rate": 5.213032581453634e-06,
+      "loss": 0.521860659122467,
+      "num_input_tokens_seen": 469924,
+      "step": 1481,
+      "train_runtime": 3965.9423,
+      "train_tokens_per_second": 118.49
+    },
+    {
+      "epoch": 0.06813793103448276,
+      "grad_norm": 10.086437225341797,
+      "learning_rate": 5.203007518796993e-06,
+      "loss": 0.5799345970153809,
+      "num_input_tokens_seen": 470244,
+      "step": 1482,
+      "train_runtime": 3967.8889,
+      "train_tokens_per_second": 118.512
+    },
+    {
+      "epoch": 0.06818390804597702,
+      "grad_norm": 8.745047569274902,
+      "learning_rate": 5.192982456140351e-06,
+      "loss": 0.4893014430999756,
+      "num_input_tokens_seen": 470484,
+      "step": 1483,
+      "train_runtime": 3969.8344,
+      "train_tokens_per_second": 118.515
+    },
+    {
+      "epoch": 0.06822988505747127,
+      "grad_norm": 8.214205741882324,
+      "learning_rate": 5.182957393483709e-06,
+      "loss": 0.3816409707069397,
+      "num_input_tokens_seen": 470798,
+      "step": 1484,
+      "train_runtime": 3971.825,
+      "train_tokens_per_second": 118.534
+    },
+    {
+      "epoch": 0.06827586206896552,
+      "grad_norm": 6.99646520614624,
+      "learning_rate": 5.172932330827069e-06,
+      "loss": 0.5191115140914917,
+      "num_input_tokens_seen": 471250,
+      "step": 1485,
+      "train_runtime": 3973.8055,
+      "train_tokens_per_second": 118.589
+    },
+    {
+      "epoch": 0.06832183908045977,
+      "grad_norm": 7.990946292877197,
+      "learning_rate": 5.162907268170426e-06,
+      "loss": 0.49176281690597534,
+      "num_input_tokens_seen": 471494,
+      "step": 1486,
+      "train_runtime": 3975.7526,
+      "train_tokens_per_second": 118.592
+    },
+    {
+      "epoch": 0.06836781609195403,
+      "grad_norm": 6.933746814727783,
+      "learning_rate": 5.152882205513785e-06,
+      "loss": 0.3267862796783447,
+      "num_input_tokens_seen": 471750,
+      "step": 1487,
+      "train_runtime": 3977.7152,
+      "train_tokens_per_second": 118.598
+    },
+    {
+      "epoch": 0.06841379310344828,
+      "grad_norm": 9.269936561584473,
+      "learning_rate": 5.142857142857142e-06,
+      "loss": 0.5424149036407471,
+      "num_input_tokens_seen": 472076,
+      "step": 1488,
+      "train_runtime": 3979.7152,
+      "train_tokens_per_second": 118.621
+    },
+    {
+      "epoch": 0.06845977011494253,
+      "grad_norm": 6.7744035720825195,
+      "learning_rate": 5.132832080200502e-06,
+      "loss": 0.3430073857307434,
+      "num_input_tokens_seen": 472364,
+      "step": 1489,
+      "train_runtime": 3981.6783,
+      "train_tokens_per_second": 118.634
+    },
+    {
+      "epoch": 0.06850574712643678,
+      "grad_norm": 7.776864051818848,
+      "learning_rate": 5.12280701754386e-06,
+      "loss": 0.3698754608631134,
+      "num_input_tokens_seen": 472624,
+      "step": 1490,
+      "train_runtime": 3983.6629,
+      "train_tokens_per_second": 118.641
+    },
+    {
+      "epoch": 0.06855172413793104,
+      "grad_norm": 7.676121234893799,
+      "learning_rate": 5.112781954887218e-06,
+      "loss": 0.5057390928268433,
+      "num_input_tokens_seen": 473178,
+      "step": 1491,
+      "train_runtime": 3985.7442,
+      "train_tokens_per_second": 118.718
+    },
+    {
+      "epoch": 0.06859770114942529,
+      "grad_norm": 6.1846418380737305,
+      "learning_rate": 5.102756892230577e-06,
+      "loss": 0.3319104313850403,
+      "num_input_tokens_seen": 473548,
+      "step": 1492,
+      "train_runtime": 3987.727,
+      "train_tokens_per_second": 118.751
+    },
+    {
+      "epoch": 0.06864367816091954,
+      "grad_norm": 8.331839561462402,
+      "learning_rate": 5.092731829573935e-06,
+      "loss": 0.6448594927787781,
+      "num_input_tokens_seen": 473872,
+      "step": 1493,
+      "train_runtime": 3989.7313,
+      "train_tokens_per_second": 118.773
+    },
+    {
+      "epoch": 0.0686896551724138,
+      "grad_norm": 9.190635681152344,
+      "learning_rate": 5.082706766917293e-06,
+      "loss": 0.5440618991851807,
+      "num_input_tokens_seen": 474106,
+      "step": 1494,
+      "train_runtime": 3991.6556,
+      "train_tokens_per_second": 118.774
+    },
+    {
+      "epoch": 0.06873563218390805,
+      "grad_norm": 10.09354305267334,
+      "learning_rate": 5.072681704260653e-06,
+      "loss": 0.34777310490608215,
+      "num_input_tokens_seen": 474378,
+      "step": 1495,
+      "train_runtime": 3993.656,
+      "train_tokens_per_second": 118.783
+    },
+    {
+      "epoch": 0.0687816091954023,
+      "grad_norm": 7.709604740142822,
+      "learning_rate": 5.06265664160401e-06,
+      "loss": 0.4458523988723755,
+      "num_input_tokens_seen": 474652,
+      "step": 1496,
+      "train_runtime": 3995.6046,
+      "train_tokens_per_second": 118.794
+    },
+    {
+      "epoch": 0.06882758620689655,
+      "grad_norm": 6.953578472137451,
+      "learning_rate": 5.052631578947369e-06,
+      "loss": 0.40357089042663574,
+      "num_input_tokens_seen": 475020,
+      "step": 1497,
+      "train_runtime": 3997.676,
+      "train_tokens_per_second": 118.824
+    },
+    {
+      "epoch": 0.0688735632183908,
+      "grad_norm": 8.717723846435547,
+      "learning_rate": 5.042606516290726e-06,
+      "loss": 0.5088080763816833,
+      "num_input_tokens_seen": 475368,
+      "step": 1498,
+      "train_runtime": 3999.6783,
+      "train_tokens_per_second": 118.852
+    },
+    {
+      "epoch": 0.06891954022988506,
+      "grad_norm": 6.654493808746338,
+      "learning_rate": 5.032581453634086e-06,
+      "loss": 0.5015754699707031,
+      "num_input_tokens_seen": 475868,
+      "step": 1499,
+      "train_runtime": 4001.7448,
+      "train_tokens_per_second": 118.915
+    },
+    {
+      "epoch": 0.06896551724137931,
+      "grad_norm": 7.362853050231934,
+      "learning_rate": 5.022556390977444e-06,
+      "loss": 0.46414509415626526,
+      "num_input_tokens_seen": 476138,
+      "step": 1500,
+      "train_runtime": 4003.7518,
+      "train_tokens_per_second": 118.923
+    },
+    {
+      "epoch": 0.06901149425287356,
+      "grad_norm": 7.295174598693848,
+      "learning_rate": 5.012531328320802e-06,
+      "loss": 0.44448113441467285,
+      "num_input_tokens_seen": 476464,
+      "step": 1501,
+      "train_runtime": 4019.5965,
+      "train_tokens_per_second": 118.535
+    },
+    {
+      "epoch": 0.06905747126436781,
+      "grad_norm": 8.375143051147461,
+      "learning_rate": 5.002506265664161e-06,
+      "loss": 0.395553320646286,
+      "num_input_tokens_seen": 476832,
+      "step": 1502,
+      "train_runtime": 4021.7011,
+      "train_tokens_per_second": 118.565
+    },
+    {
+      "epoch": 0.06910344827586207,
+      "grad_norm": 9.805458068847656,
+      "learning_rate": 4.992481203007519e-06,
+      "loss": 0.42004987597465515,
+      "num_input_tokens_seen": 477098,
+      "step": 1503,
+      "train_runtime": 4023.7052,
+      "train_tokens_per_second": 118.572
+    },
+    {
+      "epoch": 0.06914942528735632,
+      "grad_norm": 8.395523071289062,
+      "learning_rate": 4.982456140350877e-06,
+      "loss": 0.5958109498023987,
+      "num_input_tokens_seen": 477516,
+      "step": 1504,
+      "train_runtime": 4025.7994,
+      "train_tokens_per_second": 118.614
+    },
+    {
+      "epoch": 0.06919540229885057,
+      "grad_norm": 8.359265327453613,
+      "learning_rate": 4.972431077694236e-06,
+      "loss": 0.4998864531517029,
+      "num_input_tokens_seen": 477846,
+      "step": 1505,
+      "train_runtime": 4027.7803,
+      "train_tokens_per_second": 118.638
+    },
+    {
+      "epoch": 0.06924137931034482,
+      "grad_norm": 6.486353397369385,
+      "learning_rate": 4.962406015037594e-06,
+      "loss": 0.3851095139980316,
+      "num_input_tokens_seen": 478138,
+      "step": 1506,
+      "train_runtime": 4029.8107,
+      "train_tokens_per_second": 118.65
+    },
+    {
+      "epoch": 0.06928735632183908,
+      "grad_norm": 6.7579803466796875,
+      "learning_rate": 4.952380952380953e-06,
+      "loss": 0.355318158864975,
+      "num_input_tokens_seen": 478414,
+      "step": 1507,
+      "train_runtime": 4031.8359,
+      "train_tokens_per_second": 118.659
+    },
+    {
+      "epoch": 0.06933333333333333,
+      "grad_norm": 12.411815643310547,
+      "learning_rate": 4.942355889724311e-06,
+      "loss": 0.42763108015060425,
+      "num_input_tokens_seen": 478646,
+      "step": 1508,
+      "train_runtime": 4033.7831,
+      "train_tokens_per_second": 118.659
+    },
+    {
+      "epoch": 0.06937931034482758,
+      "grad_norm": 9.093607902526855,
+      "learning_rate": 4.93233082706767e-06,
+      "loss": 0.4881199598312378,
+      "num_input_tokens_seen": 479012,
+      "step": 1509,
+      "train_runtime": 4035.8165,
+      "train_tokens_per_second": 118.69
+    },
+    {
+      "epoch": 0.06942528735632184,
+      "grad_norm": 8.364423751831055,
+      "learning_rate": 4.922305764411027e-06,
+      "loss": 0.45255759358406067,
+      "num_input_tokens_seen": 479384,
+      "step": 1510,
+      "train_runtime": 4037.8086,
+      "train_tokens_per_second": 118.724
+    },
+    {
+      "epoch": 0.06947126436781609,
+      "grad_norm": 7.509194374084473,
+      "learning_rate": 4.912280701754386e-06,
+      "loss": 0.5315737128257751,
+      "num_input_tokens_seen": 479778,
+      "step": 1511,
+      "train_runtime": 4039.8268,
+      "train_tokens_per_second": 118.762
+    },
+    {
+      "epoch": 0.06951724137931034,
+      "grad_norm": 7.366269588470459,
+      "learning_rate": 4.902255639097745e-06,
+      "loss": 0.4919632077217102,
+      "num_input_tokens_seen": 480118,
+      "step": 1512,
+      "train_runtime": 4041.7992,
+      "train_tokens_per_second": 118.788
+    },
+    {
+      "epoch": 0.0695632183908046,
+      "grad_norm": 7.457450866699219,
+      "learning_rate": 4.892230576441103e-06,
+      "loss": 0.37449610233306885,
+      "num_input_tokens_seen": 480422,
+      "step": 1513,
+      "train_runtime": 4043.7798,
+      "train_tokens_per_second": 118.805
+    },
+    {
+      "epoch": 0.06960919540229885,
+      "grad_norm": 7.189851760864258,
+      "learning_rate": 4.882205513784461e-06,
+      "loss": 0.456065833568573,
+      "num_input_tokens_seen": 480710,
+      "step": 1514,
+      "train_runtime": 4045.8441,
+      "train_tokens_per_second": 118.816
+    },
+    {
+      "epoch": 0.0696551724137931,
+      "grad_norm": 7.268008708953857,
+      "learning_rate": 4.87218045112782e-06,
+      "loss": 0.4126277565956116,
+      "num_input_tokens_seen": 480948,
+      "step": 1515,
+      "train_runtime": 4047.8221,
+      "train_tokens_per_second": 118.816
+    },
+    {
+      "epoch": 0.06970114942528735,
+      "grad_norm": 6.931264877319336,
+      "learning_rate": 4.862155388471178e-06,
+      "loss": 0.35446545481681824,
+      "num_input_tokens_seen": 481306,
+      "step": 1516,
+      "train_runtime": 4049.8866,
+      "train_tokens_per_second": 118.844
+    },
+    {
+      "epoch": 0.0697471264367816,
+      "grad_norm": 9.292113304138184,
+      "learning_rate": 4.852130325814537e-06,
+      "loss": 0.34105855226516724,
+      "num_input_tokens_seen": 481570,
+      "step": 1517,
+      "train_runtime": 4051.9034,
+      "train_tokens_per_second": 118.85
+    },
+    {
+      "epoch": 0.06979310344827586,
+      "grad_norm": 8.461747169494629,
+      "learning_rate": 4.842105263157895e-06,
+      "loss": 0.5056231021881104,
+      "num_input_tokens_seen": 481892,
+      "step": 1518,
+      "train_runtime": 4053.8728,
+      "train_tokens_per_second": 118.872
+    },
+    {
+      "epoch": 0.06983908045977011,
+      "grad_norm": 7.729432582855225,
+      "learning_rate": 4.832080200501254e-06,
+      "loss": 0.5439703464508057,
+      "num_input_tokens_seen": 482198,
+      "step": 1519,
+      "train_runtime": 4055.8291,
+      "train_tokens_per_second": 118.89
+    },
+    {
+      "epoch": 0.06988505747126436,
+      "grad_norm": 7.601734161376953,
+      "learning_rate": 4.8220551378446114e-06,
+      "loss": 0.46440935134887695,
+      "num_input_tokens_seen": 482550,
+      "step": 1520,
+      "train_runtime": 4057.8424,
+      "train_tokens_per_second": 118.918
+    },
+    {
+      "epoch": 0.06993103448275861,
+      "grad_norm": 7.580126762390137,
+      "learning_rate": 4.81203007518797e-06,
+      "loss": 0.4427540898323059,
+      "num_input_tokens_seen": 482938,
+      "step": 1521,
+      "train_runtime": 4059.9159,
+      "train_tokens_per_second": 118.953
+    },
+    {
+      "epoch": 0.06997701149425288,
+      "grad_norm": 9.245956420898438,
+      "learning_rate": 4.802005012531329e-06,
+      "loss": 0.2203454077243805,
+      "num_input_tokens_seen": 483178,
+      "step": 1522,
+      "train_runtime": 4061.8498,
+      "train_tokens_per_second": 118.955
+    },
+    {
+      "epoch": 0.07002298850574713,
+      "grad_norm": 24.038677215576172,
+      "learning_rate": 4.791979949874687e-06,
+      "loss": 0.3537423014640808,
+      "num_input_tokens_seen": 483402,
+      "step": 1523,
+      "train_runtime": 4063.7902,
+      "train_tokens_per_second": 118.953
+    },
+    {
+      "epoch": 0.07006896551724139,
+      "grad_norm": 8.508315086364746,
+      "learning_rate": 4.781954887218045e-06,
+      "loss": 0.5330834984779358,
+      "num_input_tokens_seen": 483802,
+      "step": 1524,
+      "train_runtime": 4065.7865,
+      "train_tokens_per_second": 118.993
+    },
+    {
+      "epoch": 0.07011494252873564,
+      "grad_norm": 12.299150466918945,
+      "learning_rate": 4.771929824561404e-06,
+      "loss": 0.5105612277984619,
+      "num_input_tokens_seen": 484070,
+      "step": 1525,
+      "train_runtime": 4067.7933,
+      "train_tokens_per_second": 119.001
+    },
+    {
+      "epoch": 0.07016091954022989,
+      "grad_norm": 6.736196041107178,
+      "learning_rate": 4.761904761904762e-06,
+      "loss": 0.3398650586605072,
+      "num_input_tokens_seen": 484344,
+      "step": 1526,
+      "train_runtime": 4069.7464,
+      "train_tokens_per_second": 119.011
+    },
+    {
+      "epoch": 0.07020689655172414,
+      "grad_norm": 5.740973949432373,
+      "learning_rate": 4.751879699248121e-06,
+      "loss": 0.2813899517059326,
+      "num_input_tokens_seen": 484544,
+      "step": 1527,
+      "train_runtime": 4071.7248,
+      "train_tokens_per_second": 119.002
+    },
+    {
+      "epoch": 0.0702528735632184,
+      "grad_norm": 7.960937023162842,
+      "learning_rate": 4.741854636591479e-06,
+      "loss": 0.4507075250148773,
+      "num_input_tokens_seen": 484866,
+      "step": 1528,
+      "train_runtime": 4073.7501,
+      "train_tokens_per_second": 119.022
+    },
+    {
+      "epoch": 0.07029885057471265,
+      "grad_norm": 7.307352066040039,
+      "learning_rate": 4.731829573934838e-06,
+      "loss": 0.5230740308761597,
+      "num_input_tokens_seen": 485206,
+      "step": 1529,
+      "train_runtime": 4075.8439,
+      "train_tokens_per_second": 119.044
+    },
+    {
+      "epoch": 0.0703448275862069,
+      "grad_norm": 6.8969244956970215,
+      "learning_rate": 4.7218045112781955e-06,
+      "loss": 0.3683476448059082,
+      "num_input_tokens_seen": 485502,
+      "step": 1530,
+      "train_runtime": 4077.7995,
+      "train_tokens_per_second": 119.06
+    },
+    {
+      "epoch": 0.07039080459770115,
+      "grad_norm": 6.411011219024658,
+      "learning_rate": 4.711779448621554e-06,
+      "loss": 0.41447699069976807,
+      "num_input_tokens_seen": 485962,
+      "step": 1531,
+      "train_runtime": 4092.7217,
+      "train_tokens_per_second": 118.738
+    },
+    {
+      "epoch": 0.0704367816091954,
+      "grad_norm": 8.083466529846191,
+      "learning_rate": 4.7017543859649125e-06,
+      "loss": 0.5244938731193542,
+      "num_input_tokens_seen": 486228,
+      "step": 1532,
+      "train_runtime": 4094.744,
+      "train_tokens_per_second": 118.744
+    },
+    {
+      "epoch": 0.07048275862068966,
+      "grad_norm": 7.008942127227783,
+      "learning_rate": 4.691729323308271e-06,
+      "loss": 0.4396969974040985,
+      "num_input_tokens_seen": 486514,
+      "step": 1533,
+      "train_runtime": 4096.7295,
+      "train_tokens_per_second": 118.757
+    },
+    {
+      "epoch": 0.07052873563218391,
+      "grad_norm": 9.5209379196167,
+      "learning_rate": 4.6817042606516294e-06,
+      "loss": 0.38825932145118713,
+      "num_input_tokens_seen": 486760,
+      "step": 1534,
+      "train_runtime": 4098.7097,
+      "train_tokens_per_second": 118.759
+    },
+    {
+      "epoch": 0.07057471264367816,
+      "grad_norm": 6.287064075469971,
+      "learning_rate": 4.671679197994988e-06,
+      "loss": 0.33252447843551636,
+      "num_input_tokens_seen": 487050,
+      "step": 1535,
+      "train_runtime": 4100.8288,
+      "train_tokens_per_second": 118.769
+    },
+    {
+      "epoch": 0.07062068965517242,
+      "grad_norm": 7.780177116394043,
+      "learning_rate": 4.661654135338346e-06,
+      "loss": 0.40082573890686035,
+      "num_input_tokens_seen": 487384,
+      "step": 1536,
+      "train_runtime": 4103.0639,
+      "train_tokens_per_second": 118.785
+    },
+    {
+      "epoch": 0.07066666666666667,
+      "grad_norm": 6.677377700805664,
+      "learning_rate": 4.651629072681705e-06,
+      "loss": 0.4884091019630432,
+      "num_input_tokens_seen": 488052,
+      "step": 1537,
+      "train_runtime": 4105.1073,
+      "train_tokens_per_second": 118.889
+    },
+    {
+      "epoch": 0.07071264367816092,
+      "grad_norm": 10.25025463104248,
+      "learning_rate": 4.641604010025063e-06,
+      "loss": 0.3952670991420746,
+      "num_input_tokens_seen": 488380,
+      "step": 1538,
+      "train_runtime": 4107.0889,
+      "train_tokens_per_second": 118.911
+    },
+    {
+      "epoch": 0.07075862068965517,
+      "grad_norm": 8.572774887084961,
+      "learning_rate": 4.631578947368421e-06,
+      "loss": 0.7552651166915894,
+      "num_input_tokens_seen": 488652,
+      "step": 1539,
+      "train_runtime": 4109.0954,
+      "train_tokens_per_second": 118.92
+    },
+    {
+      "epoch": 0.07080459770114943,
+      "grad_norm": 7.492312431335449,
+      "learning_rate": 4.6215538847117795e-06,
+      "loss": 0.5208468437194824,
+      "num_input_tokens_seen": 488996,
+      "step": 1540,
+      "train_runtime": 4111.1038,
+      "train_tokens_per_second": 118.945
+    },
+    {
+      "epoch": 0.07085057471264368,
+      "grad_norm": 6.831908226013184,
+      "learning_rate": 4.611528822055138e-06,
+      "loss": 0.4159829914569855,
+      "num_input_tokens_seen": 489436,
+      "step": 1541,
+      "train_runtime": 4113.081,
+      "train_tokens_per_second": 118.995
+    },
+    {
+      "epoch": 0.07089655172413793,
+      "grad_norm": 6.706018447875977,
+      "learning_rate": 4.6015037593984965e-06,
+      "loss": 0.42937764525413513,
+      "num_input_tokens_seen": 489840,
+      "step": 1542,
+      "train_runtime": 4115.2212,
+      "train_tokens_per_second": 119.031
+    },
+    {
+      "epoch": 0.07094252873563218,
+      "grad_norm": 6.830271244049072,
+      "learning_rate": 4.591478696741855e-06,
+      "loss": 0.31548362970352173,
+      "num_input_tokens_seen": 490070,
+      "step": 1543,
+      "train_runtime": 4117.2331,
+      "train_tokens_per_second": 119.029
+    },
+    {
+      "epoch": 0.07098850574712644,
+      "grad_norm": 6.314537048339844,
+      "learning_rate": 4.5814536340852135e-06,
+      "loss": 0.3962181508541107,
+      "num_input_tokens_seen": 490426,
+      "step": 1544,
+      "train_runtime": 4119.2344,
+      "train_tokens_per_second": 119.058
+    },
+    {
+      "epoch": 0.07103448275862069,
+      "grad_norm": 7.005821704864502,
+      "learning_rate": 4.571428571428572e-06,
+      "loss": 0.35218262672424316,
+      "num_input_tokens_seen": 490662,
+      "step": 1545,
+      "train_runtime": 4121.1795,
+      "train_tokens_per_second": 119.059
+    },
+    {
+      "epoch": 0.07108045977011494,
+      "grad_norm": 7.298124313354492,
+      "learning_rate": 4.56140350877193e-06,
+      "loss": 0.40392208099365234,
+      "num_input_tokens_seen": 490972,
+      "step": 1546,
