| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.0, | |
| "global_step": 504, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.25e-06, | |
| "loss": 0.5164, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.5394, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.7500000000000005e-06, | |
| "loss": 0.4026, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 5e-06, | |
| "loss": 0.3397, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 6.25e-06, | |
| "loss": 0.3042, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.2701, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 8.750000000000001e-06, | |
| "loss": 0.2333, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1e-05, | |
| "loss": 0.2235, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.125e-05, | |
| "loss": 0.202, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.25e-05, | |
| "loss": 0.1832, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.375e-05, | |
| "loss": 0.154, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.5000000000000002e-05, | |
| "loss": 0.1657, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.6250000000000002e-05, | |
| "loss": 0.2064, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.7500000000000002e-05, | |
| "loss": 0.1445, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.8750000000000002e-05, | |
| "loss": 0.1793, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2e-05, | |
| "loss": 0.1781, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.9999792781461744e-05, | |
| "loss": 0.1996, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.999917113443488e-05, | |
| "loss": 0.1394, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.999813508468276e-05, | |
| "loss": 0.1886, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.9996684675143132e-05, | |
| "loss": 0.1687, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.9994819965926346e-05, | |
| "loss": 0.1609, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.9992541034312862e-05, | |
| "loss": 0.1671, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.9989847974750057e-05, | |
| "loss": 0.1605, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.9986740898848306e-05, | |
| "loss": 0.1957, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.9983219935376353e-05, | |
| "loss": 0.1873, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.997928523025598e-05, | |
| "loss": 0.1566, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.997493694655595e-05, | |
| "loss": 0.1675, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.9970175264485268e-05, | |
| "loss": 0.1712, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.996500038138569e-05, | |
| "loss": 0.167, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.995941251172356e-05, | |
| "loss": 0.1804, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.9953411887080917e-05, | |
| "loss": 0.157, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9946998756145894e-05, | |
| "loss": 0.177, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9940173384702407e-05, | |
| "loss": 0.1696, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9932936055619164e-05, | |
| "loss": 0.1601, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.992528706883791e-05, | |
| "loss": 0.1725, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.9917226741361014e-05, | |
| "loss": 0.1496, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.9908755407238342e-05, | |
| "loss": 0.1496, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.9899873417553384e-05, | |
| "loss": 0.1651, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.9890581140408722e-05, | |
| "loss": 0.1885, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.9880878960910772e-05, | |
| "loss": 0.1547, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.987076728115383e-05, | |
| "loss": 0.1567, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.9860246520203393e-05, | |
| "loss": 0.136, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.98493171140788e-05, | |
| "loss": 0.1568, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.9837979515735168e-05, | |
| "loss": 0.1597, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.98262341950446e-05, | |
| "loss": 0.1334, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.9814081638776743e-05, | |
| "loss": 0.1502, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.980152235057858e-05, | |
| "loss": 0.165, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.978855685095358e-05, | |
| "loss": 0.1355, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.977518567724012e-05, | |
| "loss": 0.1548, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.9761409383589216e-05, | |
| "loss": 0.1526, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.9747228540941555e-05, | |
| "loss": 0.162, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.9732643737003827e-05, | |
| "loss": 0.1658, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.971765557622439e-05, | |
| "loss": 0.1681, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.9702264679768193e-05, | |
| "loss": 0.1717, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.9686471685491055e-05, | |
| "loss": 0.18, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.9670277247913205e-05, | |
| "loss": 0.1321, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.965368203819219e-05, | |
| "loss": 0.1558, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.9636686744095018e-05, | |
| "loss": 0.1422, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.96192920699697e-05, | |
| "loss": 0.1428, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.960149873671602e-05, | |
| "loss": 0.1679, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.958330748175568e-05, | |
| "loss": 0.153, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.9564719059001735e-05, | |
| "loss": 0.1863, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.9545734238827334e-05, | |
| "loss": 0.1534, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.9526353808033827e-05, | |
| "loss": 0.1549, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.9506578569818115e-05, | |
| "loss": 0.1367, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.948640934373939e-05, | |
| "loss": 0.1602, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.946584696568516e-05, | |
| "loss": 0.1721, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.9444892287836614e-05, | |
| "loss": 0.1538, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.942354617863329e-05, | |
| "loss": 0.1656, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.94018095227371e-05, | |
| "loss": 0.1677, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.9379683220995657e-05, | |
| "loss": 0.1483, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.9357168190404937e-05, | |
| "loss": 0.1546, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.933426536407129e-05, | |
| "loss": 0.152, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.9310975691172756e-05, | |
| "loss": 0.1486, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.9287300136919726e-05, | |
| "loss": 0.1588, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.9263239682514953e-05, | |
| "loss": 0.1521, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.9238795325112867e-05, | |
| "loss": 0.1608, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.9213968077778284e-05, | |
| "loss": 0.1464, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.918875896944437e-05, | |
| "loss": 0.1585, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.916316904487005e-05, | |
| "loss": 0.1444, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.9137199364596673e-05, | |
| "loss": 0.1444, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.9110851004904076e-05, | |
| "loss": 0.1727, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.9084125057765978e-05, | |
| "loss": 0.1503, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.9057022630804715e-05, | |
| "loss": 0.1251, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.902954484724535e-05, | |
| "loss": 0.1502, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.9001692845869113e-05, | |
| "loss": 0.172, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.89734677809662e-05, | |
| "loss": 0.1562, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.8944870822287957e-05, | |
| "loss": 0.1557, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.8915903154998374e-05, | |
| "loss": 0.1718, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.888656597962499e-05, | |
| "loss": 0.1472, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.8856860512009115e-05, | |
| "loss": 0.1415, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.8826787983255474e-05, | |
| "loss": 0.1643, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.8796349639681154e-05, | |
| "loss": 0.1272, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.8765546742763972e-05, | |
| "loss": 0.1327, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.8734380569090175e-05, | |
| "loss": 0.1572, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.8702852410301556e-05, | |
| "loss": 0.1683, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.867096357304191e-05, | |
| "loss": 0.1769, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.863871537890289e-05, | |
| "loss": 0.1393, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.860610916436922e-05, | |
| "loss": 0.1537, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.8573146280763327e-05, | |
| "loss": 0.1379, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.853982809418932e-05, | |
| "loss": 0.1494, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.8506155985476386e-05, | |
| "loss": 0.1375, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.847213135012155e-05, | |
| "loss": 0.128, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.8437755598231857e-05, | |
| "loss": 0.1252, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.8403030154465916e-05, | |
| "loss": 0.1521, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.8367956457974872e-05, | |
| "loss": 0.1327, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.8332535962342738e-05, | |
| "loss": 0.1628, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.829677013552619e-05, | |
| "loss": 0.1388, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.8260660459793692e-05, | |
| "loss": 0.1656, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.8224208431664093e-05, | |
| "loss": 0.146, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.8187415561844586e-05, | |
| "loss": 0.1621, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.8150283375168112e-05, | |
| "loss": 0.1505, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.8112813410530167e-05, | |
| "loss": 0.1341, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.8075007220825004e-05, | |
| "loss": 0.1449, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.8036866372881295e-05, | |
| "loss": 0.1465, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.7998392447397197e-05, | |
| "loss": 0.1629, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.7959587038874826e-05, | |
| "loss": 0.1511, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.792045175555419e-05, | |
| "loss": 0.147, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.788098821934653e-05, | |
| "loss": 0.1438, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.7841198065767107e-05, | |
| "loss": 0.1516, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.7801082943867406e-05, | |
| "loss": 0.144, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.7760644516166815e-05, | |
| "loss": 0.1682, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.771988445858371e-05, | |
| "loss": 0.1737, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.7678804460366e-05, | |
| "loss": 0.1277, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.7637406224021124e-05, | |
| "loss": 0.1612, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.7595691465245484e-05, | |