+      "train_runtime": 4123.1469,
+      "train_tokens_per_second": 119.077
+    },
+    {
+      "epoch": 0.0711264367816092,
+      "grad_norm": 7.718113899230957,
+      "learning_rate": 4.551378446115289e-06,
+      "loss": 0.4594915807247162,
+      "num_input_tokens_seen": 491268,
+      "step": 1547,
+      "train_runtime": 4125.1073,
+      "train_tokens_per_second": 119.092
+    },
+    {
+      "epoch": 0.07117241379310345,
+      "grad_norm": 8.686179161071777,
+      "learning_rate": 4.541353383458647e-06,
+      "loss": 0.4476366937160492,
+      "num_input_tokens_seen": 491532,
+      "step": 1548,
+      "train_runtime": 4127.073,
+      "train_tokens_per_second": 119.099
+    },
+    {
+      "epoch": 0.0712183908045977,
+      "grad_norm": 8.165818214416504,
+      "learning_rate": 4.531328320802005e-06,
+      "loss": 0.3883603811264038,
+      "num_input_tokens_seen": 491782,
+      "step": 1549,
+      "train_runtime": 4129.0348,
+      "train_tokens_per_second": 119.103
+    },
+    {
+      "epoch": 0.07126436781609195,
+      "grad_norm": 7.7750701904296875,
+      "learning_rate": 4.5213032581453635e-06,
+      "loss": 0.49869048595428467,
+      "num_input_tokens_seen": 492204,
+      "step": 1550,
+      "train_runtime": 4131.1119,
+      "train_tokens_per_second": 119.146
+    },
+    {
+      "epoch": 0.0713103448275862,
+      "grad_norm": 7.940079689025879,
+      "learning_rate": 4.511278195488722e-06,
+      "loss": 0.4624294340610504,
+      "num_input_tokens_seen": 492568,
+      "step": 1551,
+      "train_runtime": 4133.0921,
+      "train_tokens_per_second": 119.177
+    },
+    {
+      "epoch": 0.07135632183908046,
+      "grad_norm": 8.412528038024902,
+      "learning_rate": 4.5012531328320805e-06,
+      "loss": 0.44974592328071594,
+      "num_input_tokens_seen": 492824,
+      "step": 1552,
+      "train_runtime": 4135.0591,
+      "train_tokens_per_second": 119.182
+    },
+    {
+      "epoch": 0.07140229885057471,
+      "grad_norm": 6.750128746032715,
+      "learning_rate": 4.491228070175439e-06,
+      "loss": 0.26358163356781006,
+      "num_input_tokens_seen": 493048,
+      "step": 1553,
+      "train_runtime": 4136.9973,
+      "train_tokens_per_second": 119.18
+    },
+    {
+      "epoch": 0.07144827586206896,
+      "grad_norm": 6.663267135620117,
+      "learning_rate": 4.4812030075187975e-06,
+      "loss": 0.3788079023361206,
+      "num_input_tokens_seen": 493430,
+      "step": 1554,
+      "train_runtime": 4138.9818,
+      "train_tokens_per_second": 119.215
+    },
+    {
+      "epoch": 0.07149425287356322,
+      "grad_norm": 7.058904647827148,
+      "learning_rate": 4.471177944862156e-06,
+      "loss": 0.42675814032554626,
+      "num_input_tokens_seen": 493704,
+      "step": 1555,
+      "train_runtime": 4140.9285,
+      "train_tokens_per_second": 119.225
+    },
+    {
+      "epoch": 0.07154022988505747,
+      "grad_norm": 8.06811237335205,
+      "learning_rate": 4.461152882205514e-06,
+      "loss": 0.5106055736541748,
+      "num_input_tokens_seen": 494120,
+      "step": 1556,
+      "train_runtime": 4142.8977,
+      "train_tokens_per_second": 119.269
+    },
+    {
+      "epoch": 0.07158620689655172,
+      "grad_norm": 8.179754257202148,
+      "learning_rate": 4.451127819548873e-06,
+      "loss": 0.3482970893383026,
+      "num_input_tokens_seen": 494372,
+      "step": 1557,
+      "train_runtime": 4144.8359,
+      "train_tokens_per_second": 119.274
+    },
+    {
+      "epoch": 0.07163218390804597,
+      "grad_norm": 8.394057273864746,
+      "learning_rate": 4.4411027568922314e-06,
+      "loss": 0.5103639960289001,
+      "num_input_tokens_seen": 494650,
+      "step": 1558,
+      "train_runtime": 4146.7879,
+      "train_tokens_per_second": 119.285
+    },
+    {
+      "epoch": 0.07167816091954023,
+      "grad_norm": 7.490317344665527,
+      "learning_rate": 4.431077694235589e-06,
+      "loss": 0.3746236264705658,
+      "num_input_tokens_seen": 494908,
+      "step": 1559,
+      "train_runtime": 4148.7421,
+      "train_tokens_per_second": 119.291
+    },
+    {
+      "epoch": 0.07172413793103448,
+      "grad_norm": 9.055501937866211,
+      "learning_rate": 4.4210526315789476e-06,
+      "loss": 0.39659398794174194,
+      "num_input_tokens_seen": 495170,
+      "step": 1560,
+      "train_runtime": 4150.6871,
+      "train_tokens_per_second": 119.298
+    },
+    {
+      "epoch": 0.07177011494252873,
+      "grad_norm": 8.413995742797852,
+      "learning_rate": 4.411027568922306e-06,
+      "loss": 0.48551711440086365,
+      "num_input_tokens_seen": 495452,
+      "step": 1561,
+      "train_runtime": 4165.543,
+      "train_tokens_per_second": 118.941
+    },
+    {
+      "epoch": 0.07181609195402298,
+      "grad_norm": 7.30216121673584,
+      "learning_rate": 4.4010025062656645e-06,
+      "loss": 0.47951072454452515,
+      "num_input_tokens_seen": 495762,
+      "step": 1562,
+      "train_runtime": 4167.6294,
+      "train_tokens_per_second": 118.955
+    },
+    {
+      "epoch": 0.07186206896551724,
+      "grad_norm": 8.080525398254395,
+      "learning_rate": 4.390977443609023e-06,
+      "loss": 0.31800565123558044,
+      "num_input_tokens_seen": 495974,
+      "step": 1563,
+      "train_runtime": 4169.5551,
+      "train_tokens_per_second": 118.951
+    },
+    {
+      "epoch": 0.07190804597701149,
+      "grad_norm": 7.3906354904174805,
+      "learning_rate": 4.3809523809523815e-06,
+      "loss": 0.4231320023536682,
+      "num_input_tokens_seen": 496294,
+      "step": 1564,
+      "train_runtime": 4171.5046,
+      "train_tokens_per_second": 118.972
+    },
+    {
+      "epoch": 0.07195402298850574,
+      "grad_norm": 6.36199951171875,
+      "learning_rate": 4.370927318295739e-06,
+      "loss": 0.36958885192871094,
+      "num_input_tokens_seen": 496808,
+      "step": 1565,
+      "train_runtime": 4173.5072,
+      "train_tokens_per_second": 119.038
+    },
+    {
+      "epoch": 0.072,
+      "grad_norm": 7.268085956573486,
+      "learning_rate": 4.360902255639098e-06,
+      "loss": 0.34520044922828674,
+      "num_input_tokens_seen": 497078,
+      "step": 1566,
+      "train_runtime": 4175.4481,
+      "train_tokens_per_second": 119.048
+    },
+    {
+      "epoch": 0.07204597701149425,
+      "grad_norm": 6.480217933654785,
+      "learning_rate": 4.350877192982457e-06,
+      "loss": 0.3157622218132019,
+      "num_input_tokens_seen": 497356,
+      "step": 1567,
+      "train_runtime": 4177.4065,
+      "train_tokens_per_second": 119.059
+    },
+    {
+      "epoch": 0.0720919540229885,
+      "grad_norm": 8.270898818969727,
+      "learning_rate": 4.340852130325815e-06,
+      "loss": 0.5205401182174683,
+      "num_input_tokens_seen": 497640,
+      "step": 1568,
+      "train_runtime": 4179.3743,
+      "train_tokens_per_second": 119.07
+    },
+    {
+      "epoch": 0.07213793103448275,
+      "grad_norm": 6.635157108306885,
+      "learning_rate": 4.330827067669173e-06,
+      "loss": 0.449776828289032,
+      "num_input_tokens_seen": 498084,
+      "step": 1569,
+      "train_runtime": 4181.3625,
+      "train_tokens_per_second": 119.12
+    },
+    {
+      "epoch": 0.072183908045977,
+      "grad_norm": 7.485076904296875,
+      "learning_rate": 4.320802005012532e-06,
+      "loss": 0.5208253860473633,
+      "num_input_tokens_seen": 498412,
+      "step": 1570,
+      "train_runtime": 4183.3179,
+      "train_tokens_per_second": 119.143
+    },
+    {
+      "epoch": 0.07222988505747126,
+      "grad_norm": 7.9057841300964355,
+      "learning_rate": 4.31077694235589e-06,
+      "loss": 0.4355557858943939,
+      "num_input_tokens_seen": 498670,
+      "step": 1571,
+      "train_runtime": 4185.2576,
+      "train_tokens_per_second": 119.149
+    },
+    {
+      "epoch": 0.07227586206896552,
+      "grad_norm": 7.838947296142578,
+      "learning_rate": 4.3007518796992486e-06,
+      "loss": 0.42538976669311523,
+      "num_input_tokens_seen": 498932,
+      "step": 1572,
+      "train_runtime": 4187.2155,
+      "train_tokens_per_second": 119.156
+    },
+    {
+      "epoch": 0.07232183908045978,
+      "grad_norm": 7.784214496612549,
+      "learning_rate": 4.290726817042607e-06,
+      "loss": 0.4852614402770996,
+      "num_input_tokens_seen": 499206,
+      "step": 1573,
+      "train_runtime": 4189.2225,
+      "train_tokens_per_second": 119.164
+    },
+    {
+      "epoch": 0.07236781609195403,
+      "grad_norm": 7.158159255981445,
+      "learning_rate": 4.2807017543859656e-06,
+      "loss": 0.3924868702888489,
+      "num_input_tokens_seen": 499530,
+      "step": 1574,
+      "train_runtime": 4191.2055,
+      "train_tokens_per_second": 119.185
+    },
+    {
+      "epoch": 0.07241379310344828,
+      "grad_norm": 8.813111305236816,
+      "learning_rate": 4.270676691729323e-06,
+      "loss": 0.560117244720459,
+      "num_input_tokens_seen": 499810,
+      "step": 1575,
+      "train_runtime": 4193.2386,
+      "train_tokens_per_second": 119.194
+    },
+    {
+      "epoch": 0.07245977011494253,
+      "grad_norm": 8.600942611694336,
+      "learning_rate": 4.260651629072682e-06,
+      "loss": 0.3756157159805298,
+      "num_input_tokens_seen": 500204,
+      "step": 1576,
+      "train_runtime": 4195.2333,
+      "train_tokens_per_second": 119.232
+    },
+    {
+      "epoch": 0.07250574712643679,
+      "grad_norm": 8.5283203125,
+      "learning_rate": 4.250626566416041e-06,
+      "loss": 0.38206595182418823,
+      "num_input_tokens_seen": 500500,
+      "step": 1577,
+      "train_runtime": 4197.236,
+      "train_tokens_per_second": 119.245
+    },
+    {
+      "epoch": 0.07255172413793104,
+      "grad_norm": 7.617363452911377,
+      "learning_rate": 4.240601503759399e-06,
+      "loss": 0.41842153668403625,
+      "num_input_tokens_seen": 500790,
+      "step": 1578,
+      "train_runtime": 4199.2371,
+      "train_tokens_per_second": 119.257
+    },
+    {
+      "epoch": 0.07259770114942529,
+      "grad_norm": 8.498720169067383,
+      "learning_rate": 4.230576441102757e-06,
+      "loss": 0.5772629380226135,
+      "num_input_tokens_seen": 501118,
+      "step": 1579,
+      "train_runtime": 4201.1981,
+      "train_tokens_per_second": 119.28
+    },
+    {
+      "epoch": 0.07264367816091954,
+      "grad_norm": 6.475655555725098,
+      "learning_rate": 4.220551378446116e-06,
+      "loss": 0.27916771173477173,
+      "num_input_tokens_seen": 501364,
+      "step": 1580,
+      "train_runtime": 4203.1483,
+      "train_tokens_per_second": 119.283
+    },
+    {
+      "epoch": 0.0726896551724138,
+      "grad_norm": 7.510998725891113,
+      "learning_rate": 4.210526315789474e-06,
+      "loss": 0.41097086668014526,
+      "num_input_tokens_seen": 501670,
+      "step": 1581,
+      "train_runtime": 4205.1098,
+      "train_tokens_per_second": 119.3
+    },
+    {
+      "epoch": 0.07273563218390805,
+      "grad_norm": 7.724368095397949,
+      "learning_rate": 4.200501253132833e-06,
+      "loss": 0.33816516399383545,
+      "num_input_tokens_seen": 501914,
+      "step": 1582,
+      "train_runtime": 4207.054,
+      "train_tokens_per_second": 119.303
+    },
+    {
+      "epoch": 0.0727816091954023,
+      "grad_norm": 8.160489082336426,
+      "learning_rate": 4.190476190476191e-06,
+      "loss": 0.38364431262016296,
+      "num_input_tokens_seen": 502148,
+      "step": 1583,
+      "train_runtime": 4208.99,
+      "train_tokens_per_second": 119.304
+    },
+    {
+      "epoch": 0.07282758620689656,
+      "grad_norm": 7.687803268432617,
+      "learning_rate": 4.18045112781955e-06,
+      "loss": 0.4136406183242798,
+      "num_input_tokens_seen": 502454,
+      "step": 1584,
+      "train_runtime": 4210.9467,
+      "train_tokens_per_second": 119.321
+    },
+    {
+      "epoch": 0.07287356321839081,
+      "grad_norm": 7.2596611976623535,
+      "learning_rate": 4.170426065162907e-06,
+      "loss": 0.4339594542980194,
+      "num_input_tokens_seen": 502748,
+      "step": 1585,
+      "train_runtime": 4212.8987,
+      "train_tokens_per_second": 119.335
+    },
+    {
+      "epoch": 0.07291954022988506,
+      "grad_norm": 8.879612922668457,
+      "learning_rate": 4.160401002506266e-06,
+      "loss": 0.5420070290565491,
+      "num_input_tokens_seen": 503038,
+      "step": 1586,
+      "train_runtime": 4214.8341,
+      "train_tokens_per_second": 119.349
+    },
+    {
+      "epoch": 0.07296551724137931,
+      "grad_norm": 7.438442707061768,
+      "learning_rate": 4.150375939849624e-06,
+      "loss": 0.5225954055786133,
+      "num_input_tokens_seen": 503402,
+      "step": 1587,
+      "train_runtime": 4216.821,
+      "train_tokens_per_second": 119.38
+    },
+    {
+      "epoch": 0.07301149425287357,
+      "grad_norm": 8.286493301391602,
+      "learning_rate": 4.140350877192983e-06,
+      "loss": 0.6506780385971069,
+      "num_input_tokens_seen": 503698,
+      "step": 1588,
+      "train_runtime": 4218.7745,
+      "train_tokens_per_second": 119.394
+    },
+    {
+      "epoch": 0.07305747126436782,
+      "grad_norm": 7.63776969909668,
+      "learning_rate": 4.130325814536341e-06,
+      "loss": 0.4458877742290497,
+      "num_input_tokens_seen": 504018,
+      "step": 1589,
+      "train_runtime": 4220.799,
+      "train_tokens_per_second": 119.413
+    },
+    {
+      "epoch": 0.07310344827586207,
+      "grad_norm": 6.316173553466797,
+      "learning_rate": 4.1203007518797e-06,
+      "loss": 0.27364295721054077,
+      "num_input_tokens_seen": 504274,
+      "step": 1590,
+      "train_runtime": 4222.7448,
+      "train_tokens_per_second": 119.419
+    },
+    {
+      "epoch": 0.07314942528735632,
+      "grad_norm": 8.226495742797852,
+      "learning_rate": 4.110275689223058e-06,
+      "loss": 0.5500906705856323,
+      "num_input_tokens_seen": 504668,
+      "step": 1591,
+      "train_runtime": 4237.9847,
+      "train_tokens_per_second": 119.082
+    },
+    {
+      "epoch": 0.07319540229885058,
+      "grad_norm": 11.585219383239746,
+      "learning_rate": 4.100250626566417e-06,
+      "loss": 0.5101352334022522,
+      "num_input_tokens_seen": 504918,
+      "step": 1592,
+      "train_runtime": 4239.9262,
+      "train_tokens_per_second": 119.087
+    },
+    {
+      "epoch": 0.07324137931034483,
+      "grad_norm": 6.878238201141357,
+      "learning_rate": 4.090225563909775e-06,
+      "loss": 0.4139741063117981,
+      "num_input_tokens_seen": 505274,
+      "step": 1593,
+      "train_runtime": 4241.8798,
+      "train_tokens_per_second": 119.116
+    },
+    {
+      "epoch": 0.07328735632183908,
+      "grad_norm": 7.619193077087402,
+      "learning_rate": 4.080200501253133e-06,
+      "loss": 0.4343482553958893,
+      "num_input_tokens_seen": 505582,
+      "step": 1594,
+      "train_runtime": 4243.86,
+      "train_tokens_per_second": 119.133
+    },
+    {
+      "epoch": 0.07333333333333333,
+      "grad_norm": 7.207488059997559,
+      "learning_rate": 4.070175438596491e-06,
+      "loss": 0.39172735810279846,
+      "num_input_tokens_seen": 505842,
+      "step": 1595,
+      "train_runtime": 4245.8827,
+      "train_tokens_per_second": 119.137
+    },
+    {
+      "epoch": 0.07337931034482759,
+      "grad_norm": 7.104700565338135,
+      "learning_rate": 4.06015037593985e-06,
+      "loss": 0.4126981794834137,
+      "num_input_tokens_seen": 506134,
+      "step": 1596,
+      "train_runtime": 4247.8396,
+      "train_tokens_per_second": 119.151
+    },
+    {
+      "epoch": 0.07342528735632184,
+      "grad_norm": 6.772775650024414,
+      "learning_rate": 4.050125313283208e-06,
+      "loss": 0.38334089517593384,
+      "num_input_tokens_seen": 506386,
+      "step": 1597,
+      "train_runtime": 4249.7922,
+      "train_tokens_per_second": 119.155
+    },
+    {
+      "epoch": 0.07347126436781609,
+      "grad_norm": 6.416538715362549,
+      "learning_rate": 4.040100250626567e-06,
+      "loss": 0.3199250400066376,
+      "num_input_tokens_seen": 506644,
+      "step": 1598,
+      "train_runtime": 4251.7539,
+      "train_tokens_per_second": 119.161
+    },
+    {
+      "epoch": 0.07351724137931034,
+      "grad_norm": 6.789169788360596,
+      "learning_rate": 4.030075187969925e-06,
+      "loss": 0.3942154049873352,
+      "num_input_tokens_seen": 507002,
+      "step": 1599,
+      "train_runtime": 4253.7232,
+      "train_tokens_per_second": 119.19
+    },
+    {
+      "epoch": 0.0735632183908046,
+      "grad_norm": 7.095794200897217,
+      "learning_rate": 4.020050125313284e-06,
+      "loss": 0.366854190826416,
+      "num_input_tokens_seen": 507254,
+      "step": 1600,
+      "train_runtime": 4255.8338,
+      "train_tokens_per_second": 119.19
+    },
+    {
+      "epoch": 0.0735632183908046,
+      "eval_loss": 1.6636571884155273,
+      "eval_runtime": 54.8262,
+      "eval_samples_per_second": 18.239,
+      "eval_steps_per_second": 9.12,
+      "num_input_tokens_seen": 507254,
+      "step": 1600
+    },
+    {
+      "epoch": 0.07360919540229885,
+      "grad_norm": 6.588114261627197,
+      "learning_rate": 4.010025062656641e-06,
+      "loss": 0.3823266923427582,
+      "num_input_tokens_seen": 507560,
+      "step": 1601,
+      "train_runtime": 4312.6649,
+      "train_tokens_per_second": 117.691
+    },
+    {
+      "epoch": 0.0736551724137931,
+      "grad_norm": 7.267997741699219,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.5816208720207214,
+      "num_input_tokens_seen": 508014,
+      "step": 1602,
+      "train_runtime": 4314.6651,
+      "train_tokens_per_second": 117.741
+    },
+    {
+      "epoch": 0.07370114942528735,
+      "grad_norm": 7.822113990783691,
+      "learning_rate": 3.989974937343359e-06,
+      "loss": 0.47935065627098083,
+      "num_input_tokens_seen": 508328,
+      "step": 1603,
+      "train_runtime": 4316.6245,
+      "train_tokens_per_second": 117.761
+    },
+    {
+      "epoch": 0.07374712643678161,
+      "grad_norm": 7.869327545166016,
+      "learning_rate": 3.979949874686717e-06,
+      "loss": 0.40576744079589844,
+      "num_input_tokens_seen": 508564,
+      "step": 1604,
+      "train_runtime": 4318.5603,
+      "train_tokens_per_second": 117.762
+    },
+    {
+      "epoch": 0.07379310344827586,
+      "grad_norm": 6.4306416511535645,
+      "learning_rate": 3.969924812030075e-06,
+      "loss": 0.3773975372314453,
+      "num_input_tokens_seen": 508880,
+      "step": 1605,
+      "train_runtime": 4320.4984,
+      "train_tokens_per_second": 117.783
+    },
+    {
+      "epoch": 0.07383908045977011,
+      "grad_norm": 7.573537826538086,
+      "learning_rate": 3.959899749373434e-06,
+      "loss": 0.3697569966316223,
+      "num_input_tokens_seen": 509146,
+      "step": 1606,
+      "train_runtime": 4322.4401,
+      "train_tokens_per_second": 117.791
+    },
+    {
+      "epoch": 0.07388505747126436,
+      "grad_norm": 6.8335113525390625,
+      "learning_rate": 3.949874686716792e-06,
+      "loss": 0.39870771765708923,
+      "num_input_tokens_seen": 509522,
+      "step": 1607,
+      "train_runtime": 4324.4145,
+      "train_tokens_per_second": 117.825
+    },
+    {
+      "epoch": 0.07393103448275862,
+      "grad_norm": 7.273937225341797,
+      "learning_rate": 3.939849624060151e-06,
+      "loss": 0.5059205889701843,
+      "num_input_tokens_seen": 509876,
+      "step": 1608,
+      "train_runtime": 4326.3915,
+      "train_tokens_per_second": 117.852
+    },
+    {
+      "epoch": 0.07397701149425287,
+      "grad_norm": 6.84849739074707,
+      "learning_rate": 3.929824561403509e-06,
+      "loss": 0.4065139889717102,
+      "num_input_tokens_seen": 510130,
+      "step": 1609,
+      "train_runtime": 4328.3341,
+      "train_tokens_per_second": 117.858
+    },
+    {
+      "epoch": 0.07402298850574712,
+      "grad_norm": 7.5032243728637695,
+      "learning_rate": 3.919799498746868e-06,
+      "loss": 0.4133743643760681,
+      "num_input_tokens_seen": 510448,
+      "step": 1610,
+      "train_runtime": 4330.2833,
+      "train_tokens_per_second": 117.879
+    },
+    {
+      "epoch": 0.07406896551724138,
+      "grad_norm": 8.283543586730957,
+      "learning_rate": 3.909774436090225e-06,
+      "loss": 0.5078597068786621,
+      "num_input_tokens_seen": 510720,
+      "step": 1611,
+      "train_runtime": 4332.2275,
+      "train_tokens_per_second": 117.889
+    },
+    {
+      "epoch": 0.07411494252873563,
+      "grad_norm": 8.472081184387207,
+      "learning_rate": 3.899749373433585e-06,
+      "loss": 0.5025591254234314,
+      "num_input_tokens_seen": 511014,
+      "step": 1612,
+      "train_runtime": 4334.1848,
+      "train_tokens_per_second": 117.903
+    },
+    {
+      "epoch": 0.07416091954022988,
+      "grad_norm": 11.184776306152344,
+      "learning_rate": 3.889724310776943e-06,
+      "loss": 0.44690465927124023,
+      "num_input_tokens_seen": 511288,
+      "step": 1613,
+      "train_runtime": 4336.1508,
+      "train_tokens_per_second": 117.913
+    },
+    {
+      "epoch": 0.07420689655172413,
+      "grad_norm": 10.022826194763184,
+      "learning_rate": 3.879699248120301e-06,
+      "loss": 0.5291357040405273,
+      "num_input_tokens_seen": 511618,
+      "step": 1614,
+      "train_runtime": 4338.1104,
+      "train_tokens_per_second": 117.936
+    },
+    {
+      "epoch": 0.07425287356321839,
+      "grad_norm": 10.145757675170898,
+      "learning_rate": 3.869674185463659e-06,
+      "loss": 0.41287463903427124,
+      "num_input_tokens_seen": 511884,
+      "step": 1615,
+      "train_runtime": 4340.0839,
+      "train_tokens_per_second": 117.943
+    },
+    {
+      "epoch": 0.07429885057471264,
+      "grad_norm": 7.911931037902832,
+      "learning_rate": 3.859649122807018e-06,
+      "loss": 0.5255659818649292,
+      "num_input_tokens_seen": 512226,
+      "step": 1616,
+      "train_runtime": 4342.034,
+      "train_tokens_per_second": 117.969
+    },
+    {
+      "epoch": 0.07434482758620689,
+      "grad_norm": 10.514172554016113,
+      "learning_rate": 3.849624060150376e-06,
+      "loss": 0.6998154520988464,
+      "num_input_tokens_seen": 512526,
+      "step": 1617,
+      "train_runtime": 4343.9812,
+      "train_tokens_per_second": 117.985
+    },
+    {
+      "epoch": 0.07439080459770114,
+      "grad_norm": 7.514682292938232,
+      "learning_rate": 3.839598997493735e-06,
+      "loss": 0.4628090262413025,
+      "num_input_tokens_seen": 512828,
+      "step": 1618,
+      "train_runtime": 4345.9288,
+      "train_tokens_per_second": 118.002
+    },
+    {
+      "epoch": 0.0744367816091954,
+      "grad_norm": 7.007578372955322,
+      "learning_rate": 3.829573934837093e-06,
+      "loss": 0.45868954062461853,
+      "num_input_tokens_seen": 513192,
+      "step": 1619,
+      "train_runtime": 4347.9134,
+      "train_tokens_per_second": 118.032
+    },
+    {
+      "epoch": 0.07448275862068965,
+      "grad_norm": 8.817955017089844,
+      "learning_rate": 3.819548872180452e-06,
+      "loss": 0.3704671859741211,
+      "num_input_tokens_seen": 513440,
+      "step": 1620,
+      "train_runtime": 4349.8462,
+      "train_tokens_per_second": 118.036
+    },
+    {
+      "epoch": 0.0745287356321839,
+      "grad_norm": 7.611793041229248,
+      "learning_rate": 3.80952380952381e-06,
+      "loss": 0.5009437203407288,
+      "num_input_tokens_seen": 513798,
+      "step": 1621,
+      "train_runtime": 4364.426,
+      "train_tokens_per_second": 117.724
+    },
+    {
+      "epoch": 0.07457471264367817,
+      "grad_norm": 6.247671127319336,
+      "learning_rate": 3.7994987468671683e-06,
+      "loss": 0.40714675188064575,
+      "num_input_tokens_seen": 514212,
+      "step": 1622,
+      "train_runtime": 4366.4808,
+      "train_tokens_per_second": 117.763
+    },
+    {
+      "epoch": 0.07462068965517242,
+      "grad_norm": 6.984149932861328,
+      "learning_rate": 3.789473684210527e-06,
+      "loss": 0.5318334102630615,
+      "num_input_tokens_seen": 514582,
+      "step": 1623,
+      "train_runtime": 4368.4774,
+      "train_tokens_per_second": 117.794
+    },
+    {
+      "epoch": 0.07466666666666667,
+      "grad_norm": 23.912731170654297,
+      "learning_rate": 3.779448621553885e-06,
+      "loss": 0.42978233098983765,
+      "num_input_tokens_seen": 514854,
+      "step": 1624,
+      "train_runtime": 4370.4241,
+      "train_tokens_per_second": 117.804
+    },
+    {
+      "epoch": 0.07471264367816093,
+      "grad_norm": 7.788239479064941,
+      "learning_rate": 3.7694235588972433e-06,
+      "loss": 0.36474743485450745,
+      "num_input_tokens_seen": 515114,
+      "step": 1625,
+      "train_runtime": 4372.4536,
+      "train_tokens_per_second": 117.809
+    },
+    {
+      "epoch": 0.07475862068965518,
+      "grad_norm": 8.246030807495117,
+      "learning_rate": 3.7593984962406014e-06,
+      "loss": 0.48351597785949707,
+      "num_input_tokens_seen": 515414,
+      "step": 1626,
+      "train_runtime": 4374.4976,
+      "train_tokens_per_second": 117.822
+    },
+    {
+      "epoch": 0.07480459770114943,
+      "grad_norm": 7.160177707672119,
+      "learning_rate": 3.7493734335839603e-06,
+      "loss": 0.3814699649810791,
+      "num_input_tokens_seen": 515734,
+      "step": 1627,
+      "train_runtime": 4376.5191,
+      "train_tokens_per_second": 117.841
+    },
+    {
+      "epoch": 0.07485057471264368,
+      "grad_norm": 9.433629035949707,
+      "learning_rate": 3.739348370927319e-06,
+      "loss": 0.4707894027233124,
+      "num_input_tokens_seen": 516038,
+      "step": 1628,
+      "train_runtime": 4378.4914,
+      "train_tokens_per_second": 117.857
+    },
+    {
+      "epoch": 0.07489655172413794,
+      "grad_norm": 7.79624080657959,
+      "learning_rate": 3.729323308270677e-06,
+      "loss": 0.32810497283935547,
+      "num_input_tokens_seen": 516298,
+      "step": 1629,
+      "train_runtime": 4380.4961,
+      "train_tokens_per_second": 117.863
+    },
+    {
+      "epoch": 0.07494252873563219,
+      "grad_norm": 7.606733322143555,
+      "learning_rate": 3.7192982456140354e-06,
+      "loss": 0.39508774876594543,
+      "num_input_tokens_seen": 516614,
+      "step": 1630,
+      "train_runtime": 4382.4569,
+      "train_tokens_per_second": 117.882
+    },
+    {
+      "epoch": 0.07498850574712644,
+      "grad_norm": 8.296076774597168,
+      "learning_rate": 3.7092731829573934e-06,
+      "loss": 0.3836243152618408,
+      "num_input_tokens_seen": 516862,
+      "step": 1631,
+      "train_runtime": 4384.3844,
+      "train_tokens_per_second": 117.887
+    },
+    {
+      "epoch": 0.0750344827586207,
+      "grad_norm": 7.348832130432129,
+      "learning_rate": 3.6992481203007523e-06,
+      "loss": 0.4283863306045532,
+      "num_input_tokens_seen": 517120,
+      "step": 1632,
+      "train_runtime": 4386.3302,
+      "train_tokens_per_second": 117.894
+    },
+    {
+      "epoch": 0.07508045977011495,
+      "grad_norm": 7.960624694824219,
+      "learning_rate": 3.689223057644111e-06,
+      "loss": 0.5945356488227844,
+      "num_input_tokens_seen": 517486,
+      "step": 1633,
+      "train_runtime": 4388.3672,
+      "train_tokens_per_second": 117.922
+    },
+    {
+      "epoch": 0.0751264367816092,
+      "grad_norm": 7.2380170822143555,
+      "learning_rate": 3.679197994987469e-06,
+      "loss": 0.38955235481262207,
+      "num_input_tokens_seen": 517754,
+      "step": 1634,
+      "train_runtime": 4390.3855,
+      "train_tokens_per_second": 117.929
+    },
+    {
+      "epoch": 0.07517241379310345,
+      "grad_norm": 7.709038734436035,
+      "learning_rate": 3.6691729323308274e-06,
+      "loss": 0.48962312936782837,
+      "num_input_tokens_seen": 518088,
+      "step": 1635,
+      "train_runtime": 4392.3413,
+      "train_tokens_per_second": 117.953
+    },
+    {
+      "epoch": 0.0752183908045977,
+      "grad_norm": 10.929262161254883,
+      "learning_rate": 3.6591478696741854e-06,
+      "loss": 0.4196699857711792,
+      "num_input_tokens_seen": 518334,
+      "step": 1636,
+      "train_runtime": 4394.2835,
+      "train_tokens_per_second": 117.956
+    },
+    {
+      "epoch": 0.07526436781609196,
+      "grad_norm": 7.947827339172363,
+      "learning_rate": 3.6491228070175443e-06,
+      "loss": 0.386614054441452,
+      "num_input_tokens_seen": 518610,
+      "step": 1637,
+      "train_runtime": 4396.2272,
+      "train_tokens_per_second": 117.967
+    },
+    {
+      "epoch": 0.07531034482758621,
+      "grad_norm": 7.691780090332031,
+      "learning_rate": 3.639097744360903e-06,
+      "loss": 0.40818464756011963,
+      "num_input_tokens_seen": 518904,
+      "step": 1638,
+      "train_runtime": 4398.1819,
+      "train_tokens_per_second": 117.981
+    },
+    {
+      "epoch": 0.07535632183908046,
+      "grad_norm": 7.336256980895996,
+      "learning_rate": 3.629072681704261e-06,
+      "loss": 0.432528018951416,
+      "num_input_tokens_seen": 519132,
+      "step": 1639,
+      "train_runtime": 4400.1363,
+      "train_tokens_per_second": 117.981
+    },
+    {
+      "epoch": 0.07540229885057471,
+      "grad_norm": 7.535099506378174,
+      "learning_rate": 3.6190476190476194e-06,
+      "loss": 0.45522022247314453,
+      "num_input_tokens_seen": 519504,
+      "step": 1640,
+      "train_runtime": 4402.1352,
+      "train_tokens_per_second": 118.012
+    },
+    {
+      "epoch": 0.07544827586206897,
+      "grad_norm": 7.481330871582031,
+      "learning_rate": 3.6090225563909775e-06,
+      "loss": 0.37103793025016785,
+      "num_input_tokens_seen": 519748,
+      "step": 1641,
+      "train_runtime": 4404.0998,
+      "train_tokens_per_second": 118.015
+    },
+    {
+      "epoch": 0.07549425287356322,
+      "grad_norm": 7.335155010223389,
+      "learning_rate": 3.5989974937343364e-06,
+      "loss": 0.4369334578514099,
+      "num_input_tokens_seen": 519970,
+      "step": 1642,
+      "train_runtime": 4406.0395,
+      "train_tokens_per_second": 118.013
+    },
+    {
+      "epoch": 0.07554022988505747,
+      "grad_norm": 6.278005599975586,
+      "learning_rate": 3.588972431077695e-06,
+      "loss": 0.4030069410800934,
+      "num_input_tokens_seen": 520270,
+      "step": 1643,
+      "train_runtime": 4407.9919,
+      "train_tokens_per_second": 118.029
+    },
+    {
+      "epoch": 0.07558620689655172,
+      "grad_norm": 6.86829137802124,
+      "learning_rate": 3.578947368421053e-06,
+      "loss": 0.43802517652511597,
+      "num_input_tokens_seen": 520528,
+      "step": 1644,
+      "train_runtime": 4409.9382,
+      "train_tokens_per_second": 118.035
+    },
+    {
+      "epoch": 0.07563218390804598,
+      "grad_norm": 7.912787437438965,
+      "learning_rate": 3.5689223057644114e-06,
+      "loss": 0.4219006896018982,
+      "num_input_tokens_seen": 520860,
+      "step": 1645,
+      "train_runtime": 4411.8938,
+      "train_tokens_per_second": 118.058
+    },
+    {
+      "epoch": 0.07567816091954023,
+      "grad_norm": 6.702023506164551,
+      "learning_rate": 3.5588972431077695e-06,
+      "loss": 0.37762436270713806,
+      "num_input_tokens_seen": 521168,
+      "step": 1646,
+      "train_runtime": 4413.9059,
+      "train_tokens_per_second": 118.074
+    },
+    {
+      "epoch": 0.07572413793103448,
+      "grad_norm": 8.325960159301758,
+      "learning_rate": 3.5488721804511284e-06,
+      "loss": 0.5221037268638611,
+      "num_input_tokens_seen": 521564,
+      "step": 1647,
+      "train_runtime": 4415.9159,
+      "train_tokens_per_second": 118.11
+    },
+    {
+      "epoch": 0.07577011494252874,
+      "grad_norm": 7.025114059448242,
+      "learning_rate": 3.5388471177944864e-06,
+      "loss": 0.4202582538127899,
+      "num_input_tokens_seen": 521998,
+      "step": 1648,
+      "train_runtime": 4417.9148,
+      "train_tokens_per_second": 118.155
+    },
+    {
+      "epoch": 0.07581609195402299,
+      "grad_norm": 8.078812599182129,
+      "learning_rate": 3.528822055137845e-06,
+      "loss": 0.45326465368270874,
+      "num_input_tokens_seen": 522346,
+      "step": 1649,
+      "train_runtime": 4419.8877,
+      "train_tokens_per_second": 118.181
+    },
+    {
+      "epoch": 0.07586206896551724,
+      "grad_norm": 7.274520397186279,
+      "learning_rate": 3.5187969924812034e-06,
+      "loss": 0.4773089289665222,
+      "num_input_tokens_seen": 522732,
+      "step": 1650,
+      "train_runtime": 4421.8721,
+      "train_tokens_per_second": 118.215
+    },
+    {
+      "epoch": 0.07590804597701149,
+      "grad_norm": 6.407050609588623,
+      "learning_rate": 3.5087719298245615e-06,
+      "loss": 0.38265374302864075,
+      "num_input_tokens_seen": 523048,
+      "step": 1651,
+      "train_runtime": 4436.9042,
+      "train_tokens_per_second": 117.886
+    },
+    {
+      "epoch": 0.07595402298850575,
+      "grad_norm": 7.074411392211914,
+      "learning_rate": 3.4987468671679204e-06,
+      "loss": 0.33349496126174927,
+      "num_input_tokens_seen": 523358,
+      "step": 1652,
+      "train_runtime": 4438.9778,
+      "train_tokens_per_second": 117.901
+    },
+    {
+      "epoch": 0.076,
+      "grad_norm": 6.719621658325195,
+      "learning_rate": 3.4887218045112785e-06,
+      "loss": 0.4396601915359497,
+      "num_input_tokens_seen": 524056,
+      "step": 1653,
+      "train_runtime": 4441.0833,
+      "train_tokens_per_second": 118.002
+    },
+    {
+      "epoch": 0.07604597701149425,
+      "grad_norm": 8.060017585754395,
+      "learning_rate": 3.478696741854637e-06,
+      "loss": 0.4098191261291504,
+      "num_input_tokens_seen": 524378,
+      "step": 1654,
+      "train_runtime": 4443.3098,
+      "train_tokens_per_second": 118.015
+    },
+    {
+      "epoch": 0.0760919540229885,
+      "grad_norm": 8.207612991333008,
+      "learning_rate": 3.468671679197995e-06,
+      "loss": 0.3707464933395386,
+      "num_input_tokens_seen": 524608,
+      "step": 1655,
+      "train_runtime": 4445.3321,
+      "train_tokens_per_second": 118.013
+    },
+    {
+      "epoch": 0.07613793103448276,
+      "grad_norm": 7.073892116546631,
+      "learning_rate": 3.4586466165413535e-06,
+      "loss": 0.4043232500553131,
+      "num_input_tokens_seen": 524852,
+      "step": 1656,
+      "train_runtime": 4447.3285,
+      "train_tokens_per_second": 118.015
+    },
+    {
+      "epoch": 0.07618390804597701,
+      "grad_norm": 13.12764835357666,
+      "learning_rate": 3.4486215538847124e-06,
+      "loss": 0.5430985689163208,
+      "num_input_tokens_seen": 525186,
+      "step": 1657,
+      "train_runtime": 4449.3721,
+      "train_tokens_per_second": 118.036
+    },
+    {
+      "epoch": 0.07622988505747126,
+      "grad_norm": 7.269232273101807,
+      "learning_rate": 3.4385964912280705e-06,
+      "loss": 0.5189374089241028,
+      "num_input_tokens_seen": 525508,
+      "step": 1658,
+      "train_runtime": 4451.3422,
+      "train_tokens_per_second": 118.056
+    },
+    {
+      "epoch": 0.07627586206896551,
+      "grad_norm": 9.399940490722656,
+      "learning_rate": 3.428571428571429e-06,
+      "loss": 0.6738297939300537,
+      "num_input_tokens_seen": 525862,
+      "step": 1659,
+      "train_runtime": 4453.3401,
+      "train_tokens_per_second": 118.083
+    },
+    {
+      "epoch": 0.07632183908045977,
+      "grad_norm": 7.022154331207275,
+      "learning_rate": 3.418546365914787e-06,
+      "loss": 0.34696778655052185,
+      "num_input_tokens_seen": 526126,
+      "step": 1660,
+      "train_runtime": 4455.305,
+      "train_tokens_per_second": 118.09
+    },
+    {
+      "epoch": 0.07636781609195402,
+      "grad_norm": 7.5776591300964355,
+      "learning_rate": 3.4085213032581455e-06,
+      "loss": 0.41206005215644836,
+      "num_input_tokens_seen": 526376,
+      "step": 1661,
+      "train_runtime": 4457.2725,
+      "train_tokens_per_second": 118.094
+    },
+    {
+      "epoch": 0.07641379310344827,
+      "grad_norm": 9.169903755187988,
+      "learning_rate": 3.3984962406015044e-06,
+      "loss": 0.470018595457077,
+      "num_input_tokens_seen": 526666,
+      "step": 1662,
+      "train_runtime": 4459.2584,
+      "train_tokens_per_second": 118.106
+    },
+    {
+      "epoch": 0.07645977011494252,
+      "grad_norm": 10.000397682189941,
+      "learning_rate": 3.3884711779448625e-06,
+      "loss": 0.6311056017875671,
+      "num_input_tokens_seen": 526962,
+      "step": 1663,
+      "train_runtime": 4461.1953,
+      "train_tokens_per_second": 118.121
+    },
+    {
+      "epoch": 0.07650574712643678,
+      "grad_norm": 7.303699970245361,
+      "learning_rate": 3.378446115288221e-06,
+      "loss": 0.43634146451950073,
+      "num_input_tokens_seen": 527412,
+      "step": 1664,
+      "train_runtime": 4463.197,
+      "train_tokens_per_second": 118.169
+    },
+    {
+      "epoch": 0.07655172413793103,
+      "grad_norm": 7.92708158493042,
+      "learning_rate": 3.368421052631579e-06,
+      "loss": 0.5328547954559326,
+      "num_input_tokens_seen": 527810,
+      "step": 1665,
+      "train_runtime": 4465.1683,
+      "train_tokens_per_second": 118.206
+    },
+    {
+      "epoch": 0.07659770114942528,
+      "grad_norm": 8.074692726135254,
+      "learning_rate": 3.3583959899749375e-06,
+      "loss": 0.4190375804901123,
+      "num_input_tokens_seen": 528106,
+      "step": 1666,
+      "train_runtime": 4467.1239,
+      "train_tokens_per_second": 118.221
+    },
+    {
+      "epoch": 0.07664367816091953,
+      "grad_norm": 6.452220916748047,
+      "learning_rate": 3.3483709273182964e-06,
+      "loss": 0.35974109172821045,
+      "num_input_tokens_seen": 528402,
+      "step": 1667,
+      "train_runtime": 4469.0837,
+      "train_tokens_per_second": 118.235
+    },
+    {
+      "epoch": 0.07668965517241379,
+      "grad_norm": 8.29507064819336,
+      "learning_rate": 3.3383458646616545e-06,
+      "loss": 0.4307975172996521,
+      "num_input_tokens_seen": 528720,
+      "step": 1668,
+      "train_runtime": 4471.0318,
+      "train_tokens_per_second": 118.255
+    },
+    {
+      "epoch": 0.07673563218390804,
+      "grad_norm": 7.829823017120361,
+      "learning_rate": 3.328320802005013e-06,
+      "loss": 0.582558274269104,
+      "num_input_tokens_seen": 529026,
+      "step": 1669,
+      "train_runtime": 4472.9836,
+      "train_tokens_per_second": 118.271
+    },
+    {
+      "epoch": 0.07678160919540229,
+      "grad_norm": 12.883645057678223,
+      "learning_rate": 3.318295739348371e-06,
+      "loss": 0.3684832751750946,
+      "num_input_tokens_seen": 529286,
+      "step": 1670,
+      "train_runtime": 4474.9147,
+      "train_tokens_per_second": 118.278
+    },
+    {
+      "epoch": 0.07682758620689654,
+      "grad_norm": 6.940089225769043,
+      "learning_rate": 3.3082706766917295e-06,
+      "loss": 0.3522677421569824,
+      "num_input_tokens_seen": 529572,
+      "step": 1671,
+      "train_runtime": 4476.8712,
+      "train_tokens_per_second": 118.291
+    },
+    {
+      "epoch": 0.07687356321839081,
+      "grad_norm": 10.376507759094238,
+      "learning_rate": 3.2982456140350885e-06,
+      "loss": 0.4347451329231262,
+      "num_input_tokens_seen": 529850,
+      "step": 1672,
+      "train_runtime": 4478.8345,
+      "train_tokens_per_second": 118.301
+    },
+    {
+      "epoch": 0.07691954022988506,
+      "grad_norm": 7.491230010986328,
+      "learning_rate": 3.2882205513784465e-06,
+      "loss": 0.4169405400753021,
+      "num_input_tokens_seen": 530132,
+      "step": 1673,
+      "train_runtime": 4480.8406,
+      "train_tokens_per_second": 118.311
+    },
+    {
+      "epoch": 0.07696551724137932,
+      "grad_norm": 7.8009843826293945,
+      "learning_rate": 3.278195488721805e-06,
+      "loss": 0.45866137742996216,
+      "num_input_tokens_seen": 530396,
+      "step": 1674,
+      "train_runtime": 4482.8218,
+      "train_tokens_per_second": 118.317
+    },
+    {
+      "epoch": 0.07701149425287357,
+      "grad_norm": 7.69118070602417,
+      "learning_rate": 3.268170426065163e-06,
+      "loss": 0.47137773036956787,
+      "num_input_tokens_seen": 530702,
+      "step": 1675,
+      "train_runtime": 4484.827,
+      "train_tokens_per_second": 118.333
+    },
+    {
+      "epoch": 0.07705747126436782,
+      "grad_norm": 9.579022407531738,
+      "learning_rate": 3.2581453634085216e-06,
+      "loss": 0.3288058042526245,
+      "num_input_tokens_seen": 530946,
+      "step": 1676,