| "loss": 0.1512, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.7553661912853348e-05, | |
| "loss": 0.1353, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.7511319308705198e-05, | |
| "loss": 0.1593, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.7468665407635543e-05, | |
| "loss": 0.1415, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.742570197738019e-05, | |
| "loss": 0.1307, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.7382430798502977e-05, | |
| "loss": 0.128, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.7338853664321993e-05, | |
| "loss": 0.1456, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.729497238083525e-05, | |
| "loss": 0.1495, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.725078876664583e-05, | |
| "loss": 0.1398, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.7206304652886524e-05, | |
| "loss": 0.1062, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.7161521883143936e-05, | |
| "loss": 0.1529, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.711644231338208e-05, | |
| "loss": 0.1481, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.7071067811865477e-05, | |
| "loss": 0.1523, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.7025400259081694e-05, | |
| "loss": 0.15, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.6979441547663434e-05, | |
| "loss": 0.1541, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.693319358231011e-05, | |
| "loss": 0.1647, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.6886658279708867e-05, | |
| "loss": 0.1653, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.6839837568455184e-05, | |
| "loss": 0.156, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.679273338897293e-05, | |
| "loss": 0.1407, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.6745347693433953e-05, | |
| "loss": 0.1634, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.6697682445677158e-05, | |
| "loss": 0.1511, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.6649739621127147e-05, | |
| "loss": 0.1492, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.660152120671232e-05, | |
| "loss": 0.1319, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.6553029200782545e-05, | |
| "loss": 0.1531, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.650426561302634e-05, | |
| "loss": 0.1221, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.6455232464387587e-05, | |
| "loss": 0.1474, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.6405931786981753e-05, | |
| "loss": 0.1417, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.6356365624011706e-05, | |
| "loss": 0.1389, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.6306536029683015e-05, | |
| "loss": 0.1406, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.6256445069118812e-05, | |
| "loss": 0.134, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.6206094818274228e-05, | |
| "loss": 0.1454, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.615548736385034e-05, | |
| "loss": 0.1583, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.6104624803207697e-05, | |
| "loss": 0.1602, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.6053509244279382e-05, | |
| "loss": 0.125, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.6002142805483686e-05, | |
| "loss": 0.1552, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.595052761563627e-05, | |
| "loss": 0.1512, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.589866581386199e-05, | |
| "loss": 0.1273, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.5846559549506183e-05, | |
| "loss": 0.1632, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.5794210982045638e-05, | |
| "loss": 0.1402, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.5741622280999086e-05, | |
| "loss": 0.1301, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.5688795625837274e-05, | |
| "loss": 0.1741, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.5635733205892653e-05, | |
| "loss": 0.1344, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.5582437220268648e-05, | |
| "loss": 0.1474, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.5528909877748504e-05, | |
| "loss": 0.1192, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.5475153396703756e-05, | |
| "loss": 0.1448, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.542117000500229e-05, | |
| "loss": 0.1463, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.536696193991601e-05, | |
| "loss": 0.146, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.5312531448028107e-05, | |
| "loss": 0.1646, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.525788078513999e-05, | |
| "loss": 0.165, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.5203012216177748e-05, | |
| "loss": 0.132, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.5147928015098309e-05, | |
| "loss": 0.1315, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.5092630464795202e-05, | |
| "loss": 0.1266, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.5037121857003937e-05, | |
| "loss": 0.1368, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.4981404492207025e-05, | |
| "loss": 0.1255, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.4925480679538646e-05, | |
| "loss": 0.1512, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.4869352736688938e-05, | |
| "loss": 0.1601, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.4813022989807964e-05, | |
| "loss": 0.1607, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.4756493773409276e-05, | |
| "loss": 0.1339, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.4699767430273202e-05, | |
| "loss": 0.1498, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.4642846311349717e-05, | |
| "loss": 0.1343, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.458573277566103e-05, | |
| "loss": 0.1196, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.4528429190203826e-05, | |
| "loss": 0.1553, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.4470937929851142e-05, | |