+      "train_runtime": 4486.777,
+      "train_tokens_per_second": 118.336
+    },
+    {
+      "epoch": 0.07710344827586207,
+      "grad_norm": 7.989346981048584,
+      "learning_rate": 3.24812030075188e-06,
+      "loss": 0.4825642704963684,
+      "num_input_tokens_seen": 531368,
+      "step": 1677,
+      "train_runtime": 4488.8358,
+      "train_tokens_per_second": 118.375
+    },
+    {
+      "epoch": 0.07714942528735633,
+      "grad_norm": 8.046740531921387,
+      "learning_rate": 3.2380952380952385e-06,
+      "loss": 0.4609355032444,
+      "num_input_tokens_seen": 531716,
+      "step": 1678,
+      "train_runtime": 4490.8579,
+      "train_tokens_per_second": 118.4
+    },
+    {
+      "epoch": 0.07719540229885058,
+      "grad_norm": 7.730014324188232,
+      "learning_rate": 3.2280701754385966e-06,
+      "loss": 0.40114837884902954,
+      "num_input_tokens_seen": 531972,
+      "step": 1679,
+      "train_runtime": 4492.826,
+      "train_tokens_per_second": 118.405
+    },
+    {
+      "epoch": 0.07724137931034483,
+      "grad_norm": 8.52988338470459,
+      "learning_rate": 3.218045112781955e-06,
+      "loss": 0.4026782810688019,
+      "num_input_tokens_seen": 532288,
+      "step": 1680,
+      "train_runtime": 4494.7758,
+      "train_tokens_per_second": 118.424
+    },
+    {
+      "epoch": 0.07728735632183908,
+      "grad_norm": 8.861639976501465,
+      "learning_rate": 3.2080200501253136e-06,
+      "loss": 0.499417781829834,
+      "num_input_tokens_seen": 532704,
+      "step": 1681,
+      "train_runtime": 4509.2897,
+      "train_tokens_per_second": 118.135
+    },
+    {
+      "epoch": 0.07733333333333334,
+      "grad_norm": 8.764716148376465,
+      "learning_rate": 3.197994987468672e-06,
+      "loss": 0.4563250243663788,
+      "num_input_tokens_seen": 533036,
+      "step": 1682,
+      "train_runtime": 4511.2534,
+      "train_tokens_per_second": 118.157
+    },
+    {
+      "epoch": 0.07737931034482759,
+      "grad_norm": 6.531656265258789,
+      "learning_rate": 3.1879699248120305e-06,
+      "loss": 0.3374243974685669,
+      "num_input_tokens_seen": 533296,
+      "step": 1683,
+      "train_runtime": 4513.2216,
+      "train_tokens_per_second": 118.163
+    },
+    {
+      "epoch": 0.07742528735632184,
+      "grad_norm": 7.941148281097412,
+      "learning_rate": 3.1779448621553886e-06,
+      "loss": 0.413734495639801,
+      "num_input_tokens_seen": 533566,
+      "step": 1684,
+      "train_runtime": 4515.2062,
+      "train_tokens_per_second": 118.171
+    },
+    {
+      "epoch": 0.0774712643678161,
+      "grad_norm": 8.646973609924316,
+      "learning_rate": 3.167919799498747e-06,
+      "loss": 0.44844621419906616,
+      "num_input_tokens_seen": 533830,
+      "step": 1685,
+      "train_runtime": 4517.1983,
+      "train_tokens_per_second": 118.177
+    },
+    {
+      "epoch": 0.07751724137931035,
+      "grad_norm": 8.277411460876465,
+      "learning_rate": 3.157894736842105e-06,
+      "loss": 0.4859049618244171,
+      "num_input_tokens_seen": 534216,
+      "step": 1686,
+      "train_runtime": 4519.2305,
+      "train_tokens_per_second": 118.21
+    },
+    {
+      "epoch": 0.0775632183908046,
+      "grad_norm": 9.720200538635254,
+      "learning_rate": 3.147869674185464e-06,
+      "loss": 0.582287609577179,
+      "num_input_tokens_seen": 534482,
+      "step": 1687,
+      "train_runtime": 4521.1968,
+      "train_tokens_per_second": 118.217
+    },
+    {
+      "epoch": 0.07760919540229885,
+      "grad_norm": 8.369261741638184,
+      "learning_rate": 3.1378446115288226e-06,
+      "loss": 0.35591331124305725,
+      "num_input_tokens_seen": 534746,
+      "step": 1688,
+      "train_runtime": 4523.1536,
+      "train_tokens_per_second": 118.224
+    },
+    {
+      "epoch": 0.0776551724137931,
+      "grad_norm": 6.724172115325928,
+      "learning_rate": 3.1278195488721806e-06,
+      "loss": 0.5251051783561707,
+      "num_input_tokens_seen": 535262,
+      "step": 1689,
+      "train_runtime": 4525.1722,
+      "train_tokens_per_second": 118.285
+    },
+    {
+      "epoch": 0.07770114942528736,
+      "grad_norm": 8.139476776123047,
+      "learning_rate": 3.117794486215539e-06,
+      "loss": 0.4429723620414734,
+      "num_input_tokens_seen": 535552,
+      "step": 1690,
+      "train_runtime": 4527.1279,
+      "train_tokens_per_second": 118.298
+    },
+    {
+      "epoch": 0.07774712643678161,
+      "grad_norm": 7.223095417022705,
+      "learning_rate": 3.107769423558897e-06,
+      "loss": 0.4311490058898926,
+      "num_input_tokens_seen": 535890,
+      "step": 1691,
+      "train_runtime": 4529.1251,
+      "train_tokens_per_second": 118.321
+    },
+    {
+      "epoch": 0.07779310344827586,
+      "grad_norm": 8.413722038269043,
+      "learning_rate": 3.097744360902256e-06,
+      "loss": 0.5396718978881836,
+      "num_input_tokens_seen": 536276,
+      "step": 1692,
+      "train_runtime": 4531.1699,
+      "train_tokens_per_second": 118.353
+    },
+    {
+      "epoch": 0.07783908045977012,
+      "grad_norm": 8.579139709472656,
+      "learning_rate": 3.0877192982456146e-06,
+      "loss": 0.5883654356002808,
+      "num_input_tokens_seen": 536528,
+      "step": 1693,
+      "train_runtime": 4533.1482,
+      "train_tokens_per_second": 118.357
+    },
+    {
+      "epoch": 0.07788505747126437,
+      "grad_norm": 9.403660774230957,
+      "learning_rate": 3.0776942355889726e-06,
+      "loss": 0.4726146459579468,
+      "num_input_tokens_seen": 536870,
+      "step": 1694,
+      "train_runtime": 4535.1176,
+      "train_tokens_per_second": 118.381
+    },
+    {
+      "epoch": 0.07793103448275862,
+      "grad_norm": 9.025625228881836,
+      "learning_rate": 3.067669172932331e-06,
+      "loss": 0.545159101486206,
+      "num_input_tokens_seen": 537198,
+      "step": 1695,
+      "train_runtime": 4537.1696,
+      "train_tokens_per_second": 118.399
+    },
+    {
+      "epoch": 0.07797701149425287,
+      "grad_norm": 7.852901935577393,
+      "learning_rate": 3.057644110275689e-06,
+      "loss": 0.44690242409706116,
+      "num_input_tokens_seen": 537492,
+      "step": 1696,
+      "train_runtime": 4539.1088,
+      "train_tokens_per_second": 118.414
+    },
+    {
+      "epoch": 0.07802298850574713,
+      "grad_norm": 9.632786750793457,
+      "learning_rate": 3.047619047619048e-06,
+      "loss": 0.3916051983833313,
+      "num_input_tokens_seen": 537736,
+      "step": 1697,
+      "train_runtime": 4541.0476,
+      "train_tokens_per_second": 118.417
+    },
+    {
+      "epoch": 0.07806896551724138,
+      "grad_norm": 6.765559673309326,
+      "learning_rate": 3.0375939849624066e-06,
+      "loss": 0.3171115815639496,
+      "num_input_tokens_seen": 537988,
+      "step": 1698,
+      "train_runtime": 4542.9782,
+      "train_tokens_per_second": 118.422
+    },
+    {
+      "epoch": 0.07811494252873563,
+      "grad_norm": 7.956576824188232,
+      "learning_rate": 3.0275689223057647e-06,
+      "loss": 0.38876232504844666,
+      "num_input_tokens_seen": 538274,
+      "step": 1699,
+      "train_runtime": 4544.9399,
+      "train_tokens_per_second": 118.434
+    },
+    {
+      "epoch": 0.07816091954022988,
+      "grad_norm": 8.61601734161377,
+      "learning_rate": 3.017543859649123e-06,
+      "loss": 0.4542631506919861,
+      "num_input_tokens_seen": 538530,
+      "step": 1700,
+      "train_runtime": 4546.8849,
+      "train_tokens_per_second": 118.439
+    },
+    {
+      "epoch": 0.07820689655172414,
+      "grad_norm": 7.048073768615723,
+      "learning_rate": 3.007518796992481e-06,
+      "loss": 0.4090226888656616,
+      "num_input_tokens_seen": 538766,
+      "step": 1701,
+      "train_runtime": 4548.8137,
+      "train_tokens_per_second": 118.441
+    },
+    {
+      "epoch": 0.07825287356321839,
+      "grad_norm": 7.331066608428955,
+      "learning_rate": 2.9974937343358397e-06,
+      "loss": 0.31398913264274597,
+      "num_input_tokens_seen": 539010,
+      "step": 1702,
+      "train_runtime": 4550.7477,
+      "train_tokens_per_second": 118.444
+    },
+    {
+      "epoch": 0.07829885057471264,
+      "grad_norm": 7.632887840270996,
+      "learning_rate": 2.9874686716791986e-06,
+      "loss": 0.29063817858695984,
+      "num_input_tokens_seen": 539254,
+      "step": 1703,
+      "train_runtime": 4552.6852,
+      "train_tokens_per_second": 118.447
+    },
+    {
+      "epoch": 0.0783448275862069,
+      "grad_norm": 6.532099723815918,
+      "learning_rate": 2.9774436090225567e-06,
+      "loss": 0.36357584595680237,
+      "num_input_tokens_seen": 539512,
+      "step": 1704,
+      "train_runtime": 4554.6408,
+      "train_tokens_per_second": 118.453
+    },
+    {
+      "epoch": 0.07839080459770115,
+      "grad_norm": 6.803408145904541,
+      "learning_rate": 2.967418546365915e-06,
+      "loss": 0.36048248410224915,
+      "num_input_tokens_seen": 539836,
+      "step": 1705,
+      "train_runtime": 4556.5982,
+      "train_tokens_per_second": 118.473
+    },
+    {
+      "epoch": 0.0784367816091954,
+      "grad_norm": 7.635514259338379,
+      "learning_rate": 2.9573934837092732e-06,
+      "loss": 0.37317177653312683,
+      "num_input_tokens_seen": 540176,
+      "step": 1706,
+      "train_runtime": 4558.5535,
+      "train_tokens_per_second": 118.497
+    },
+    {
+      "epoch": 0.07848275862068965,
+      "grad_norm": 7.402206897735596,
+      "learning_rate": 2.9473684210526317e-06,
+      "loss": 0.32631269097328186,
+      "num_input_tokens_seen": 540440,
+      "step": 1707,
+      "train_runtime": 4560.5148,
+      "train_tokens_per_second": 118.504
+    },
+    {
+      "epoch": 0.0785287356321839,
+      "grad_norm": 7.177687168121338,
+      "learning_rate": 2.93734335839599e-06,
+      "loss": 0.31662577390670776,
+      "num_input_tokens_seen": 540724,
+      "step": 1708,
+      "train_runtime": 4562.554,
+      "train_tokens_per_second": 118.513
+    },
+    {
+      "epoch": 0.07857471264367816,
+      "grad_norm": 7.495624542236328,
+      "learning_rate": 2.9273182957393487e-06,
+      "loss": 0.46745815873146057,
+      "num_input_tokens_seen": 541018,
+      "step": 1709,
+      "train_runtime": 4564.5109,
+      "train_tokens_per_second": 118.527
+    },
+    {
+      "epoch": 0.07862068965517241,
+      "grad_norm": 7.3868327140808105,
+      "learning_rate": 2.9172932330827068e-06,
+      "loss": 0.4397212266921997,
+      "num_input_tokens_seen": 541326,
+      "step": 1710,
+      "train_runtime": 4566.4711,
+      "train_tokens_per_second": 118.544
+    },
+    {
+      "epoch": 0.07866666666666666,
+      "grad_norm": 6.463175296783447,
+      "learning_rate": 2.9072681704260652e-06,
+      "loss": 0.35786816477775574,
+      "num_input_tokens_seen": 541606,
+      "step": 1711,
+      "train_runtime": 4580.9489,
+      "train_tokens_per_second": 118.23
+    },
+    {
+      "epoch": 0.07871264367816092,
+      "grad_norm": 6.365063190460205,
+      "learning_rate": 2.8972431077694237e-06,
+      "loss": 0.36957696080207825,
+      "num_input_tokens_seen": 541880,
+      "step": 1712,
+      "train_runtime": 4582.9043,
+      "train_tokens_per_second": 118.239
+    },
+    {
+      "epoch": 0.07875862068965517,
+      "grad_norm": 7.89999532699585,
+      "learning_rate": 2.8872180451127822e-06,
+      "loss": 0.5105806589126587,
+      "num_input_tokens_seen": 542188,
+      "step": 1713,
+      "train_runtime": 4584.8725,
+      "train_tokens_per_second": 118.256
+    },
+    {
+      "epoch": 0.07880459770114942,
+      "grad_norm": 7.104886054992676,
+      "learning_rate": 2.8771929824561407e-06,
+      "loss": 0.42578667402267456,
+      "num_input_tokens_seen": 542484,
+      "step": 1714,
+      "train_runtime": 4586.7943,
+      "train_tokens_per_second": 118.271
+    },
+    {
+      "epoch": 0.07885057471264367,
+      "grad_norm": 8.263605117797852,
+      "learning_rate": 2.8671679197994988e-06,
+      "loss": 0.5237813591957092,
+      "num_input_tokens_seen": 542846,
+      "step": 1715,
+      "train_runtime": 4588.7706,
+      "train_tokens_per_second": 118.299
+    },
+    {
+      "epoch": 0.07889655172413793,
+      "grad_norm": 9.0348482131958,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 0.4177630543708801,
+      "num_input_tokens_seen": 543206,
+      "step": 1716,
+      "train_runtime": 4590.9383,
+      "train_tokens_per_second": 118.321
+    },
+    {
+      "epoch": 0.07894252873563218,
+      "grad_norm": 8.561115264892578,
+      "learning_rate": 2.8471177944862153e-06,
+      "loss": 0.5585188865661621,
+      "num_input_tokens_seen": 543490,
+      "step": 1717,
+      "train_runtime": 4593.0554,
+      "train_tokens_per_second": 118.329
+    },
+    {
+      "epoch": 0.07898850574712643,
+      "grad_norm": 7.145869255065918,
+      "learning_rate": 2.8370927318295742e-06,
+      "loss": 0.40125787258148193,
+      "num_input_tokens_seen": 543936,
+      "step": 1718,
+      "train_runtime": 4595.0332,
+      "train_tokens_per_second": 118.375
+    },
+    {
+      "epoch": 0.07903448275862068,
+      "grad_norm": 6.850518226623535,
+      "learning_rate": 2.8270676691729327e-06,
+      "loss": 0.3150305449962616,
+      "num_input_tokens_seen": 544194,
+      "step": 1719,
+      "train_runtime": 4596.9835,
+      "train_tokens_per_second": 118.381
+    },
+    {
+      "epoch": 0.07908045977011494,
+      "grad_norm": 8.219104766845703,
+      "learning_rate": 2.8170426065162908e-06,
+      "loss": 0.4087820053100586,
+      "num_input_tokens_seen": 544448,
+      "step": 1720,
+      "train_runtime": 4598.9388,
+      "train_tokens_per_second": 118.386
+    },
+    {
+      "epoch": 0.07912643678160919,
+      "grad_norm": 7.8992133140563965,
+      "learning_rate": 2.8070175438596493e-06,
+      "loss": 0.47887152433395386,
+      "num_input_tokens_seen": 544712,
+      "step": 1721,
+      "train_runtime": 4600.8891,
+      "train_tokens_per_second": 118.393
+    },
+    {
+      "epoch": 0.07917241379310345,
+      "grad_norm": 7.767656326293945,
+      "learning_rate": 2.7969924812030073e-06,
+      "loss": 0.4307166039943695,
+      "num_input_tokens_seen": 544992,
+      "step": 1722,
+      "train_runtime": 4602.8681,
+      "train_tokens_per_second": 118.403
+    },
+    {
+      "epoch": 0.07921839080459771,
+      "grad_norm": 6.98056697845459,
+      "learning_rate": 2.7869674185463662e-06,
+      "loss": 0.38568735122680664,
+      "num_input_tokens_seen": 545300,
+      "step": 1723,
+      "train_runtime": 4604.8485,
+      "train_tokens_per_second": 118.419
+    },
+    {
+      "epoch": 0.07926436781609196,
+      "grad_norm": 7.0472259521484375,
+      "learning_rate": 2.7769423558897247e-06,
+      "loss": 0.43514782190322876,
+      "num_input_tokens_seen": 545594,
+      "step": 1724,
+      "train_runtime": 4606.7906,
+      "train_tokens_per_second": 118.433
+    },
+    {
+      "epoch": 0.07931034482758621,
+      "grad_norm": 6.9541120529174805,
+      "learning_rate": 2.766917293233083e-06,
+      "loss": 0.399605393409729,
+      "num_input_tokens_seen": 545880,
+      "step": 1725,
+      "train_runtime": 4608.8016,
+      "train_tokens_per_second": 118.443
+    },
+    {
+      "epoch": 0.07935632183908047,
+      "grad_norm": 8.352391242980957,
+      "learning_rate": 2.7568922305764413e-06,
+      "loss": 0.43901270627975464,
+      "num_input_tokens_seen": 546150,
+      "step": 1726,
+      "train_runtime": 4610.7567,
+      "train_tokens_per_second": 118.451
+    },
+    {
+      "epoch": 0.07940229885057472,
+      "grad_norm": 6.75828218460083,
+      "learning_rate": 2.7468671679197994e-06,
+      "loss": 0.3787592053413391,
+      "num_input_tokens_seen": 546404,
+      "step": 1727,
+      "train_runtime": 4612.7529,
+      "train_tokens_per_second": 118.455
+    },
+    {
+      "epoch": 0.07944827586206897,
+      "grad_norm": 8.66209888458252,
+      "learning_rate": 2.7368421052631583e-06,
+      "loss": 0.4774472117424011,
+      "num_input_tokens_seen": 546698,
+      "step": 1728,
+      "train_runtime": 4614.7464,
+      "train_tokens_per_second": 118.468
+    },
+    {
+      "epoch": 0.07949425287356322,
+      "grad_norm": 6.795530796051025,
+      "learning_rate": 2.7268170426065167e-06,
+      "loss": 0.3278595805168152,
+      "num_input_tokens_seen": 546910,
+      "step": 1729,
+      "train_runtime": 4616.74,
+      "train_tokens_per_second": 118.462
+    },
+    {
+      "epoch": 0.07954022988505748,
+      "grad_norm": 8.187334060668945,
+      "learning_rate": 2.716791979949875e-06,
+      "loss": 0.49100738763809204,
+      "num_input_tokens_seen": 547170,
+      "step": 1730,
+      "train_runtime": 4618.6763,
+      "train_tokens_per_second": 118.469
+    },
+    {
+      "epoch": 0.07958620689655173,
+      "grad_norm": 7.767502784729004,
+      "learning_rate": 2.7067669172932333e-06,
+      "loss": 0.39085710048675537,
+      "num_input_tokens_seen": 547484,
+      "step": 1731,
+      "train_runtime": 4620.6439,
+      "train_tokens_per_second": 118.487
+    },
+    {
+      "epoch": 0.07963218390804598,
+      "grad_norm": 7.032953262329102,
+      "learning_rate": 2.6967418546365914e-06,
+      "loss": 0.3649653196334839,
+      "num_input_tokens_seen": 547756,
+      "step": 1732,
+      "train_runtime": 4622.6777,
+      "train_tokens_per_second": 118.493
+    },
+    {
+      "epoch": 0.07967816091954023,
+      "grad_norm": 7.271172046661377,
+      "learning_rate": 2.6867167919799503e-06,
+      "loss": 0.26579615473747253,
+      "num_input_tokens_seen": 547982,
+      "step": 1733,
+      "train_runtime": 4624.7391,
+      "train_tokens_per_second": 118.489
+    },
+    {
+      "epoch": 0.07972413793103449,
+      "grad_norm": 7.960947513580322,
+      "learning_rate": 2.6766917293233088e-06,
+      "loss": 0.47854000329971313,
+      "num_input_tokens_seen": 548272,
+      "step": 1734,
+      "train_runtime": 4626.71,
+      "train_tokens_per_second": 118.501
+    },
+    {
+      "epoch": 0.07977011494252874,
+      "grad_norm": 8.42829418182373,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 0.5312048196792603,
+      "num_input_tokens_seen": 548510,
+      "step": 1735,
+      "train_runtime": 4628.7365,
+      "train_tokens_per_second": 118.501
+    },
+    {
+      "epoch": 0.07981609195402299,
+      "grad_norm": 6.893427848815918,
+      "learning_rate": 2.6566416040100253e-06,
+      "loss": 0.4053395390510559,
+      "num_input_tokens_seen": 548808,
+      "step": 1736,
+      "train_runtime": 4630.6921,
+      "train_tokens_per_second": 118.515
+    },
+    {
+      "epoch": 0.07986206896551724,
+      "grad_norm": 7.288859844207764,
+      "learning_rate": 2.6466165413533834e-06,
+      "loss": 0.3709200620651245,
+      "num_input_tokens_seen": 549078,
+      "step": 1737,
+      "train_runtime": 4632.6999,
+      "train_tokens_per_second": 118.522
+    },
+    {
+      "epoch": 0.0799080459770115,
+      "grad_norm": 7.377291679382324,
+      "learning_rate": 2.6365914786967423e-06,
+      "loss": 0.3193652331829071,
+      "num_input_tokens_seen": 549364,
+      "step": 1738,
+      "train_runtime": 4634.6636,
+      "train_tokens_per_second": 118.534
+    },
+    {
+      "epoch": 0.07995402298850575,
+      "grad_norm": 6.757475852966309,
+      "learning_rate": 2.6265664160401004e-06,
+      "loss": 0.4178864657878876,
+      "num_input_tokens_seen": 549708,
+      "step": 1739,
+      "train_runtime": 4636.6359,
+      "train_tokens_per_second": 118.558
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 10.461713790893555,
+      "learning_rate": 2.616541353383459e-06,
+      "loss": 0.5244767665863037,
+      "num_input_tokens_seen": 549980,
+      "step": 1740,
+      "train_runtime": 4638.6478,
+      "train_tokens_per_second": 118.565
+    },
+    {
+      "epoch": 0.08004597701149425,
+      "grad_norm": 7.775877475738525,
+      "learning_rate": 2.606516290726817e-06,
+      "loss": 0.3699747622013092,
+      "num_input_tokens_seen": 550286,
+      "step": 1741,
+      "train_runtime": 4652.6103,
+      "train_tokens_per_second": 118.275
+    },
+    {
+      "epoch": 0.0800919540229885,
+      "grad_norm": 7.458606719970703,
+      "learning_rate": 2.5964912280701754e-06,
+      "loss": 0.48922139406204224,
+      "num_input_tokens_seen": 550842,
+      "step": 1742,
+      "train_runtime": 4654.6736,
+      "train_tokens_per_second": 118.342
+    },
+    {
+      "epoch": 0.08013793103448276,
+      "grad_norm": 7.654501914978027,
+      "learning_rate": 2.5864661654135343e-06,
+      "loss": 0.4521372318267822,
+      "num_input_tokens_seen": 551142,
+      "step": 1743,
+      "train_runtime": 4656.6192,
+      "train_tokens_per_second": 118.357
+    },
+    {
+      "epoch": 0.08018390804597701,
+      "grad_norm": 9.197833061218262,
+      "learning_rate": 2.5764411027568924e-06,
+      "loss": 0.44839343428611755,
+      "num_input_tokens_seen": 551448,
+      "step": 1744,
+      "train_runtime": 4658.5871,
+      "train_tokens_per_second": 118.372
+    },
+    {
+      "epoch": 0.08022988505747126,
+      "grad_norm": 6.440124988555908,
+      "learning_rate": 2.566416040100251e-06,
+      "loss": 0.3701336085796356,
+      "num_input_tokens_seen": 551786,
+      "step": 1745,
+      "train_runtime": 4660.6495,
+      "train_tokens_per_second": 118.393
+    },
+    {
+      "epoch": 0.08027586206896552,
+      "grad_norm": 7.900654315948486,
+      "learning_rate": 2.556390977443609e-06,
+      "loss": 0.5311776995658875,
+      "num_input_tokens_seen": 552126,
+      "step": 1746,
+      "train_runtime": 4662.648,
+      "train_tokens_per_second": 118.415
+    },
+    {
+      "epoch": 0.08032183908045977,
+      "grad_norm": 7.551472187042236,
+      "learning_rate": 2.5463659147869674e-06,
+      "loss": 0.4569704234600067,
+      "num_input_tokens_seen": 552360,
+      "step": 1747,
+      "train_runtime": 4664.6816,
+      "train_tokens_per_second": 118.413
+    },
+    {
+      "epoch": 0.08036781609195402,
+      "grad_norm": 9.04428482055664,
+      "learning_rate": 2.5363408521303263e-06,
+      "loss": 0.43082064390182495,
+      "num_input_tokens_seen": 552640,
+      "step": 1748,
+      "train_runtime": 4666.6903,
+      "train_tokens_per_second": 118.422
+    },
+    {
+      "epoch": 0.08041379310344828,
+      "grad_norm": 7.74245023727417,
+      "learning_rate": 2.5263157894736844e-06,
+      "loss": 0.4223887622356415,
+      "num_input_tokens_seen": 552894,
+      "step": 1749,
+      "train_runtime": 4668.7186,
+      "train_tokens_per_second": 118.425
+    },
+    {
+      "epoch": 0.08045977011494253,
+      "grad_norm": 5.408883571624756,
+      "learning_rate": 2.516290726817043e-06,
+      "loss": 0.2123432457447052,
+      "num_input_tokens_seen": 553146,
+      "step": 1750,
+      "train_runtime": 4670.6825,
+      "train_tokens_per_second": 118.429
+    },
+    {
+      "epoch": 0.08050574712643678,
+      "grad_norm": 7.574296951293945,
+      "learning_rate": 2.506265664160401e-06,
+      "loss": 0.4172983169555664,
+      "num_input_tokens_seen": 553470,
+      "step": 1751,
+      "train_runtime": 4672.7165,
+      "train_tokens_per_second": 118.447
+    },
+    {
+      "epoch": 0.08055172413793103,
+      "grad_norm": 8.019220352172852,
+      "learning_rate": 2.4962406015037594e-06,
+      "loss": 0.43811845779418945,
+      "num_input_tokens_seen": 553784,
+      "step": 1752,
+      "train_runtime": 4674.6719,
+      "train_tokens_per_second": 118.465
+    },
+    {
+      "epoch": 0.08059770114942529,
+      "grad_norm": 8.313697814941406,
+      "learning_rate": 2.486215538847118e-06,
+      "loss": 0.450636625289917,
+      "num_input_tokens_seen": 554220,
+      "step": 1753,
+      "train_runtime": 4676.6844,
+      "train_tokens_per_second": 118.507
+    },
+    {
+      "epoch": 0.08064367816091954,
+      "grad_norm": 7.183359146118164,
+      "learning_rate": 2.4761904761904764e-06,
+      "loss": 0.41533249616622925,
+      "num_input_tokens_seen": 554530,
+      "step": 1754,
+      "train_runtime": 4678.6585,
+      "train_tokens_per_second": 118.523
+    },
+    {
+      "epoch": 0.08068965517241379,
+      "grad_norm": 8.74924373626709,
+      "learning_rate": 2.466165413533835e-06,
+      "loss": 0.5330443978309631,
+      "num_input_tokens_seen": 554792,
+      "step": 1755,
+      "train_runtime": 4680.6052,
+      "train_tokens_per_second": 118.53
+    },
+    {
+      "epoch": 0.08073563218390804,
+      "grad_norm": 7.614817142486572,
+      "learning_rate": 2.456140350877193e-06,
+      "loss": 0.3097129166126251,
+      "num_input_tokens_seen": 555046,
+      "step": 1756,
+      "train_runtime": 4682.5467,
+      "train_tokens_per_second": 118.535
+    },
+    {
+      "epoch": 0.0807816091954023,
+      "grad_norm": 7.424604892730713,
+      "learning_rate": 2.4461152882205514e-06,
+      "loss": 0.37505459785461426,
+      "num_input_tokens_seen": 555286,
+      "step": 1757,
+      "train_runtime": 4684.483,
+      "train_tokens_per_second": 118.537
+    },
+    {
+      "epoch": 0.08082758620689655,
+      "grad_norm": 6.519960403442383,
+      "learning_rate": 2.43609022556391e-06,
+      "loss": 0.33945173025131226,
+      "num_input_tokens_seen": 555576,
+      "step": 1758,
+      "train_runtime": 4686.431,
+      "train_tokens_per_second": 118.55
+    },
+    {
+      "epoch": 0.0808735632183908,
+      "grad_norm": 7.936978816986084,
+      "learning_rate": 2.4260651629072684e-06,
+      "loss": 0.45852887630462646,
+      "num_input_tokens_seen": 555886,
+      "step": 1759,
+      "train_runtime": 4688.3962,
+      "train_tokens_per_second": 118.566
+    },
+    {
+      "epoch": 0.08091954022988505,
+      "grad_norm": 7.150357246398926,
+      "learning_rate": 2.416040100250627e-06,
+      "loss": 0.5692156553268433,
+      "num_input_tokens_seen": 556340,
+      "step": 1760,
+      "train_runtime": 4690.5359,
+      "train_tokens_per_second": 118.609
+    },
+    {
+      "epoch": 0.0809655172413793,
+      "grad_norm": 10.340112686157227,
+      "learning_rate": 2.406015037593985e-06,
+      "loss": 0.40212365984916687,
+      "num_input_tokens_seen": 556622,
+      "step": 1761,
+      "train_runtime": 4692.5794,
+      "train_tokens_per_second": 118.617
+    },
+    {
+      "epoch": 0.08101149425287356,
+      "grad_norm": 5.893184185028076,
+      "learning_rate": 2.3959899749373435e-06,
+      "loss": 0.2924824655056,
+      "num_input_tokens_seen": 556964,
+      "step": 1762,
+      "train_runtime": 4694.6371,
+      "train_tokens_per_second": 118.638
+    },
+    {
+      "epoch": 0.08105747126436781,
+      "grad_norm": 6.778119087219238,
+      "learning_rate": 2.385964912280702e-06,
+      "loss": 0.2572908103466034,
+      "num_input_tokens_seen": 557278,
+      "step": 1763,
+      "train_runtime": 4696.5951,
+      "train_tokens_per_second": 118.656
+    },
+    {
+      "epoch": 0.08110344827586206,
+      "grad_norm": 8.0773286819458,
+      "learning_rate": 2.3759398496240604e-06,
+      "loss": 0.40051108598709106,
+      "num_input_tokens_seen": 557568,
+      "step": 1764,
+      "train_runtime": 4698.5756,
+      "train_tokens_per_second": 118.667
+    },
+    {
+      "epoch": 0.08114942528735632,
+      "grad_norm": 7.1524577140808105,
+      "learning_rate": 2.365914786967419e-06,
+      "loss": 0.44823575019836426,
+      "num_input_tokens_seen": 557982,
+      "step": 1765,
+      "train_runtime": 4700.5846,
+      "train_tokens_per_second": 118.705
+    },
+    {
+      "epoch": 0.08119540229885057,
+      "grad_norm": 6.727236270904541,
+      "learning_rate": 2.355889724310777e-06,
+      "loss": 0.3018498718738556,
+      "num_input_tokens_seen": 558242,
+      "step": 1766,
+      "train_runtime": 4702.5665,
+      "train_tokens_per_second": 118.71
+    },
+    {
+      "epoch": 0.08124137931034482,
+      "grad_norm": 7.099841594696045,
+      "learning_rate": 2.3458646616541355e-06,
+      "loss": 0.4439983367919922,
+      "num_input_tokens_seen": 558520,
+      "step": 1767,
+      "train_runtime": 4704.5323,
+      "train_tokens_per_second": 118.72
+    },
+    {
+      "epoch": 0.08128735632183907,
+      "grad_norm": 7.709004878997803,
+      "learning_rate": 2.335839598997494e-06,
+      "loss": 0.3934343457221985,
+      "num_input_tokens_seen": 558834,
+      "step": 1768,
+      "train_runtime": 4706.5099,
+      "train_tokens_per_second": 118.736
+    },
+    {
+      "epoch": 0.08133333333333333,
+      "grad_norm": 7.746722221374512,
+      "learning_rate": 2.3258145363408524e-06,
+      "loss": 0.45177793502807617,
+      "num_input_tokens_seen": 559120,
+      "step": 1769,
+      "train_runtime": 4708.4668,
+      "train_tokens_per_second": 118.748
+    },
+    {
+      "epoch": 0.08137931034482758,
+      "grad_norm": 8.621835708618164,
+      "learning_rate": 2.3157894736842105e-06,
+      "loss": 0.454231858253479,
+      "num_input_tokens_seen": 559408,
+      "step": 1770,
+      "train_runtime": 4710.4036,
+      "train_tokens_per_second": 118.76
+    },
+    {
+      "epoch": 0.08142528735632185,
+      "grad_norm": 7.304722309112549,
+      "learning_rate": 2.305764411027569e-06,
+      "loss": 0.4125288724899292,
+      "num_input_tokens_seen": 559714,
+      "step": 1771,
+      "train_runtime": 4726.2743,
+      "train_tokens_per_second": 118.426
+    },
+    {
+      "epoch": 0.0814712643678161,
+      "grad_norm": 6.720957279205322,
+      "learning_rate": 2.2957393483709275e-06,
+      "loss": 0.4494350254535675,
+      "num_input_tokens_seen": 560012,
+      "step": 1772,
+      "train_runtime": 4728.2414,
+      "train_tokens_per_second": 118.44
+    },
+    {
+      "epoch": 0.08151724137931035,
+      "grad_norm": 6.90883731842041,
+      "learning_rate": 2.285714285714286e-06,
+      "loss": 0.3727937936782837,
+      "num_input_tokens_seen": 560356,
+      "step": 1773,
+      "train_runtime": 4730.2751,
+      "train_tokens_per_second": 118.462
+    },
+    {
+      "epoch": 0.0815632183908046,
+      "grad_norm": 7.366011619567871,
+      "learning_rate": 2.2756892230576445e-06,
+      "loss": 0.3889319896697998,
+      "num_input_tokens_seen": 560642,
+      "step": 1774,
+      "train_runtime": 4732.2628,
+      "train_tokens_per_second": 118.472
+    },
+    {
+      "epoch": 0.08160919540229886,
+      "grad_norm": 6.236909866333008,
+      "learning_rate": 2.2656641604010025e-06,
+      "loss": 0.32321351766586304,
+      "num_input_tokens_seen": 561052,
+      "step": 1775,
+      "train_runtime": 4734.2775,
+      "train_tokens_per_second": 118.508
+    },
+    {
+      "epoch": 0.08165517241379311,
+      "grad_norm": 7.311210632324219,
+      "learning_rate": 2.255639097744361e-06,
+      "loss": 0.33350884914398193,
+      "num_input_tokens_seen": 561298,
+      "step": 1776,
+      "train_runtime": 4736.2756,
+      "train_tokens_per_second": 118.51
+    },
+    {
+      "epoch": 0.08170114942528736,
+      "grad_norm": 8.70530891418457,
+      "learning_rate": 2.2456140350877195e-06,
+      "loss": 0.42279165983200073,
+      "num_input_tokens_seen": 561548,
+      "step": 1777,
+      "train_runtime": 4738.2288,
+      "train_tokens_per_second": 118.514
+    },
+    {
+      "epoch": 0.08174712643678161,
+      "grad_norm": 7.851328372955322,
+      "learning_rate": 2.235588972431078e-06,
+      "loss": 0.46526530385017395,
+      "num_input_tokens_seen": 561814,
+      "step": 1778,
+      "train_runtime": 4740.2878,
+      "train_tokens_per_second": 118.519
+    },
+    {
+      "epoch": 0.08179310344827587,
+      "grad_norm": 7.489623546600342,
+      "learning_rate": 2.2255639097744365e-06,
+      "loss": 0.4107331931591034,
+      "num_input_tokens_seen": 562144,
+      "step": 1779,
+      "train_runtime": 4742.3276,
+      "train_tokens_per_second": 118.538
+    },
+    {
+      "epoch": 0.08183908045977012,
+      "grad_norm": 8.210550308227539,
+      "learning_rate": 2.2155388471177945e-06,
+      "loss": 0.38923993706703186,
+      "num_input_tokens_seen": 562398,
+      "step": 1780,
+      "train_runtime": 4744.4241,
+      "train_tokens_per_second": 118.539
+    },
+    {
+      "epoch": 0.08188505747126437,
+      "grad_norm": 6.557780742645264,
+      "learning_rate": 2.205513784461153e-06,
+      "loss": 0.3759402930736542,
+      "num_input_tokens_seen": 562630,
+      "step": 1781,
+      "train_runtime": 4746.3925,
+      "train_tokens_per_second": 118.538
+    },
+    {
+      "epoch": 0.08193103448275862,
+      "grad_norm": 6.851749420166016,
+      "learning_rate": 2.1954887218045115e-06,
+      "loss": 0.37134015560150146,
+      "num_input_tokens_seen": 562914,
+      "step": 1782,
+      "train_runtime": 4748.3343,
+      "train_tokens_per_second": 118.55
+    },
+    {
+      "epoch": 0.08197701149425288,
+      "grad_norm": 8.408563613891602,
+      "learning_rate": 2.1854636591478696e-06,
+      "loss": 0.5807152986526489,
+      "num_input_tokens_seen": 563260,
+      "step": 1783,
+      "train_runtime": 4750.3555,
+      "train_tokens_per_second": 118.572
+    },
+    {
+      "epoch": 0.08202298850574713,
+      "grad_norm": 8.055425643920898,
+      "learning_rate": 2.1754385964912285e-06,
+      "loss": 0.43624207377433777,
+      "num_input_tokens_seen": 563564,
+      "step": 1784,
+      "train_runtime": 4752.4042,
+      "train_tokens_per_second": 118.585
+    },
+    {
+      "epoch": 0.08206896551724138,
+      "grad_norm": 11.250126838684082,
+      "learning_rate": 2.1654135338345866e-06,
+      "loss": 0.577454149723053,
+      "num_input_tokens_seen": 563888,
+      "step": 1785,
+      "train_runtime": 4754.339,
+      "train_tokens_per_second": 118.605
+    },
+    {
+      "epoch": 0.08211494252873563,
+      "grad_norm": 7.877490997314453,
+      "learning_rate": 2.155388471177945e-06,
+      "loss": 0.4524950385093689,
+      "num_input_tokens_seen": 564144,
+      "step": 1786,
+      "train_runtime": 4756.2715,
+      "train_tokens_per_second": 118.611
+    },
+    {
+      "epoch": 0.08216091954022989,
+      "grad_norm": 6.374622821807861,
+      "learning_rate": 2.1453634085213035e-06,
+      "loss": 0.361733615398407,
+      "num_input_tokens_seen": 564424,
+      "step": 1787,
+      "train_runtime": 4758.2183,
+      "train_tokens_per_second": 118.621
+    },
+    {
+      "epoch": 0.08220689655172414,
+      "grad_norm": 7.350593090057373,
+      "learning_rate": 2.1353383458646616e-06,
+      "loss": 0.4720909595489502,
+      "num_input_tokens_seen": 564762,
+      "step": 1788,
+      "train_runtime": 4760.1879,
+      "train_tokens_per_second": 118.643
+    },
+    {
+      "epoch": 0.08225287356321839,
+      "grad_norm": 7.337474346160889,
+      "learning_rate": 2.1253132832080205e-06,
+      "loss": 0.586622953414917,
+      "num_input_tokens_seen": 565154,
+      "step": 1789,
+      "train_runtime": 4762.1594,
+      "train_tokens_per_second": 118.676
+    },
+    {
+      "epoch": 0.08229885057471265,
+      "grad_norm": 7.515060901641846,
+      "learning_rate": 2.1152882205513786e-06,
+      "loss": 0.5617174506187439,
+      "num_input_tokens_seen": 565478,
+      "step": 1790,
+      "train_runtime": 4764.1204,
+      "train_tokens_per_second": 118.695
+    },
+    {
+      "epoch": 0.0823448275862069,
+      "grad_norm": 8.223000526428223,
+      "learning_rate": 2.105263157894737e-06,
+      "loss": 0.46494728326797485,
+      "num_input_tokens_seen": 565770,
+      "step": 1791,
+      "train_runtime": 4766.0626,
+      "train_tokens_per_second": 118.708
+    },
+    {
+      "epoch": 0.08239080459770115,
+      "grad_norm": 7.07399320602417,
+      "learning_rate": 2.0952380952380955e-06,
+      "loss": 0.3907439112663269,
+      "num_input_tokens_seen": 566174,
+      "step": 1792,
+      "train_runtime": 4768.1724,
+      "train_tokens_per_second": 118.74
+    },
+    {
+      "epoch": 0.0824367816091954,
+      "grad_norm": 7.540831089019775,
+      "learning_rate": 2.0852130325814536e-06,
+      "loss": 0.437356173992157,
+      "num_input_tokens_seen": 566468,
+      "step": 1793,
+      "train_runtime": 4770.1422,
+      "train_tokens_per_second": 118.753
+    },
+    {
+      "epoch": 0.08248275862068966,
+      "grad_norm": 8.21686840057373,
+      "learning_rate": 2.075187969924812e-06,
+      "loss": 0.5119478106498718,
+      "num_input_tokens_seen": 566840,
+      "step": 1794,
+      "train_runtime": 4772.1541,
+      "train_tokens_per_second": 118.781
+    },
+    {
+      "epoch": 0.08252873563218391,
+      "grad_norm": 8.327613830566406,
+      "learning_rate": 2.0651629072681706e-06,
+      "loss": 0.29968926310539246,
+      "num_input_tokens_seen": 567078,
+      "step": 1795,
+      "train_runtime": 4774.1261,
+      "train_tokens_per_second": 118.782
+    },
+    {
+      "epoch": 0.08257471264367816,
+      "grad_norm": 9.069775581359863,
+      "learning_rate": 2.055137844611529e-06,
+      "loss": 0.4626874029636383,
+      "num_input_tokens_seen": 567354,
+      "step": 1796,
+      "train_runtime": 4776.0735,
+      "train_tokens_per_second": 118.791
+    },
+    {
+      "epoch": 0.08262068965517241,
+      "grad_norm": 7.5980706214904785,
+      "learning_rate": 2.0451127819548876e-06,
+      "loss": 0.39771243929862976,
+      "num_input_tokens_seen": 567706,
+      "step": 1797,
+      "train_runtime": 4778.0678,
+      "train_tokens_per_second": 118.815
+    },
+    {
+      "epoch": 0.08266666666666667,
+      "grad_norm": 7.28903865814209,
+      "learning_rate": 2.0350877192982456e-06,
+      "loss": 0.3334161043167114,
+      "num_input_tokens_seen": 567974,
+      "step": 1798,
+      "train_runtime": 4780.0528,
+      "train_tokens_per_second": 118.822
+    },
+    {
+      "epoch": 0.08271264367816092,
+      "grad_norm": 7.9057536125183105,
+      "learning_rate": 2.025062656641604e-06,
+      "loss": 0.48750609159469604,
+      "num_input_tokens_seen": 568244,
+      "step": 1799,
+      "train_runtime": 4782.0153,
+      "train_tokens_per_second": 118.829
+    },
+    {
+      "epoch": 0.08275862068965517,
+      "grad_norm": 7.5187554359436035,
+      "learning_rate": 2.0150375939849626e-06,
+      "loss": 0.46189796924591064,
+      "num_input_tokens_seen": 568570,
+      "step": 1800,
+      "train_runtime": 4784.0347,
+      "train_tokens_per_second": 118.847
+    },
+    {
+      "epoch": 0.08275862068965517,
+      "eval_loss": 1.6565169095993042,
+      "eval_runtime": 54.3567,
+      "eval_samples_per_second": 18.397,
+      "eval_steps_per_second": 9.198,
+      "num_input_tokens_seen": 568570,
+      "step": 1800
+    },
+    {
+      "epoch": 0.08280459770114942,
+      "grad_norm": 7.315585136413574,
+      "learning_rate": 2.0050125313283207e-06,
+      "loss": 0.3831065595149994,
+      "num_input_tokens_seen": 568850,
+      "step": 1801,
+      "train_runtime": 4853.8606,
+      "train_tokens_per_second": 117.195
+    },
+    {
+      "epoch": 0.08285057471264368,
+      "grad_norm": 8.172098159790039,
+      "learning_rate": 1.9949874686716796e-06,
+      "loss": 0.4771665930747986,
+      "num_input_tokens_seen": 569132,
+      "step": 1802,
+      "train_runtime": 4855.9201,
+      "train_tokens_per_second": 117.204
+    },
+    {
+      "epoch": 0.08289655172413793,
+      "grad_norm": 8.046319007873535,
+      "learning_rate": 1.9849624060150376e-06,
+      "loss": 0.41621360182762146,
+      "num_input_tokens_seen": 569426,
+      "step": 1803,
+      "train_runtime": 4857.9495,
+      "train_tokens_per_second": 117.215
+    },
+    {
+      "epoch": 0.08294252873563218,
+      "grad_norm": 7.207624912261963,
+      "learning_rate": 1.974937343358396e-06,
+      "loss": 0.4766409397125244,
+      "num_input_tokens_seen": 569806,
+      "step": 1804,
+      "train_runtime": 4859.9565,
+      "train_tokens_per_second": 117.245
+    },
+    {
+      "epoch": 0.08298850574712643,
+      "grad_norm": 7.081984996795654,
+      "learning_rate": 1.9649122807017546e-06,
+      "loss": 0.4001272916793823,
+      "num_input_tokens_seen": 570082,
+      "step": 1805,
+      "train_runtime": 4861.9675,
+      "train_tokens_per_second": 117.253
+    },
+    {
+      "epoch": 0.08303448275862069,
+      "grad_norm": 6.745321273803711,
+      "learning_rate": 1.9548872180451127e-06,
+      "loss": 0.3025839328765869,