| "loss": 0.1601, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.4413261377253961e-05, | |
| "loss": 0.1533, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.4355401922742478e-05, | |
| "loss": 0.1399, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.4297361964227004e-05, | |
| "loss": 0.1245, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.423914390709861e-05, | |
| "loss": 0.148, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.4180750164129435e-05, | |
| "loss": 0.1221, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.4122183155372699e-05, | |
| "loss": 0.1457, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.4063445308062384e-05, | |
| "loss": 0.1251, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.4004539056512667e-05, | |
| "loss": 0.1174, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.394546684201701e-05, | |
| "loss": 0.1157, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.3886231112747006e-05, | |
| "loss": 0.13, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.3826834323650899e-05, | |
| "loss": 0.1444, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.3767278936351853e-05, | |
| "loss": 0.133, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.3707567419045926e-05, | |
| "loss": 0.1203, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.3647702246399787e-05, | |
| "loss": 0.165, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.3587685899448148e-05, | |
| "loss": 0.1404, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.352752086549095e-05, | |
| "loss": 0.1532, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.3467209637990266e-05, | |
| "loss": 0.1585, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.3406754716466978e-05, | |
| "loss": 0.1354, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.3346158606397182e-05, | |
| "loss": 0.1734, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.3285423819108349e-05, | |
| "loss": 0.1408, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.3224552871675244e-05, | |
| "loss": 0.1199, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.316354828681562e-05, | |
| "loss": 0.1352, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.3102412592785654e-05, | |
| "loss": 0.1449, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.3041148323275182e-05, | |
| "loss": 0.1371, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.2979758017302673e-05, | |
| "loss": 0.1328, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.2918244219110022e-05, | |
| "loss": 0.1347, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.28566094780571e-05, | |
| "loss": 0.1549, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.2794856348516095e-05, | |
| "loss": 0.1292, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.2732987389765659e-05, | |
| "loss": 0.1489, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.2671005165884818e-05, | |
| "loss": 0.1314, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.2608912245646754e-05, | |
| "loss": 0.1399, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.2546711202412287e-05, | |
| "loss": 0.1289, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.248440461402328e-05, | |
| "loss": 0.129, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.2421995062695758e-05, | |
| "loss": 0.1208, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.2359485134912928e-05, | |
| "loss": 0.1617, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.2296877421317958e-05, | |
| "loss": 0.1122, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.2234174516606627e-05, | |
| "loss": 0.1259, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.2171379019419786e-05, | |
| "loss": 0.1567, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.2108493532235665e-05, | |
| "loss": 0.1408, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.2045520661262011e-05, | |
| "loss": 0.1294, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.1982463016328074e-05, | |
| "loss": 0.1457, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.1919323210776459e-05, | |
| "loss": 0.1494, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.1856103861354809e-05, | |
| "loss": 0.1284, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.1792807588107358e-05, | |
| "loss": 0.1153, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.1729437014266352e-05, | |
| "loss": 0.1179, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.166599476614332e-05, | |
| "loss": 0.1536, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.1602483473020257e-05, | |
| "loss": 0.1372, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.153890576704062e-05, | |
| "loss": 0.1251, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.1475264283100271e-05, | |
| "loss": 0.1472, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.1411561658738261e-05, | |
| "loss": 0.1578, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.134780053402754e-05, | |
| "loss": 0.1235, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.1283983551465512e-05, | |
| "loss": 0.1605, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.1220113355864549e-05, | |
| "loss": 0.1616, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.1156192594242359e-05, | |
| "loss": 0.1535, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.109222391571231e-05, | |
| "loss": 0.1299, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.1028209971373605e-05, | |
| "loss": 0.1902, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.0964153414201446e-05, | |
| "loss": 0.1329, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.0900056898937055e-05, | |
| "loss": 0.1413, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.0835923081977673e-05, | |
| "loss": 0.1249, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.0771754621266466e-05, | |
| "loss": 0.15, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.0707554176182355e-05, | |
| "loss": 0.1264, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.0643324407429818e-05, | |
| "loss": 0.1492, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.0579067976928614e-05, | |