+      "num_input_tokens_seen": 570326,
+      "step": 1806,
+      "train_runtime": 4863.9405,
+      "train_tokens_per_second": 117.256
+    },
+    {
+      "epoch": 0.08308045977011494,
+      "grad_norm": 7.696715354919434,
+      "learning_rate": 1.9448621553884716e-06,
+      "loss": 0.5356572866439819,
+      "num_input_tokens_seen": 570696,
+      "step": 1807,
+      "train_runtime": 4865.9073,
+      "train_tokens_per_second": 117.285
+    },
+    {
+      "epoch": 0.08312643678160919,
+      "grad_norm": 6.422617435455322,
+      "learning_rate": 1.9348370927318297e-06,
+      "loss": 0.3579423725605011,
+      "num_input_tokens_seen": 571062,
+      "step": 1808,
+      "train_runtime": 4867.8939,
+      "train_tokens_per_second": 117.312
+    },
+    {
+      "epoch": 0.08317241379310344,
+      "grad_norm": 8.224241256713867,
+      "learning_rate": 1.924812030075188e-06,
+      "loss": 0.468651682138443,
+      "num_input_tokens_seen": 571330,
+      "step": 1809,
+      "train_runtime": 4869.8745,
+      "train_tokens_per_second": 117.319
+    },
+    {
+      "epoch": 0.0832183908045977,
+      "grad_norm": 8.08243465423584,
+      "learning_rate": 1.9147869674185466e-06,
+      "loss": 0.4702770709991455,
+      "num_input_tokens_seen": 571606,
+      "step": 1810,
+      "train_runtime": 4871.8252,
+      "train_tokens_per_second": 117.329
+    },
+    {
+      "epoch": 0.08326436781609195,
+      "grad_norm": 7.297043323516846,
+      "learning_rate": 1.904761904761905e-06,
+      "loss": 0.4983699917793274,
+      "num_input_tokens_seen": 572024,
+      "step": 1811,
+      "train_runtime": 4873.8609,
+      "train_tokens_per_second": 117.366
+    },
+    {
+      "epoch": 0.0833103448275862,
+      "grad_norm": 8.084451675415039,
+      "learning_rate": 1.8947368421052634e-06,
+      "loss": 0.5442982316017151,
+      "num_input_tokens_seen": 572358,
+      "step": 1812,
+      "train_runtime": 4875.8308,
+      "train_tokens_per_second": 117.387
+    },
+    {
+      "epoch": 0.08335632183908046,
+      "grad_norm": 9.324164390563965,
+      "learning_rate": 1.8847117794486217e-06,
+      "loss": 0.43849676847457886,
+      "num_input_tokens_seen": 572674,
+      "step": 1813,
+      "train_runtime": 4877.99,
+      "train_tokens_per_second": 117.4
+    },
+    {
+      "epoch": 0.08340229885057471,
+      "grad_norm": 6.999230861663818,
+      "learning_rate": 1.8746867167919802e-06,
+      "loss": 0.4893054664134979,
+      "num_input_tokens_seen": 572970,
+      "step": 1814,
+      "train_runtime": 4879.996,
+      "train_tokens_per_second": 117.412
+    },
+    {
+      "epoch": 0.08344827586206896,
+      "grad_norm": 6.0354228019714355,
+      "learning_rate": 1.8646616541353384e-06,
+      "loss": 0.3791593611240387,
+      "num_input_tokens_seen": 573376,
+      "step": 1815,
+      "train_runtime": 4881.9929,
+      "train_tokens_per_second": 117.447
+    },
+    {
+      "epoch": 0.08349425287356321,
+      "grad_norm": 8.382869720458984,
+      "learning_rate": 1.8546365914786967e-06,
+      "loss": 0.4673730731010437,
+      "num_input_tokens_seen": 573718,
+      "step": 1816,
+      "train_runtime": 4883.9513,
+      "train_tokens_per_second": 117.47
+    },
+    {
+      "epoch": 0.08354022988505747,
+      "grad_norm": 8.546704292297363,
+      "learning_rate": 1.8446115288220554e-06,
+      "loss": 0.5125331878662109,
+      "num_input_tokens_seen": 574024,
+      "step": 1817,
+      "train_runtime": 4885.9807,
+      "train_tokens_per_second": 117.484
+    },
+    {
+      "epoch": 0.08358620689655172,
+      "grad_norm": 7.281277656555176,
+      "learning_rate": 1.8345864661654137e-06,
+      "loss": 0.4075167775154114,
+      "num_input_tokens_seen": 574286,
+      "step": 1818,
+      "train_runtime": 4888.0502,
+      "train_tokens_per_second": 117.488
+    },
+    {
+      "epoch": 0.08363218390804597,
+      "grad_norm": 9.473976135253906,
+      "learning_rate": 1.8245614035087722e-06,
+      "loss": 0.36136457324028015,
+      "num_input_tokens_seen": 574578,
+      "step": 1819,
+      "train_runtime": 4889.9952,
+      "train_tokens_per_second": 117.501
+    },
+    {
+      "epoch": 0.08367816091954022,
+      "grad_norm": 6.530069351196289,
+      "learning_rate": 1.8145363408521305e-06,
+      "loss": 0.3033841848373413,
+      "num_input_tokens_seen": 574868,
+      "step": 1820,
+      "train_runtime": 4891.9461,
+      "train_tokens_per_second": 117.513
+    },
+    {
+      "epoch": 0.08372413793103449,
+      "grad_norm": 6.166937828063965,
+      "learning_rate": 1.8045112781954887e-06,
+      "loss": 0.3236815333366394,
+      "num_input_tokens_seen": 575116,
+      "step": 1821,
+      "train_runtime": 4893.8952,
+      "train_tokens_per_second": 117.517
+    },
+    {
+      "epoch": 0.08377011494252874,
+      "grad_norm": 6.645321369171143,
+      "learning_rate": 1.7944862155388474e-06,
+      "loss": 0.38862884044647217,
+      "num_input_tokens_seen": 575502,
+      "step": 1822,
+      "train_runtime": 4895.8832,
+      "train_tokens_per_second": 117.548
+    },
+    {
+      "epoch": 0.083816091954023,
+      "grad_norm": 6.6288161277771,
+      "learning_rate": 1.7844611528822057e-06,
+      "loss": 0.43700075149536133,
+      "num_input_tokens_seen": 575778,
+      "step": 1823,
+      "train_runtime": 4897.8428,
+      "train_tokens_per_second": 117.557
+    },
+    {
+      "epoch": 0.08386206896551725,
+      "grad_norm": 7.225349426269531,
+      "learning_rate": 1.7744360902255642e-06,
+      "loss": 0.3360759913921356,
+      "num_input_tokens_seen": 576008,
+      "step": 1824,
+      "train_runtime": 4899.8445,
+      "train_tokens_per_second": 117.556
+    },
+    {
+      "epoch": 0.0839080459770115,
+      "grad_norm": 7.799300193786621,
+      "learning_rate": 1.7644110275689225e-06,
+      "loss": 0.5063899159431458,
+      "num_input_tokens_seen": 576312,
+      "step": 1825,
+      "train_runtime": 4901.8442,
+      "train_tokens_per_second": 117.57
+    },
+    {
+      "epoch": 0.08395402298850575,
+      "grad_norm": 7.065634727478027,
+      "learning_rate": 1.7543859649122807e-06,
+      "loss": 0.4996592700481415,
+      "num_input_tokens_seen": 576820,
+      "step": 1826,
+      "train_runtime": 4903.8709,
+      "train_tokens_per_second": 117.625
+    },
+    {
+      "epoch": 0.084,
+      "grad_norm": 7.042445659637451,
+      "learning_rate": 1.7443609022556392e-06,
+      "loss": 0.46579164266586304,
+      "num_input_tokens_seen": 577132,
+      "step": 1827,
+      "train_runtime": 4905.8338,
+      "train_tokens_per_second": 117.642
+    },
+    {
+      "epoch": 0.08404597701149426,
+      "grad_norm": 6.733147621154785,
+      "learning_rate": 1.7343358395989975e-06,
+      "loss": 0.36569154262542725,
+      "num_input_tokens_seen": 577438,
+      "step": 1828,
+      "train_runtime": 4907.8546,
+      "train_tokens_per_second": 117.656
+    },
+    {
+      "epoch": 0.08409195402298851,
+      "grad_norm": 7.03594970703125,
+      "learning_rate": 1.7243107769423562e-06,
+      "loss": 0.5295254588127136,
+      "num_input_tokens_seen": 577744,
+      "step": 1829,
+      "train_runtime": 4909.9602,
+      "train_tokens_per_second": 117.668
+    },
+    {
+      "epoch": 0.08413793103448276,
+      "grad_norm": 7.294963836669922,
+      "learning_rate": 1.7142857142857145e-06,
+      "loss": 0.375788152217865,
+      "num_input_tokens_seen": 578076,
+      "step": 1830,
+      "train_runtime": 4911.939,
+      "train_tokens_per_second": 117.688
+    },
+    {
+      "epoch": 0.08418390804597702,
+      "grad_norm": 7.433239459991455,
+      "learning_rate": 1.7042606516290728e-06,
+      "loss": 0.5001753568649292,
+      "num_input_tokens_seen": 578482,
+      "step": 1831,
+      "train_runtime": 4926.6953,
+      "train_tokens_per_second": 117.418
+    },
+    {
+      "epoch": 0.08422988505747127,
+      "grad_norm": 6.932636260986328,
+      "learning_rate": 1.6942355889724312e-06,
+      "loss": 0.4600294828414917,
+      "num_input_tokens_seen": 578882,
+      "step": 1832,
+      "train_runtime": 4928.7787,
+      "train_tokens_per_second": 117.449
+    },
+    {
+      "epoch": 0.08427586206896552,
+      "grad_norm": 8.25571346282959,
+      "learning_rate": 1.6842105263157895e-06,
+      "loss": 0.5654494166374207,
+      "num_input_tokens_seen": 579196,
+      "step": 1833,
+      "train_runtime": 4930.7354,
+      "train_tokens_per_second": 117.466
+    },
+    {
+      "epoch": 0.08432183908045977,
+      "grad_norm": 7.823970794677734,
+      "learning_rate": 1.6741854636591482e-06,
+      "loss": 0.3812708556652069,
+      "num_input_tokens_seen": 579422,
+      "step": 1834,
+      "train_runtime": 4932.6728,
+      "train_tokens_per_second": 117.466
+    },
+    {
+      "epoch": 0.08436781609195403,
+      "grad_norm": 6.841559886932373,
+      "learning_rate": 1.6641604010025065e-06,
+      "loss": 0.39847031235694885,
+      "num_input_tokens_seen": 579706,
+      "step": 1835,
+      "train_runtime": 4934.6248,
+      "train_tokens_per_second": 117.477
+    },
+    {
+      "epoch": 0.08441379310344828,
+      "grad_norm": 8.656652450561523,
+      "learning_rate": 1.6541353383458648e-06,
+      "loss": 0.43860891461372375,
+      "num_input_tokens_seen": 580002,
+      "step": 1836,
+      "train_runtime": 4936.5727,
+      "train_tokens_per_second": 117.491
+    },
+    {
+      "epoch": 0.08445977011494253,
+      "grad_norm": 7.259665489196777,
+      "learning_rate": 1.6441102756892233e-06,
+      "loss": 0.4278905689716339,
+      "num_input_tokens_seen": 580286,
+      "step": 1837,
+      "train_runtime": 4938.5271,
+      "train_tokens_per_second": 117.502
+    },
+    {
+      "epoch": 0.08450574712643678,
+      "grad_norm": 12.496138572692871,
+      "learning_rate": 1.6340852130325815e-06,
+      "loss": 0.31393980979919434,
+      "num_input_tokens_seen": 580586,
+      "step": 1838,
+      "train_runtime": 4940.4933,
+      "train_tokens_per_second": 117.516
+    },
+    {
+      "epoch": 0.08455172413793104,
+      "grad_norm": 8.727643013000488,
+      "learning_rate": 1.62406015037594e-06,
+      "loss": 0.49319982528686523,
+      "num_input_tokens_seen": 580840,
+      "step": 1839,
+      "train_runtime": 4942.4492,
+      "train_tokens_per_second": 117.521
+    },
+    {
+      "epoch": 0.08459770114942529,
+      "grad_norm": 8.640833854675293,
+      "learning_rate": 1.6140350877192983e-06,
+      "loss": 0.37768036127090454,
+      "num_input_tokens_seen": 581104,
+      "step": 1840,
+      "train_runtime": 4944.4025,
+      "train_tokens_per_second": 117.528
+    },
+    {
+      "epoch": 0.08464367816091954,
+      "grad_norm": 7.445074081420898,
+      "learning_rate": 1.6040100250626568e-06,
+      "loss": 0.49109211564064026,
+      "num_input_tokens_seen": 581510,
+      "step": 1841,
+      "train_runtime": 4946.3893,
+      "train_tokens_per_second": 117.563
+    },
+    {
+      "epoch": 0.0846896551724138,
+      "grad_norm": 7.998035907745361,
+      "learning_rate": 1.5939849624060153e-06,
+      "loss": 0.5314834713935852,
+      "num_input_tokens_seen": 581792,
+      "step": 1842,
+      "train_runtime": 4948.3354,
+      "train_tokens_per_second": 117.573
+    },
+    {
+      "epoch": 0.08473563218390805,
+      "grad_norm": 7.3700456619262695,
+      "learning_rate": 1.5839598997493736e-06,
+      "loss": 0.47153419256210327,
+      "num_input_tokens_seen": 582100,
+      "step": 1843,
+      "train_runtime": 4950.3029,
+      "train_tokens_per_second": 117.589
+    },
+    {
+      "epoch": 0.0847816091954023,
+      "grad_norm": 7.144804954528809,
+      "learning_rate": 1.573934837092732e-06,
+      "loss": 0.35979849100112915,
+      "num_input_tokens_seen": 582374,
+      "step": 1844,
+      "train_runtime": 4952.2579,
+      "train_tokens_per_second": 117.598
+    },
+    {
+      "epoch": 0.08482758620689655,
+      "grad_norm": 7.0776753425598145,
+      "learning_rate": 1.5639097744360903e-06,
+      "loss": 0.4229793846607208,
+      "num_input_tokens_seen": 582718,
+      "step": 1845,
+      "train_runtime": 4954.2512,
+      "train_tokens_per_second": 117.62
+    },
+    {
+      "epoch": 0.0848735632183908,
+      "grad_norm": 8.390013694763184,
+      "learning_rate": 1.5538847117794486e-06,
+      "loss": 0.3618282377719879,
+      "num_input_tokens_seen": 582982,
+      "step": 1846,
+      "train_runtime": 4956.2201,
+      "train_tokens_per_second": 117.626
+    },
+    {
+      "epoch": 0.08491954022988506,
+      "grad_norm": 9.31496524810791,
+      "learning_rate": 1.5438596491228073e-06,
+      "loss": 0.44351887702941895,
+      "num_input_tokens_seen": 583306,
+      "step": 1847,
+      "train_runtime": 4958.2346,
+      "train_tokens_per_second": 117.644
+    },
+    {
+      "epoch": 0.08496551724137931,
+      "grad_norm": 9.846113204956055,
+      "learning_rate": 1.5338345864661656e-06,
+      "loss": 0.5924310088157654,
+      "num_input_tokens_seen": 583566,
+      "step": 1848,
+      "train_runtime": 4960.1792,
+      "train_tokens_per_second": 117.65
+    },
+    {
+      "epoch": 0.08501149425287356,
+      "grad_norm": 7.809129238128662,
+      "learning_rate": 1.523809523809524e-06,
+      "loss": 0.4247511923313141,
+      "num_input_tokens_seen": 583834,
+      "step": 1849,
+      "train_runtime": 4962.1238,
+      "train_tokens_per_second": 117.658
+    },
+    {
+      "epoch": 0.08505747126436781,
+      "grad_norm": 8.016154289245605,
+      "learning_rate": 1.5137844611528823e-06,
+      "loss": 0.543587863445282,
+      "num_input_tokens_seen": 584152,
+      "step": 1850,
+      "train_runtime": 4964.0897,
+      "train_tokens_per_second": 117.676
+    },
+    {
+      "epoch": 0.08510344827586207,
+      "grad_norm": 9.180368423461914,
+      "learning_rate": 1.5037593984962406e-06,
+      "loss": 0.48159167170524597,
+      "num_input_tokens_seen": 584422,
+      "step": 1851,
+      "train_runtime": 4966.0273,
+      "train_tokens_per_second": 117.684
+    },
+    {
+      "epoch": 0.08514942528735632,
+      "grad_norm": 7.4745564460754395,
+      "learning_rate": 1.4937343358395993e-06,
+      "loss": 0.47470101714134216,
+      "num_input_tokens_seen": 584712,
+      "step": 1852,
+      "train_runtime": 4967.9784,
+      "train_tokens_per_second": 117.696
+    },
+    {
+      "epoch": 0.08519540229885057,
+      "grad_norm": 8.902558326721191,
+      "learning_rate": 1.4837092731829576e-06,
+      "loss": 0.44762665033340454,
+      "num_input_tokens_seen": 585014,
+      "step": 1853,
+      "train_runtime": 4969.9345,
+      "train_tokens_per_second": 117.711
+    },
+    {
+      "epoch": 0.08524137931034483,
+      "grad_norm": 7.327677249908447,
+      "learning_rate": 1.4736842105263159e-06,
+      "loss": 0.38496196269989014,
+      "num_input_tokens_seen": 585294,
+      "step": 1854,
+      "train_runtime": 4971.8812,
+      "train_tokens_per_second": 117.721
+    },
+    {
+      "epoch": 0.08528735632183908,
+      "grad_norm": 8.085262298583984,
+      "learning_rate": 1.4636591478696743e-06,
+      "loss": 0.3590407967567444,
+      "num_input_tokens_seen": 585560,
+      "step": 1855,
+      "train_runtime": 4973.8233,
+      "train_tokens_per_second": 117.728
+    },
+    {
+      "epoch": 0.08533333333333333,
+      "grad_norm": 7.7812676429748535,
+      "learning_rate": 1.4536340852130326e-06,
+      "loss": 0.38731035590171814,
+      "num_input_tokens_seen": 585866,
+      "step": 1856,
+      "train_runtime": 4975.8074,
+      "train_tokens_per_second": 117.743
+    },
+    {
+      "epoch": 0.08537931034482758,
+      "grad_norm": 8.900866508483887,
+      "learning_rate": 1.4436090225563911e-06,
+      "loss": 0.5512892603874207,
+      "num_input_tokens_seen": 586138,
+      "step": 1857,
+      "train_runtime": 4977.7921,
+      "train_tokens_per_second": 117.751
+    },
+    {
+      "epoch": 0.08542528735632184,
+      "grad_norm": 7.379859924316406,
+      "learning_rate": 1.4335839598997494e-06,
+      "loss": 0.5174167156219482,
+      "num_input_tokens_seen": 586422,
+      "step": 1858,
+      "train_runtime": 4979.7679,
+      "train_tokens_per_second": 117.761
+    },
+    {
+      "epoch": 0.08547126436781609,
+      "grad_norm": 6.956198692321777,
+      "learning_rate": 1.4235588972431077e-06,
+      "loss": 0.39228907227516174,
+      "num_input_tokens_seen": 586712,
+      "step": 1859,
+      "train_runtime": 4981.7305,
+      "train_tokens_per_second": 117.773
+    },
+    {
+      "epoch": 0.08551724137931034,
+      "grad_norm": 7.236478328704834,
+      "learning_rate": 1.4135338345864664e-06,
+      "loss": 0.36122918128967285,
+      "num_input_tokens_seen": 586952,
+      "step": 1860,
+      "train_runtime": 4983.6668,
+      "train_tokens_per_second": 117.775
+    },
+    {
+      "epoch": 0.0855632183908046,
+      "grad_norm": 8.94511604309082,
+      "learning_rate": 1.4035087719298246e-06,
+      "loss": 0.5683632493019104,
+      "num_input_tokens_seen": 587284,
+      "step": 1861,
+      "train_runtime": 4998.5733,
+      "train_tokens_per_second": 117.49
+    },
+    {
+      "epoch": 0.08560919540229885,
+      "grad_norm": 8.432382583618164,
+      "learning_rate": 1.3934837092731831e-06,
+      "loss": 0.4612915813922882,
+      "num_input_tokens_seen": 587680,
+      "step": 1862,
+      "train_runtime": 5000.6149,
+      "train_tokens_per_second": 117.522
+    },
+    {
+      "epoch": 0.0856551724137931,
+      "grad_norm": 7.545629501342773,
+      "learning_rate": 1.3834586466165414e-06,
+      "loss": 0.3651968836784363,
+      "num_input_tokens_seen": 587916,
+      "step": 1863,
+      "train_runtime": 5002.5899,
+      "train_tokens_per_second": 117.522
+    },
+    {
+      "epoch": 0.08570114942528735,
+      "grad_norm": 8.204903602600098,
+      "learning_rate": 1.3734335839598997e-06,
+      "loss": 0.2659253478050232,
+      "num_input_tokens_seen": 588150,
+      "step": 1864,
+      "train_runtime": 5004.5266,
+      "train_tokens_per_second": 117.524
+    },
+    {
+      "epoch": 0.0857471264367816,
+      "grad_norm": 6.838364124298096,
+      "learning_rate": 1.3634085213032584e-06,
+      "loss": 0.3213936984539032,
+      "num_input_tokens_seen": 588402,
+      "step": 1865,
+      "train_runtime": 5006.467,
+      "train_tokens_per_second": 117.528
+    },
+    {
+      "epoch": 0.08579310344827586,
+      "grad_norm": 6.686530590057373,
+      "learning_rate": 1.3533834586466167e-06,
+      "loss": 0.49802929162979126,
+      "num_input_tokens_seen": 589038,
+      "step": 1866,
+      "train_runtime": 5008.5146,
+      "train_tokens_per_second": 117.607
+    },
+    {
+      "epoch": 0.08583908045977011,
+      "grad_norm": 7.597280025482178,
+      "learning_rate": 1.3433583959899751e-06,
+      "loss": 0.47507423162460327,
+      "num_input_tokens_seen": 589412,
+      "step": 1867,
+      "train_runtime": 5010.5024,
+      "train_tokens_per_second": 117.635
+    },
+    {
+      "epoch": 0.08588505747126436,
+      "grad_norm": 8.68179988861084,
+      "learning_rate": 1.3333333333333334e-06,
+      "loss": 0.4467782974243164,
+      "num_input_tokens_seen": 589702,
+      "step": 1868,
+      "train_runtime": 5012.4823,
+      "train_tokens_per_second": 117.647
+    },
+    {
+      "epoch": 0.08593103448275861,
+      "grad_norm": 7.25511360168457,
+      "learning_rate": 1.3233082706766917e-06,
+      "loss": 0.4221172332763672,
+      "num_input_tokens_seen": 589960,
+      "step": 1869,
+      "train_runtime": 5014.4225,
+      "train_tokens_per_second": 117.653
+    },
+    {
+      "epoch": 0.08597701149425287,
+      "grad_norm": 8.154078483581543,
+      "learning_rate": 1.3132832080200502e-06,
+      "loss": 0.5756381154060364,
+      "num_input_tokens_seen": 590244,
+      "step": 1870,
+      "train_runtime": 5016.3608,
+      "train_tokens_per_second": 117.664
+    },
+    {
+      "epoch": 0.08602298850574713,
+      "grad_norm": 9.70241928100586,
+      "learning_rate": 1.3032581453634085e-06,
+      "loss": 0.3953090310096741,
+      "num_input_tokens_seen": 590474,
+      "step": 1871,
+      "train_runtime": 5018.3019,