| "loss": 0.1338, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.0514787547703466e-05, | |
| "loss": 0.1563, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.0450485783773689e-05, | |
| "loss": 0.1575, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 1.0386165350042786e-05, | |
| "loss": 0.1259, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 1.032182891218801e-05, | |
| "loss": 0.1152, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 1.0257479136549889e-05, | |
| "loss": 0.1465, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 1.0193118690021699e-05, | |
| "loss": 0.1363, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 1.0128750239938983e-05, | |
| "loss": 0.136, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 1.006437645396896e-05, | |
| "loss": 0.1206, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 1e-05, | |
| "loss": 0.1262, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.935623546031043e-06, | |
| "loss": 0.1145, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.87124976006102e-06, | |
| "loss": 0.1382, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.8068813099783e-06, | |
| "loss": 0.1371, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.742520863450116e-06, | |
| "loss": 0.1186, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.678171087811991e-06, | |
| "loss": 0.1217, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 9.613834649957216e-06, | |
| "loss": 0.1322, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 9.549514216226312e-06, | |
| "loss": 0.1493, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 9.485212452296535e-06, | |
| "loss": 0.135, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 9.42093202307139e-06, | |
| "loss": 0.142, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 9.356675592570185e-06, | |
| "loss": 0.1236, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 9.292445823817647e-06, | |
| "loss": 0.1179, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 9.228245378733537e-06, | |
| "loss": 0.0993, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 9.16407691802233e-06, | |
| "loss": 0.1144, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 9.099943101062948e-06, | |
| "loss": 0.1468, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 9.03584658579856e-06, | |
| "loss": 0.1192, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 8.971790028626395e-06, | |
| "loss": 0.1293, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 8.907776084287694e-06, | |
| "loss": 0.1459, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 8.843807405757644e-06, | |
| "loss": 0.1307, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 8.779886644135456e-06, | |
| "loss": 0.133, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 8.71601644853449e-06, | |
| "loss": 0.1355, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 8.652199465972462e-06, | |
| "loss": 0.1453, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 8.58843834126174e-06, | |
| "loss": 0.1291, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 8.524735716899734e-06, | |
| "loss": 0.1237, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 8.461094232959381e-06, | |
| "loss": 0.1272, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 8.397516526979745e-06, | |
| "loss": 0.1171, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 8.334005233856681e-06, | |
| "loss": 0.1084, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 8.270562985733652e-06, | |
| "loss": 0.1345, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 8.207192411892645e-06, | |
| "loss": 0.1066, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 8.143896138645193e-06, | |
| "loss": 0.1358, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 8.080676789223545e-06, | |
| "loss": 0.1473, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 8.017536983671929e-06, | |
| "loss": 0.1214, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 7.954479338737995e-06, | |
| "loss": 0.1345, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 7.891506467764337e-06, | |
| "loss": 0.1327, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 7.828620980580217e-06, | |
| "loss": 0.1284, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 7.765825483393376e-06, | |
| "loss": 0.1175, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 7.703122578682047e-06, | |
| "loss": 0.1175, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 7.640514865087078e-06, | |
| "loss": 0.1263, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 7.578004937304242e-06, | |
| "loss": 0.1153, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 7.5155953859767215e-06, | |
| "loss": 0.1519, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 7.453288797587714e-06, | |
| "loss": 0.135, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 7.391087754353252e-06, | |
| "loss": 0.131, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 7.328994834115181e-06, | |
| "loss": 0.1358, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 7.267012610234345e-06, | |
| "loss": 0.108, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 7.2051436514839064e-06, | |
| "loss": 0.1403, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 7.143390521942903e-06, | |
| "loss": 0.1373, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 7.081755780889978e-06, | |
| "loss": 0.1428, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 7.020241982697331e-06, | |
| "loss": 0.1318, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 6.958851676724823e-06, | |
| "loss": 0.1188, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 6.89758740721435e-06, | |
| "loss": 0.1347, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 6.8364517131843865e-06, | |
| "loss": 0.103, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 6.7754471283247594e-06, | |
| "loss": 0.1057, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 6.714576180891653e-06, | |
| "loss": 0.1431, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 6.65384139360282e-06, | |
| "loss": 0.1294, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 6.593245283533025e-06, | |
| "loss": 0.143, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 6.5327903620097366e-06, | |
| "loss": 0.1358, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 6.472479134509052e-06, | |
| "loss": 0.1391, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 6.412314100551854e-06, | |
| "loss": 0.1153, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 6.352297753600218e-06, | |
| "loss": 0.1265, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 6.292432580954075e-06, | |
| "loss": 0.1206, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 6.232721063648148e-06, | |
| "loss": 0.1343, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 6.173165676349103e-06, | |
| "loss": 0.1404, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 6.113768887252998e-06, | |
| "loss": 0.1537, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 6.054533157982994e-06, | |
| "loss": 0.1197, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 5.995460943487334e-06, | |
| "loss": 0.1178, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 5.936554691937617e-06, | |
| "loss": 0.1435, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 5.8778168446273045e-06, | |
| "loss": 0.1281, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 5.819249835870567e-06, | |
| "loss": 0.1247, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 5.760856092901394e-06, | |
| "loss": 0.1076, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 5.702638035773001e-06, | |
| "loss": 0.1446, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 5.644598077257524e-06, | |
| "loss": 0.1153, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 5.586738622746042e-06, | |
| "loss": 0.147, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 5.529062070148859e-06, | |
| "loss": 0.1187, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 5.4715708097961785e-06, | |
| "loss": 0.1395, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 5.414267224338971e-06, | |
| "loss": 0.1032, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 5.357153688650287e-06, | |
| "loss": 0.1204, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 5.300232569726805e-06, | |
| "loss": 0.1279, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 5.243506226590722e-06, | |
| "loss": 0.1239, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 5.186977010192037e-06, | |
| "loss": 0.1369, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 5.130647263311064e-06, | |
| "loss": 0.1312, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 5.074519320461358e-06, | |
| "loss": 0.1395, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 5.0185955077929774e-06, | |
| "loss": 0.1153, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 4.962878142996065e-06, | |
| "loss": 0.113, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 4.907369535204799e-06, | |
| "loss": 0.1229, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 4.852071984901696e-06, | |
| "loss": 0.1218, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 4.796987783822254e-06, | |
| "loss": 0.1345, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 4.742119214860009e-06, | |
| "loss": 0.1209, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.6874685519718945e-06, | |
| "loss": 0.114, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.633038060083996e-06, | |
| "loss": 0.1127, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.578829994997712e-06, | |
| "loss": 0.1078, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 4.524846603296245e-06, | |
| "loss": 0.1236, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 4.471090122251496e-06, | |
| "loss": 0.1418, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 4.417562779731355e-06, | |
| "loss": 0.1117, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 4.36426679410735e-06, | |
| "loss": 0.1368, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 4.311204374162731e-06, | |
| "loss": 0.1028, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 4.258377719000918e-06, | |
| "loss": 0.1209, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 4.205789017954364e-06, | |
| "loss": 0.1529, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 4.153440450493823e-06, | |
| "loss": 0.1364, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 4.101334186138015e-06, | |
| "loss": 0.1222, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 4.0494723843637275e-06, | |
| "loss": 0.1219, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 3.997857194516319e-06, | |
| "loss": 0.0985, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 3.946490755720621e-06, | |
| "loss": 0.1177, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 3.895375196792308e-06, | |
| "loss": 0.1376, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 3.84451263614966e-06, | |
| "loss": 0.1421, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 3.793905181725772e-06, | |
| "loss": 0.1477, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 3.7435549308811923e-06, | |
| "loss": 0.1207, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 3.6934639703169905e-06, | |
| "loss": 0.1137, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 3.643634375988293e-06, | |
| "loss": 0.1265, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 3.594068213018249e-06, | |
| "loss": 0.1175, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 3.5447675356124167e-06, | |
| "loss": 0.1542, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 3.4957343869736604e-06, | |
| "loss": 0.1313, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 3.4469707992174607e-06, | |
| "loss": 0.1163, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 3.398478793287682e-06, | |
| "loss": 0.1475, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 3.3502603788728547e-06, | |
| "loss": 0.1368, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 3.3023175543228416e-06, | |