+      "train_tokens_per_second": 117.664
+    },
+    {
+      "epoch": 0.08606896551724139,
+      "grad_norm": 7.584150791168213,
+      "learning_rate": 1.2932330827067672e-06,
+      "loss": 0.41417622566223145,
+      "num_input_tokens_seen": 590734,
+      "step": 1872,
+      "train_runtime": 5020.2562,
+      "train_tokens_per_second": 117.67
+    },
+    {
+      "epoch": 0.08611494252873564,
+      "grad_norm": 8.666409492492676,
+      "learning_rate": 1.2832080200501254e-06,
+      "loss": 0.37340855598449707,
+      "num_input_tokens_seen": 590990,
+      "step": 1873,
+      "train_runtime": 5022.4241,
+      "train_tokens_per_second": 117.67
+    },
+    {
+      "epoch": 0.08616091954022989,
+      "grad_norm": 6.290865421295166,
+      "learning_rate": 1.2731829573934837e-06,
+      "loss": 0.23415996134281158,
+      "num_input_tokens_seen": 591214,
+      "step": 1874,
+      "train_runtime": 5024.4497,
+      "train_tokens_per_second": 117.667
+    },
+    {
+      "epoch": 0.08620689655172414,
+      "grad_norm": 6.804515838623047,
+      "learning_rate": 1.2631578947368422e-06,
+      "loss": 0.4050792157649994,
+      "num_input_tokens_seen": 591598,
+      "step": 1875,
+      "train_runtime": 5026.4811,
+      "train_tokens_per_second": 117.696
+    },
+    {
+      "epoch": 0.0862528735632184,
+      "grad_norm": 6.425148963928223,
+      "learning_rate": 1.2531328320802005e-06,
+      "loss": 0.3941546082496643,
+      "num_input_tokens_seen": 591914,
+      "step": 1876,
+      "train_runtime": 5028.4398,
+      "train_tokens_per_second": 117.713
+    },
+    {
+      "epoch": 0.08629885057471265,
+      "grad_norm": 7.374588489532471,
+      "learning_rate": 1.243107769423559e-06,
+      "loss": 0.4062543213367462,
+      "num_input_tokens_seen": 592212,
+      "step": 1877,
+      "train_runtime": 5030.3954,
+      "train_tokens_per_second": 117.727
+    },
+    {
+      "epoch": 0.0863448275862069,
+      "grad_norm": 8.183197021484375,
+      "learning_rate": 1.2330827067669174e-06,
+      "loss": 0.5620994567871094,
+      "num_input_tokens_seen": 592544,
+      "step": 1878,
+      "train_runtime": 5032.3515,
+      "train_tokens_per_second": 117.747
+    },
+    {
+      "epoch": 0.08639080459770115,
+      "grad_norm": 6.94822883605957,
+      "learning_rate": 1.2230576441102757e-06,
+      "loss": 0.4957292377948761,
+      "num_input_tokens_seen": 593032,
+      "step": 1879,
+      "train_runtime": 5034.4133,
+      "train_tokens_per_second": 117.796
+    },
+    {
+      "epoch": 0.0864367816091954,
+      "grad_norm": 7.928312301635742,
+      "learning_rate": 1.2130325814536342e-06,
+      "loss": 0.4197303354740143,
+      "num_input_tokens_seen": 593330,
+      "step": 1880,
+      "train_runtime": 5036.3615,
+      "train_tokens_per_second": 117.809
+    },
+    {
+      "epoch": 0.08648275862068966,
+      "grad_norm": 9.07723331451416,
+      "learning_rate": 1.2030075187969925e-06,
+      "loss": 0.5152315497398376,
+      "num_input_tokens_seen": 593664,
+      "step": 1881,
+      "train_runtime": 5038.3243,
+      "train_tokens_per_second": 117.83
+    },
+    {
+      "epoch": 0.08652873563218391,
+      "grad_norm": 7.6124162673950195,
+      "learning_rate": 1.192982456140351e-06,
+      "loss": 0.47064727544784546,
+      "num_input_tokens_seen": 593956,
+      "step": 1882,
+      "train_runtime": 5040.2788,
+      "train_tokens_per_second": 117.842
+    },
+    {
+      "epoch": 0.08657471264367816,
+      "grad_norm": 7.852885723114014,
+      "learning_rate": 1.1829573934837095e-06,
+      "loss": 0.5047283172607422,
+      "num_input_tokens_seen": 594282,
+      "step": 1883,
+      "train_runtime": 5042.2495,
+      "train_tokens_per_second": 117.86
+    },
+    {
+      "epoch": 0.08662068965517242,
+      "grad_norm": 7.417337894439697,
+      "learning_rate": 1.1729323308270677e-06,
+      "loss": 0.3724783658981323,
+      "num_input_tokens_seen": 594646,
+      "step": 1884,
+      "train_runtime": 5044.2823,
+      "train_tokens_per_second": 117.885
+    },
+    {
+      "epoch": 0.08666666666666667,
+      "grad_norm": 7.350129127502441,
+      "learning_rate": 1.1629072681704262e-06,
+      "loss": 0.35512396693229675,
+      "num_input_tokens_seen": 594942,
+      "step": 1885,
+      "train_runtime": 5046.2716,
+      "train_tokens_per_second": 117.897
+    },
+    {
+      "epoch": 0.08671264367816092,
+      "grad_norm": 7.525505542755127,
+      "learning_rate": 1.1528822055137845e-06,
+      "loss": 0.33822762966156006,
+      "num_input_tokens_seen": 595250,
+      "step": 1886,
+      "train_runtime": 5048.229,
+      "train_tokens_per_second": 117.913
+    },
+    {
+      "epoch": 0.08675862068965517,
+      "grad_norm": 7.463708877563477,
+      "learning_rate": 1.142857142857143e-06,
+      "loss": 0.5150831937789917,
+      "num_input_tokens_seen": 595512,
+      "step": 1887,
+      "train_runtime": 5050.1689,
+      "train_tokens_per_second": 117.919
+    },
+    {
+      "epoch": 0.08680459770114943,
+      "grad_norm": 8.473981857299805,
+      "learning_rate": 1.1328320802005013e-06,
+      "loss": 0.5618161559104919,
+      "num_input_tokens_seen": 595840,
+      "step": 1888,
+      "train_runtime": 5052.1223,
+      "train_tokens_per_second": 117.939
+    },
+    {
+      "epoch": 0.08685057471264368,
+      "grad_norm": 7.199127197265625,
+      "learning_rate": 1.1228070175438598e-06,
+      "loss": 0.48283204436302185,
+      "num_input_tokens_seen": 596104,
+      "step": 1889,
+      "train_runtime": 5054.0612,
+      "train_tokens_per_second": 117.946
+    },
+    {
+      "epoch": 0.08689655172413793,
+      "grad_norm": 7.588725566864014,
+      "learning_rate": 1.1127819548872182e-06,
+      "loss": 0.4118557274341583,
+      "num_input_tokens_seen": 596476,
+      "step": 1890,
+      "train_runtime": 5056.0368,
+      "train_tokens_per_second": 117.973
+    },
+    {
+      "epoch": 0.08694252873563219,
+      "grad_norm": 8.241991996765137,
+      "learning_rate": 1.1027568922305765e-06,
+      "loss": 0.45901328325271606,
+      "num_input_tokens_seen": 596710,
+      "step": 1891,
+      "train_runtime": 5072.0845,
+      "train_tokens_per_second": 117.646
+    },
+    {
+      "epoch": 0.08698850574712644,
+      "grad_norm": 7.463146686553955,
+      "learning_rate": 1.0927318295739348e-06,
+      "loss": 0.3588930368423462,
+      "num_input_tokens_seen": 596952,
+      "step": 1892,
+      "train_runtime": 5074.1848,
+      "train_tokens_per_second": 117.645
+    },
+    {
+      "epoch": 0.08703448275862069,
+      "grad_norm": 7.683949947357178,
+      "learning_rate": 1.0827067669172933e-06,
+      "loss": 0.3187759518623352,
+      "num_input_tokens_seen": 597184,
+      "step": 1893,
+      "train_runtime": 5076.1802,
+      "train_tokens_per_second": 117.644
+    },
+    {
+      "epoch": 0.08708045977011494,
+      "grad_norm": 8.679635047912598,
+      "learning_rate": 1.0726817042606518e-06,
+      "loss": 0.5397830009460449,
+      "num_input_tokens_seen": 597508,
+      "step": 1894,
+      "train_runtime": 5078.1416,
+      "train_tokens_per_second": 117.663
+    },
+    {
+      "epoch": 0.0871264367816092,
+      "grad_norm": 8.731266975402832,
+      "learning_rate": 1.0626566416040103e-06,
+      "loss": 0.4531298875808716,
+      "num_input_tokens_seen": 597800,
+      "step": 1895,
+      "train_runtime": 5080.1004,
+      "train_tokens_per_second": 117.675
+    },
+    {
+      "epoch": 0.08717241379310345,
+      "grad_norm": 7.020224094390869,
+      "learning_rate": 1.0526315789473685e-06,
+      "loss": 0.3842434883117676,
+      "num_input_tokens_seen": 598136,
+      "step": 1896,
+      "train_runtime": 5082.1025,
+      "train_tokens_per_second": 117.695
+    },
+    {
+      "epoch": 0.0872183908045977,
+      "grad_norm": 8.90903377532959,
+      "learning_rate": 1.0426065162907268e-06,
+      "loss": 0.5930857062339783,
+      "num_input_tokens_seen": 598476,
+      "step": 1897,
+      "train_runtime": 5084.0886,
+      "train_tokens_per_second": 117.715
+    },
+    {
+      "epoch": 0.08726436781609195,
+      "grad_norm": 6.783843517303467,
+      "learning_rate": 1.0325814536340853e-06,
+      "loss": 0.32657402753829956,
+      "num_input_tokens_seen": 598712,
+      "step": 1898,
+      "train_runtime": 5086.0794,
+      "train_tokens_per_second": 117.716
+    },
+    {
+      "epoch": 0.0873103448275862,
+      "grad_norm": 7.516112804412842,
+      "learning_rate": 1.0225563909774438e-06,
+      "loss": 0.5041102766990662,
+      "num_input_tokens_seen": 599018,
+      "step": 1899,
+      "train_runtime": 5088.0885,
+      "train_tokens_per_second": 117.729
+    },
+    {
+      "epoch": 0.08735632183908046,
+      "grad_norm": 7.043312072753906,
+      "learning_rate": 1.012531328320802e-06,
+      "loss": 0.43204811215400696,
+      "num_input_tokens_seen": 599366,
+      "step": 1900,
+      "train_runtime": 5090.0766,
+      "train_tokens_per_second": 117.752
+    },
+    {
+      "epoch": 0.08740229885057471,
+      "grad_norm": 6.694372177124023,
+      "learning_rate": 1.0025062656641603e-06,
+      "loss": 0.33615103363990784,
+      "num_input_tokens_seen": 599606,
+      "step": 1901,
+      "train_runtime": 5092.0391,
+      "train_tokens_per_second": 117.754
+    },
+    {
+      "epoch": 0.08744827586206896,
+      "grad_norm": 8.443764686584473,
+      "learning_rate": 9.924812030075188e-07,
+      "loss": 0.5095217823982239,
+      "num_input_tokens_seen": 599936,
+      "step": 1902,
+      "train_runtime": 5094.0233,
+      "train_tokens_per_second": 117.773
+    },
+    {
+      "epoch": 0.08749425287356322,
+      "grad_norm": 7.17056131362915,
+      "learning_rate": 9.824561403508773e-07,
+      "loss": 0.5088427662849426,
+      "num_input_tokens_seen": 600434,
+      "step": 1903,
+      "train_runtime": 5096.0214,
+      "train_tokens_per_second": 117.824
+    },
+    {
+      "epoch": 0.08754022988505747,
+      "grad_norm": 8.491886138916016,
+      "learning_rate": 9.724310776942358e-07,
+      "loss": 0.5271230936050415,
+      "num_input_tokens_seen": 600742,
+      "step": 1904,
+      "train_runtime": 5098.0065,
+      "train_tokens_per_second": 117.839
+    },
+    {
+      "epoch": 0.08758620689655172,
+      "grad_norm": 7.883543968200684,
+      "learning_rate": 9.62406015037594e-07,
+      "loss": 0.3756459951400757,
+      "num_input_tokens_seen": 601022,
+      "step": 1905,
+      "train_runtime": 5100.0065,
+      "train_tokens_per_second": 117.847
+    },
+    {
+      "epoch": 0.08763218390804597,
+      "grad_norm": 8.870765686035156,
+      "learning_rate": 9.523809523809525e-07,
+      "loss": 0.5292370319366455,
+      "num_input_tokens_seen": 601344,
+      "step": 1906,
+      "train_runtime": 5102.0018,
+      "train_tokens_per_second": 117.864
+    },
+    {
+      "epoch": 0.08767816091954023,
+      "grad_norm": 8.045775413513184,
+      "learning_rate": 9.423558897243108e-07,
+      "loss": 0.37872469425201416,
+      "num_input_tokens_seen": 601664,
+      "step": 1907,
+      "train_runtime": 5104.0149,
+      "train_tokens_per_second": 117.881
+    },
+    {
+      "epoch": 0.08772413793103448,
+      "grad_norm": 6.757016658782959,
+      "learning_rate": 9.323308270676692e-07,
+      "loss": 0.42241162061691284,
+      "num_input_tokens_seen": 601984,
+      "step": 1908,
+      "train_runtime": 5106.0276,
+      "train_tokens_per_second": 117.897
+    },
+    {
+      "epoch": 0.08777011494252873,
+      "grad_norm": 7.576857089996338,
+      "learning_rate": 9.223057644110277e-07,
+      "loss": 0.3688413202762604,
+      "num_input_tokens_seen": 602248,
+      "step": 1909,
+      "train_runtime": 5108.0042,
+      "train_tokens_per_second": 117.903
+    },
+    {
+      "epoch": 0.08781609195402298,
+      "grad_norm": 8.834125518798828,
+      "learning_rate": 9.122807017543861e-07,
+      "loss": 0.41126561164855957,
+      "num_input_tokens_seen": 602624,
+      "step": 1910,
+      "train_runtime": 5110.1223,
+      "train_tokens_per_second": 117.928
+    },
+    {
+      "epoch": 0.08786206896551724,
+      "grad_norm": 6.076545238494873,
+      "learning_rate": 9.022556390977444e-07,
+      "loss": 0.34722718596458435,
+      "num_input_tokens_seen": 602950,
+      "step": 1911,
+      "train_runtime": 5112.1099,
+      "train_tokens_per_second": 117.945
+    },
+    {
+      "epoch": 0.08790804597701149,
+      "grad_norm": 7.597630023956299,
+      "learning_rate": 8.922305764411029e-07,
+      "loss": 0.5116064548492432,
+      "num_input_tokens_seen": 603286,
+      "step": 1912,
+      "train_runtime": 5114.1567,
+      "train_tokens_per_second": 117.964
+    },
+    {
+      "epoch": 0.08795402298850574,
+      "grad_norm": 7.749081134796143,
+      "learning_rate": 8.822055137844612e-07,
+      "loss": 0.39845266938209534,
+      "num_input_tokens_seen": 603570,
+      "step": 1913,
+      "train_runtime": 5116.1182,
+      "train_tokens_per_second": 117.974
+    },
+    {
+      "epoch": 0.088,
+      "grad_norm": 7.113033294677734,
+      "learning_rate": 8.721804511278196e-07,
+      "loss": 0.3241654336452484,
+      "num_input_tokens_seen": 603972,
+      "step": 1914,
+      "train_runtime": 5118.1066,
+      "train_tokens_per_second": 118.007
+    },
+    {
+      "epoch": 0.08804597701149425,
+      "grad_norm": 8.191364288330078,
+      "learning_rate": 8.621553884711781e-07,
+      "loss": 0.4271330237388611,
+      "num_input_tokens_seen": 604316,
+      "step": 1915,
+      "train_runtime": 5120.072,
+      "train_tokens_per_second": 118.029
+    },
+    {
+      "epoch": 0.0880919540229885,
+      "grad_norm": 7.335124492645264,
+      "learning_rate": 8.521303258145364e-07,
+      "loss": 0.3741306960582733,
+      "num_input_tokens_seen": 604654,
+      "step": 1916,
+      "train_runtime": 5122.0937,
+      "train_tokens_per_second": 118.048
+    },
+    {
+      "epoch": 0.08813793103448275,
+      "grad_norm": 7.051912307739258,
+      "learning_rate": 8.421052631578948e-07,
+      "loss": 0.3410656750202179,
+      "num_input_tokens_seen": 604896,
+      "step": 1917,
+      "train_runtime": 5124.1377,
+      "train_tokens_per_second": 118.048
+    },
+    {
+      "epoch": 0.088183908045977,
+      "grad_norm": 7.493203639984131,
+      "learning_rate": 8.320802005012532e-07,
+      "loss": 0.5200784206390381,
+      "num_input_tokens_seen": 605164,
+      "step": 1918,
+      "train_runtime": 5126.1371,
+      "train_tokens_per_second": 118.055
+    },
+    {
+      "epoch": 0.08822988505747126,
+      "grad_norm": 7.78281307220459,
+      "learning_rate": 8.220551378446116e-07,
+      "loss": 0.4933725893497467,
+      "num_input_tokens_seen": 605422,
+      "step": 1919,
+      "train_runtime": 5128.0848,
+      "train_tokens_per_second": 118.06
+    },
+    {
+      "epoch": 0.08827586206896551,
+      "grad_norm": 7.283194541931152,
+      "learning_rate": 8.1203007518797e-07,
+      "loss": 0.3743380010128021,
+      "num_input_tokens_seen": 605828,
+      "step": 1920,
+      "train_runtime": 5130.1645,
+      "train_tokens_per_second": 118.091
+    },
+    {
+      "epoch": 0.08832183908045978,
+      "grad_norm": 7.83817720413208,
+      "learning_rate": 8.020050125313284e-07,
+      "loss": 0.5046571493148804,
+      "num_input_tokens_seen": 606094,
+      "step": 1921,
+      "train_runtime": 5144.6477,
+      "train_tokens_per_second": 117.811
+    },
+    {
+      "epoch": 0.08836781609195403,
+      "grad_norm": 6.266878604888916,
+      "learning_rate": 7.919799498746868e-07,
+      "loss": 0.3677210211753845,
+      "num_input_tokens_seen": 606382,
+      "step": 1922,
+      "train_runtime": 5146.6131,
+      "train_tokens_per_second": 117.822
+    },
+    {
+      "epoch": 0.08841379310344828,
+      "grad_norm": 7.051089763641357,
+      "learning_rate": 7.819548872180452e-07,
+      "loss": 0.3863835036754608,
+      "num_input_tokens_seen": 606682,
+      "step": 1923,
+      "train_runtime": 5148.556,
+      "train_tokens_per_second": 117.835
+    },
+    {
+      "epoch": 0.08845977011494253,
+      "grad_norm": 8.699736595153809,
+      "learning_rate": 7.719298245614036e-07,
+      "loss": 0.47785136103630066,
+      "num_input_tokens_seen": 607000,
+      "step": 1924,
+      "train_runtime": 5150.5086,
+      "train_tokens_per_second": 117.852
+    },
+    {
+      "epoch": 0.08850574712643679,
+      "grad_norm": 7.745916366577148,
+      "learning_rate": 7.61904761904762e-07,
+      "loss": 0.4454568028450012,
+      "num_input_tokens_seen": 607248,
+      "step": 1925,
+      "train_runtime": 5152.4522,
+      "train_tokens_per_second": 117.856
+    },
+    {
+      "epoch": 0.08855172413793104,
+      "grad_norm": 9.198076248168945,
+      "learning_rate": 7.518796992481203e-07,
+      "loss": 0.3640105724334717,
+      "num_input_tokens_seen": 607568,
+      "step": 1926,
+      "train_runtime": 5154.403,
+      "train_tokens_per_second": 117.874
+    },
+    {
+      "epoch": 0.08859770114942529,
+      "grad_norm": 6.8515729904174805,
+      "learning_rate": 7.418546365914788e-07,
+      "loss": 0.3322884440422058,
+      "num_input_tokens_seen": 607936,
+      "step": 1927,
+      "train_runtime": 5156.3833,
+      "train_tokens_per_second": 117.9
+    },
+    {
+      "epoch": 0.08864367816091955,
+      "grad_norm": 6.681099891662598,
+      "learning_rate": 7.318295739348372e-07,
+      "loss": 0.5460650324821472,
+      "num_input_tokens_seen": 608304,
+      "step": 1928,
+      "train_runtime": 5158.3551,
+      "train_tokens_per_second": 117.926
+    },
+    {
+      "epoch": 0.0886896551724138,
+      "grad_norm": 7.97550106048584,
+      "learning_rate": 7.218045112781956e-07,
+      "loss": 0.42101067304611206,
+      "num_input_tokens_seen": 608522,
+      "step": 1929,
+      "train_runtime": 5160.2828,
+      "train_tokens_per_second": 117.924
+    },
+    {
+      "epoch": 0.08873563218390805,
+      "grad_norm": 5.614116191864014,
+      "learning_rate": 7.117794486215538e-07,
+      "loss": 0.3722851574420929,
+      "num_input_tokens_seen": 609350,
+      "step": 1930,
+      "train_runtime": 5162.3601,
+      "train_tokens_per_second": 118.037
+    },
+    {
+      "epoch": 0.0887816091954023,
+      "grad_norm": 7.282567977905273,
+      "learning_rate": 7.017543859649123e-07,
+      "loss": 0.49573031067848206,
+      "num_input_tokens_seen": 609714,
+      "step": 1931,
+      "train_runtime": 5164.3322,
+      "train_tokens_per_second": 118.063
+    },
+    {
+      "epoch": 0.08882758620689656,
+      "grad_norm": 8.980059623718262,
+      "learning_rate": 6.917293233082707e-07,
+      "loss": 0.545059084892273,
+      "num_input_tokens_seen": 610050,
+      "step": 1932,
+      "train_runtime": 5166.3002,
+      "train_tokens_per_second": 118.083
+    },
+    {
+      "epoch": 0.08887356321839081,
+      "grad_norm": 7.481197357177734,
+      "learning_rate": 6.817042606516292e-07,
+      "loss": 0.5490928888320923,
+      "num_input_tokens_seen": 610360,
+      "step": 1933,
+      "train_runtime": 5168.2608,
+      "train_tokens_per_second": 118.098
+    },
+    {
+      "epoch": 0.08891954022988506,
+      "grad_norm": 7.990349769592285,
+      "learning_rate": 6.716791979949876e-07,
+      "loss": 0.5157485604286194,
+      "num_input_tokens_seen": 610642,
+      "step": 1934,
+      "train_runtime": 5170.231,
+      "train_tokens_per_second": 118.107
+    },
+    {
+      "epoch": 0.08896551724137931,
+      "grad_norm": 8.92601203918457,
+      "learning_rate": 6.616541353383458e-07,
+      "loss": 0.349660187959671,
+      "num_input_tokens_seen": 610958,
+      "step": 1935,
+      "train_runtime": 5172.3834,
+      "train_tokens_per_second": 118.119
+    },
+    {
+      "epoch": 0.08901149425287357,
+      "grad_norm": 6.907639980316162,
+      "learning_rate": 6.516290726817042e-07,
+      "loss": 0.4537245035171509,
+      "num_input_tokens_seen": 611284,
+      "step": 1936,
+      "train_runtime": 5174.3828,
+      "train_tokens_per_second": 118.137
+    },
+    {
+      "epoch": 0.08905747126436782,
+      "grad_norm": 8.429737091064453,
+      "learning_rate": 6.416040100250627e-07,
+      "loss": 0.49810856580734253,
+      "num_input_tokens_seen": 611564,
+      "step": 1937,
+      "train_runtime": 5176.3494,
+      "train_tokens_per_second": 118.146
+    },
+    {
+      "epoch": 0.08910344827586207,
+      "grad_norm": 9.396894454956055,
+      "learning_rate": 6.315789473684211e-07,
+      "loss": 0.4992244839668274,
+      "num_input_tokens_seen": 611804,
+      "step": 1938,
+      "train_runtime": 5178.3308,
+      "train_tokens_per_second": 118.147
+    },
+    {
+      "epoch": 0.08914942528735632,
+      "grad_norm": 8.50183391571045,
+      "learning_rate": 6.215538847117795e-07,
+      "loss": 0.389660120010376,
+      "num_input_tokens_seen": 612092,
+      "step": 1939,
+      "train_runtime": 5180.2942,
+      "train_tokens_per_second": 118.158
+    },
+    {
+      "epoch": 0.08919540229885058,
+      "grad_norm": 7.522252559661865,
+      "learning_rate": 6.115288220551379e-07,
+      "loss": 0.4363986849784851,
+      "num_input_tokens_seen": 612416,
+      "step": 1940,
+      "train_runtime": 5182.2591,
+      "train_tokens_per_second": 118.175
+    },
+    {
+      "epoch": 0.08924137931034483,
+      "grad_norm": 7.143253326416016,
+      "learning_rate": 6.015037593984962e-07,
+      "loss": 0.4299575388431549,
+      "num_input_tokens_seen": 612742,
+      "step": 1941,
+      "train_runtime": 5184.2142,
+      "train_tokens_per_second": 118.194
+    },
+    {
+      "epoch": 0.08928735632183908,
+      "grad_norm": 7.878231525421143,
+      "learning_rate": 5.914786967418547e-07,
+      "loss": 0.5424295663833618,
+      "num_input_tokens_seen": 613132,
+      "step": 1942,
+      "train_runtime": 5186.1972,
+      "train_tokens_per_second": 118.224
+    },
+    {
+      "epoch": 0.08933333333333333,
+      "grad_norm": 7.318743705749512,
+      "learning_rate": 5.814536340852131e-07,
+      "loss": 0.3889036476612091,
+      "num_input_tokens_seen": 613416,
+      "step": 1943,
+      "train_runtime": 5188.1402,
+      "train_tokens_per_second": 118.234
+    },
+    {
+      "epoch": 0.08937931034482759,
+      "grad_norm": 7.778266429901123,
+      "learning_rate": 5.714285714285715e-07,
+      "loss": 0.44363105297088623,
+      "num_input_tokens_seen": 613664,
+      "step": 1944,
+      "train_runtime": 5190.0768,
+      "train_tokens_per_second": 118.238
+    },
+    {
+      "epoch": 0.08942528735632184,
+      "grad_norm": 8.16820240020752,
+      "learning_rate": 5.614035087719299e-07,
+      "loss": 0.4059368371963501,
+      "num_input_tokens_seen": 613982,
+      "step": 1945,
+      "train_runtime": 5192.0434,
+      "train_tokens_per_second": 118.254
+    },
+    {
+      "epoch": 0.08947126436781609,
+      "grad_norm": 7.664226055145264,
+      "learning_rate": 5.513784461152883e-07,
+      "loss": 0.4748205542564392,
+      "num_input_tokens_seen": 614360,
+      "step": 1946,
+      "train_runtime": 5193.9873,
+      "train_tokens_per_second": 118.283
+    },
+    {
+      "epoch": 0.08951724137931034,
+      "grad_norm": 8.522714614868164,
+      "learning_rate": 5.413533834586466e-07,
+      "loss": 0.3064422607421875,
+      "num_input_tokens_seen": 614600,
+      "step": 1947,
+      "train_runtime": 5195.9205,
+      "train_tokens_per_second": 118.285
+    },
+    {
+      "epoch": 0.0895632183908046,
+      "grad_norm": 6.944787502288818,
+      "learning_rate": 5.313283208020051e-07,
+      "loss": 0.5053823590278625,
+      "num_input_tokens_seen": 614884,
+      "step": 1948,
+      "train_runtime": 5197.8832,
+      "train_tokens_per_second": 118.295
+    },
+    {
+      "epoch": 0.08960919540229885,
+      "grad_norm": 7.006589889526367,
+      "learning_rate": 5.213032581453634e-07,
+      "loss": 0.5162798166275024,
+      "num_input_tokens_seen": 615168,
+      "step": 1949,
+      "train_runtime": 5199.824,
+      "train_tokens_per_second": 118.306
+    },
+    {
+      "epoch": 0.0896551724137931,
+      "grad_norm": 7.0208587646484375,
+      "learning_rate": 5.112781954887219e-07,
+      "loss": 0.3463684618473053,
+      "num_input_tokens_seen": 615406,
+      "step": 1950,
+      "train_runtime": 5201.8173,
+      "train_tokens_per_second": 118.306
+    },
+    {
+      "epoch": 0.08970114942528735,
+      "grad_norm": 8.017606735229492,
+      "learning_rate": 5.012531328320802e-07,
+      "loss": 0.4508306682109833,
+      "num_input_tokens_seen": 615704,
+      "step": 1951,
+      "train_runtime": 5216.5198,
+      "train_tokens_per_second": 118.03
+    },
+    {
+      "epoch": 0.08974712643678161,
+      "grad_norm": 7.9229607582092285,
+      "learning_rate": 4.912280701754387e-07,
+      "loss": 0.46270182728767395,
+      "num_input_tokens_seen": 616054,
+      "step": 1952,
+      "train_runtime": 5218.5133,
+      "train_tokens_per_second": 118.052
+    },
+    {
+      "epoch": 0.08979310344827586,
+      "grad_norm": 6.196159839630127,
+      "learning_rate": 4.81203007518797e-07,
+      "loss": 0.4044492840766907,
+      "num_input_tokens_seen": 616488,
+      "step": 1953,
+      "train_runtime": 5220.5335,
+      "train_tokens_per_second": 118.089
+    },
+    {
+      "epoch": 0.08983908045977011,
+      "grad_norm": 8.939637184143066,
+      "learning_rate": 4.711779448621554e-07,
+      "loss": 0.4814576208591461,
+      "num_input_tokens_seen": 616764,
+      "step": 1954,
+      "train_runtime": 5222.5264,
+      "train_tokens_per_second": 118.097
+    },
+    {
+      "epoch": 0.08988505747126437,
+      "grad_norm": 9.98863410949707,
+      "learning_rate": 4.6115288220551385e-07,
+      "loss": 0.44973185658454895,
+      "num_input_tokens_seen": 617038,
+      "step": 1955,
+      "train_runtime": 5224.468,
+      "train_tokens_per_second": 118.105
+    },
+    {
+      "epoch": 0.08993103448275862,
+      "grad_norm": 8.118160247802734,
+      "learning_rate": 4.511278195488722e-07,
+      "loss": 0.8060194253921509,
+      "num_input_tokens_seen": 617458,
+      "step": 1956,
+      "train_runtime": 5226.5359,
+      "train_tokens_per_second": 118.139
+    },
+    {
+      "epoch": 0.08997701149425287,
+      "grad_norm": 8.639585494995117,
+      "learning_rate": 4.411027568922306e-07,
+      "loss": 0.41593390703201294,
+      "num_input_tokens_seen": 617768,
+      "step": 1957,
+      "train_runtime": 5228.726,
+      "train_tokens_per_second": 118.149
+    },
+    {
+      "epoch": 0.09002298850574712,
+      "grad_norm": 6.963311672210693,
+      "learning_rate": 4.3107769423558905e-07,
+      "loss": 0.4132186770439148,
+      "num_input_tokens_seen": 618078,
+      "step": 1958,
+      "train_runtime": 5230.6867,
+      "train_tokens_per_second": 118.164
+    },
+    {
+      "epoch": 0.09006896551724138,
+      "grad_norm": 6.928063869476318,
+      "learning_rate": 4.210526315789474e-07,
+      "loss": 0.46738407015800476,
+      "num_input_tokens_seen": 618374,
+      "step": 1959,
+      "train_runtime": 5232.6462,
+      "train_tokens_per_second": 118.176
+    },
+    {
+      "epoch": 0.09011494252873563,
+      "grad_norm": 6.987063407897949,
+      "learning_rate": 4.110275689223058e-07,
+      "loss": 0.3887450695037842,
+      "num_input_tokens_seen": 618640,
+      "step": 1960,
+      "train_runtime": 5234.6425,
+      "train_tokens_per_second": 118.182
+    },
+    {
+      "epoch": 0.09016091954022988,
+      "grad_norm": 9.057330131530762,
+      "learning_rate": 4.010025062656642e-07,
+      "loss": 0.5830927491188049,
+      "num_input_tokens_seen": 619014,
+      "step": 1961,
+      "train_runtime": 5236.6084,
+      "train_tokens_per_second": 118.209
+    },
+    {
+      "epoch": 0.09020689655172413,
+      "grad_norm": 8.036870956420898,
+      "learning_rate": 3.909774436090226e-07,
+      "loss": 0.39967769384384155,
+      "num_input_tokens_seen": 619272,
+      "step": 1962,
+      "train_runtime": 5238.6378,
+      "train_tokens_per_second": 118.212
+    },
+    {
+      "epoch": 0.09025287356321839,
+      "grad_norm": 8.644590377807617,
+      "learning_rate": 3.80952380952381e-07,
+      "loss": 0.45566412806510925,
+      "num_input_tokens_seen": 619544,
+      "step": 1963,
+      "train_runtime": 5240.6292,
+      "train_tokens_per_second": 118.219
+    },
+    {
+      "epoch": 0.09029885057471264,
+      "grad_norm": 8.846774101257324,
+      "learning_rate": 3.709273182957394e-07,
+      "loss": 0.35211920738220215,
+      "num_input_tokens_seen": 619796,
+      "step": 1964,
+      "train_runtime": 5242.6112,
+      "train_tokens_per_second": 118.223
+    },
+    {
+      "epoch": 0.09034482758620689,
+      "grad_norm": 7.386560916900635,
+      "learning_rate": 3.609022556390978e-07,
+      "loss": 0.4106683135032654,
+      "num_input_tokens_seen": 620070,
+      "step": 1965,
+      "train_runtime": 5244.601,
+      "train_tokens_per_second": 118.23
+    },
+    {
+      "epoch": 0.09039080459770114,
+      "grad_norm": 7.662647247314453,
+      "learning_rate": 3.5087719298245616e-07,
+      "loss": 0.414768248796463,
+      "num_input_tokens_seen": 620352,
+      "step": 1966,
+      "train_runtime": 5246.5551,
+      "train_tokens_per_second": 118.24
+    },
+    {
+      "epoch": 0.0904367816091954,
+      "grad_norm": 7.379946708679199,
+      "learning_rate": 3.408521303258146e-07,
+      "loss": 0.5533338189125061,
+      "num_input_tokens_seen": 620708,
+      "step": 1967,
+      "train_runtime": 5248.5732,
+      "train_tokens_per_second": 118.262
+    },
+    {
+      "epoch": 0.09048275862068965,
+      "grad_norm": 5.688102722167969,
+      "learning_rate": 3.308270676691729e-07,
+      "loss": 0.30439692735671997,
+      "num_input_tokens_seen": 621016,
+      "step": 1968,
+      "train_runtime": 5250.5572,
+      "train_tokens_per_second": 118.276
+    },
+    {
+      "epoch": 0.0905287356321839,
+      "grad_norm": 7.519008636474609,
+      "learning_rate": 3.2080200501253136e-07,
+      "loss": 0.45001494884490967,
+      "num_input_tokens_seen": 621292,
+      "step": 1969,
+      "train_runtime": 5252.5832,
+      "train_tokens_per_second": 118.283
+    },
+    {
+      "epoch": 0.09057471264367815,
+      "grad_norm": 8.505365371704102,
+      "learning_rate": 3.1077694235588974e-07,
+      "loss": 0.4692716598510742,
+      "num_input_tokens_seen": 621558,
+      "step": 1970,
+      "train_runtime": 5254.5359,
+      "train_tokens_per_second": 118.29
+    },
+    {
+      "epoch": 0.09062068965517242,
+      "grad_norm": 7.983728408813477,
+      "learning_rate": 3.007518796992481e-07,
+      "loss": 0.4891626238822937,
+      "num_input_tokens_seen": 621856,
+      "step": 1971,
+      "train_runtime": 5256.5377,
+      "train_tokens_per_second": 118.301
+    },
+    {
+      "epoch": 0.09066666666666667,
+      "grad_norm": 7.890754699707031,
+      "learning_rate": 2.9072681704260656e-07,
+      "loss": 0.48190662264823914,
+      "num_input_tokens_seen": 622162,
+      "step": 1972,
+      "train_runtime": 5258.4965,
+      "train_tokens_per_second": 118.316
+    },
+    {
+      "epoch": 0.09071264367816093,
+      "grad_norm": 7.7978596687316895,
+      "learning_rate": 2.8070175438596494e-07,
+      "loss": 0.46703723073005676,
+      "num_input_tokens_seen": 622446,
+      "step": 1973,
+      "train_runtime": 5260.4611,
+      "train_tokens_per_second": 118.325
+    },
+    {
+      "epoch": 0.09075862068965518,
+      "grad_norm": 9.207006454467773,
+      "learning_rate": 2.706766917293233e-07,
+      "loss": 0.4706171154975891,
+      "num_input_tokens_seen": 622830,
+      "step": 1974,
+      "train_runtime": 5262.4203,
+      "train_tokens_per_second": 118.354
+    },
+    {
+      "epoch": 0.09080459770114943,
+      "grad_norm": 7.43474817276001,
+      "learning_rate": 2.606516290726817e-07,
+      "loss": 0.43948253989219666,
+      "num_input_tokens_seen": 623100,
+      "step": 1975,
+      "train_runtime": 5264.3793,
+      "train_tokens_per_second": 118.362
+    },
+    {
+      "epoch": 0.09085057471264368,
+      "grad_norm": 7.330848693847656,
+      "learning_rate": 2.506265664160401e-07,
+      "loss": 0.4305328130722046,
+      "num_input_tokens_seen": 623370,
+      "step": 1976,
+      "train_runtime": 5266.3846,
+      "train_tokens_per_second": 118.368
+    },
+    {
+      "epoch": 0.09089655172413794,
+      "grad_norm": 9.419845581054688,
+      "learning_rate": 2.406015037593985e-07,
+      "loss": 0.4397188127040863,
+      "num_input_tokens_seen": 623682,
+      "step": 1977,
+      "train_runtime": 5268.3391,
+      "train_tokens_per_second": 118.383
+    },
+    {
+      "epoch": 0.09094252873563219,
+      "grad_norm": 7.668386936187744,
+      "learning_rate": 2.3057644110275693e-07,
+      "loss": 0.3994528353214264,
+      "num_input_tokens_seen": 624000,
+      "step": 1978,
+      "train_runtime": 5270.3557,
+      "train_tokens_per_second": 118.398
+    },
+    {
+      "epoch": 0.09098850574712644,
+      "grad_norm": 9.285679817199707,
+      "learning_rate": 2.205513784461153e-07,
+      "loss": 0.6182694435119629,
+      "num_input_tokens_seen": 624358,
+      "step": 1979,
+      "train_runtime": 5272.457,
+      "train_tokens_per_second": 118.419
+    },
+    {
+      "epoch": 0.0910344827586207,
+      "grad_norm": 7.558021068572998,
+      "learning_rate": 2.105263157894737e-07,
+      "loss": 0.5416414141654968,
+      "num_input_tokens_seen": 624646,
+      "step": 1980,
+      "train_runtime": 5274.4287,
+      "train_tokens_per_second": 118.429
+    },
+    {
+      "epoch": 0.09108045977011495,
+      "grad_norm": 8.10972785949707,
+      "learning_rate": 2.005012531328321e-07,
+      "loss": 0.41339346766471863,
+      "num_input_tokens_seen": 624996,
+      "step": 1981,
+      "train_runtime": 5290.1793,
+      "train_tokens_per_second": 118.143
+    },
+    {
+      "epoch": 0.0911264367816092,
+      "grad_norm": 8.973311424255371,
+      "learning_rate": 1.904761904761905e-07,
+      "loss": 0.4418382942676544,
+      "num_input_tokens_seen": 625278,
+      "step": 1982,
+      "train_runtime": 5292.1961,
+      "train_tokens_per_second": 118.151
+    },
+    {
+      "epoch": 0.09117241379310345,
+      "grad_norm": 7.290642738342285,
+      "learning_rate": 1.804511278195489e-07,
+      "loss": 0.466846227645874,
+      "num_input_tokens_seen": 625644,
+      "step": 1983,
+      "train_runtime": 5294.1789,
+      "train_tokens_per_second": 118.176
+    },
+    {
+      "epoch": 0.0912183908045977,
+      "grad_norm": 7.393955707550049,
+      "learning_rate": 1.704260651629073e-07,
+      "loss": 0.42903000116348267,
+      "num_input_tokens_seen": 625978,
+      "step": 1984,
+      "train_runtime": 5296.2594,
+      "train_tokens_per_second": 118.192
+    },
+    {
+      "epoch": 0.09126436781609196,
+      "grad_norm": 7.702242851257324,
+      "learning_rate": 1.6040100250626568e-07,
+      "loss": 0.531277596950531,
+      "num_input_tokens_seen": 626260,
+      "step": 1985,
+      "train_runtime": 5298.2079,
+      "train_tokens_per_second": 118.202
+    },
+    {
+      "epoch": 0.09131034482758621,
+      "grad_norm": 8.941267967224121,
+      "learning_rate": 1.5037593984962406e-07,
+      "loss": 0.4884231388568878,
+      "num_input_tokens_seen": 626542,
+      "step": 1986,
+      "train_runtime": 5300.164,
+      "train_tokens_per_second": 118.212
+    },
+    {
+      "epoch": 0.09135632183908046,
+      "grad_norm": 7.654355049133301,
+      "learning_rate": 1.4035087719298247e-07,
+      "loss": 0.3853590488433838,
+      "num_input_tokens_seen": 626866,
+      "step": 1987,
+      "train_runtime": 5302.1362,
+      "train_tokens_per_second": 118.229
+    },
+    {
+      "epoch": 0.09140229885057471,
+      "grad_norm": 7.743011474609375,
+      "learning_rate": 1.3032581453634085e-07,
+      "loss": 0.4582059383392334,
+      "num_input_tokens_seen": 627266,
+      "step": 1988,
+      "train_runtime": 5304.1709,
+      "train_tokens_per_second": 118.259
+    },
+    {
+      "epoch": 0.09144827586206897,
+      "grad_norm": 6.894120216369629,
+      "learning_rate": 1.2030075187969926e-07,
+      "loss": 0.43101340532302856,
+      "num_input_tokens_seen": 627574,
+      "step": 1989,
+      "train_runtime": 5306.1319,
+      "train_tokens_per_second": 118.273
+    },
+    {
+      "epoch": 0.09149425287356322,
+      "grad_norm": 8.84626293182373,
+      "learning_rate": 1.1027568922305765e-07,
+      "loss": 0.598413348197937,
+      "num_input_tokens_seen": 627888,
+      "step": 1990,
+      "train_runtime": 5308.1433,
+      "train_tokens_per_second": 118.288
+    },
+    {
+      "epoch": 0.09154022988505747,
+      "grad_norm": 10.974281311035156,
+      "learning_rate": 1.0025062656641605e-07,
+      "loss": 0.4394698739051819,
+      "num_input_tokens_seen": 628118,
+      "step": 1991,
+      "train_runtime": 5310.0853,
+      "train_tokens_per_second": 118.288
+    },
+    {
+      "epoch": 0.09158620689655173,
+      "grad_norm": 6.511246681213379,
+      "learning_rate": 9.022556390977444e-08,
+      "loss": 0.41420939564704895,
+      "num_input_tokens_seen": 628438,
+      "step": 1992,
+      "train_runtime": 5312.0269,
+      "train_tokens_per_second": 118.305
+    },
+    {
+      "epoch": 0.09163218390804598,
+      "grad_norm": 6.606043815612793,
+      "learning_rate": 8.020050125313284e-08,
+      "loss": 0.3737097978591919,
+      "num_input_tokens_seen": 628778,
+      "step": 1993,
+      "train_runtime": 5314.0027,
+      "train_tokens_per_second": 118.325
+    },
+    {
+      "epoch": 0.09167816091954023,
+      "grad_norm": 6.390109539031982,
+      "learning_rate": 7.017543859649123e-08,
+      "loss": 0.4567974805831909,
+      "num_input_tokens_seen": 629192,
+      "step": 1994,
+      "train_runtime": 5315.9833,
+      "train_tokens_per_second": 118.359
+    },
+    {
+      "epoch": 0.09172413793103448,
+      "grad_norm": 9.456647872924805,
+      "learning_rate": 6.015037593984963e-08,
+      "loss": 0.5348672866821289,
+      "num_input_tokens_seen": 629514,
+      "step": 1995,
+      "train_runtime": 5317.9562,
+      "train_tokens_per_second": 118.375
+    },
+    {
+      "epoch": 0.09177011494252874,
+      "grad_norm": 7.294930458068848,
+      "learning_rate": 5.0125313283208025e-08,
+      "loss": 0.5476741790771484,
+      "num_input_tokens_seen": 629936,
+      "step": 1996,
+      "train_runtime": 5319.9153,
+      "train_tokens_per_second": 118.411
+    },
+    {
+      "epoch": 0.09181609195402299,
+      "grad_norm": 8.284785270690918,
+      "learning_rate": 4.010025062656642e-08,
+      "loss": 0.3866909146308899,
+      "num_input_tokens_seen": 630230,
+      "step": 1997,
+      "train_runtime": 5321.8832,
+      "train_tokens_per_second": 118.422
+    },
+    {
+      "epoch": 0.09186206896551724,
+      "grad_norm": 8.145447731018066,
+      "learning_rate": 3.0075187969924815e-08,
+      "loss": 0.4199717342853546,
+      "num_input_tokens_seen": 630492,
+      "step": 1998,
+      "train_runtime": 5323.843,
+      "train_tokens_per_second": 118.428
+    },
+    {
+      "epoch": 0.0919080459770115,
+      "grad_norm": 7.085546016693115,
+      "learning_rate": 2.005012531328321e-08,
+      "loss": 0.3851563036441803,
+      "num_input_tokens_seen": 630840,
+      "step": 1999,
+      "train_runtime": 5325.8038,
+      "train_tokens_per_second": 118.45
+    },
+    {
+      "epoch": 0.09195402298850575,
+      "grad_norm": 6.691437721252441,
+      "learning_rate": 1.0025062656641605e-08,
+      "loss": 0.3576313853263855,
+      "num_input_tokens_seen": 631158,
+      "step": 2000,
+      "train_runtime": 5327.7708,
+      "train_tokens_per_second": 118.466
+    },
+    {
+      "epoch": 0.09195402298850575,
+      "eval_loss": 1.6557446718215942,
+      "eval_runtime": 53.863,
+      "eval_samples_per_second": 18.566,
+      "eval_steps_per_second": 9.283,
+      "num_input_tokens_seen": 631158,
+      "step": 2000
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 2000,
+  "num_input_tokens_seen": 631158,
+  "num_train_epochs": 1,
+  "save_steps": 30,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 6446608710499584.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}