| "loss": 0.1299, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 3.2546523065660483e-06, | |
| "loss": 0.1101, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 3.207266611027069e-06, | |
| "loss": 0.1205, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 3.1601624315448167e-06, | |
| "loss": 0.1244, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 3.113341720291134e-06, | |
| "loss": 0.1403, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 3.066806417689894e-06, | |
| "loss": 0.1372, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 3.0205584523365626e-06, | |
| "loss": 0.12, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 2.97459974091831e-06, | |
| "loss": 0.1438, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 2.9289321881345257e-06, | |
| "loss": 0.1292, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 2.8835576866179193e-06, | |
| "loss": 0.1056, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 2.8384781168560693e-06, | |
| "loss": 0.1363, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 2.7936953471134785e-06, | |
| "loss": 0.1261, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 2.7492112333541744e-06, | |
| "loss": 0.1333, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 2.705027619164754e-06, | |
| "loss": 0.1306, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 2.66114633567801e-06, | |
| "loss": 0.1286, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 2.6175692014970268e-06, | |
| "loss": 0.1244, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 2.5742980226198133e-06, | |
| "loss": 0.1102, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 2.531334592364457e-06, | |
| "loss": 0.1377, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 2.4886806912948034e-06, | |
| "loss": 0.1229, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 2.44633808714665e-06, | |
| "loss": 0.1346, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 2.404308534754517e-06, | |
| "loss": 0.1242, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 2.3625937759788775e-06, | |
| "loss": 0.1418, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 2.3211955396340003e-06, | |
| "loss": 0.1422, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 2.2801155414162933e-06, | |
| "loss": 0.1193, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 2.2393554838331853e-06, | |
| "loss": 0.1275, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 2.1989170561325957e-06, | |
| "loss": 0.1362, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 2.158801934232897e-06, | |
| "loss": 0.1406, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 2.1190117806534714e-06, | |
| "loss": 0.1318, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 2.0795482444458115e-06, | |
| "loss": 0.1176, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 2.0404129611251765e-06, | |
| "loss": 0.1346, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 2.0016075526028066e-06, | |
| "loss": 0.1309, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 1.9631336271187096e-06, | |
| "loss": 0.1315, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 1.924992779174999e-06, | |
| "loss": 0.1345, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 1.8871865894698338e-06, | |
| "loss": 0.1369, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 1.8497166248318876e-06, | |
| "loss": 0.1381, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 1.8125844381554158e-06, | |
| "loss": 0.135, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 1.7757915683359084e-06, | |
| "loss": 0.1306, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 1.7393395402063085e-06, | |
| "loss": 0.1274, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 1.703229864473811e-06, | |
| "loss": 0.1309, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 1.6674640376572637e-06, | |
| "loss": 0.1221, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 1.6320435420251323e-06, | |
| "loss": 0.1187, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 1.5969698455340842e-06, | |
| "loss": 0.1473, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 1.5622444017681438e-06, | |
| "loss": 0.1268, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 1.5278686498784512e-06, | |
| "loss": 0.131, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 1.4938440145236188e-06, | |
| "loss": 0.137, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 1.4601719058106845e-06, | |
| "loss": 0.0926, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 1.426853719236676e-06, | |
| "loss": 0.1523, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 1.3938908356307846e-06, | |
| "loss": 0.1241, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 1.3612846210971153e-06, | |
| "loss": 0.1101, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 1.3290364269580935e-06, | |
| "loss": 0.1186, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 1.2971475896984475e-06, | |
| "loss": 0.1484, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 1.2656194309098292e-06, | |
| "loss": 0.1324, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 1.2344532572360325e-06, | |
| "loss": 0.1282, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 1.2036503603188464e-06, | |
| "loss": 0.1177, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 1.1732120167445248e-06, | |
| "loss": 0.1162, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 1.1431394879908863e-06, | |
| "loss": 0.1312, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 1.113434020375015e-06, | |
| "loss": 0.1327, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 1.0840968450016276e-06, | |
| "loss": 0.1176, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 1.0551291777120465e-06, | |
| "loss": 0.1267, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 1.0265322190337989e-06, | |
| "loss": 0.1361, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 9.983071541308898e-07, | |
| "loss": 0.1278, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 9.704551527546503e-07, | |
| "loss": 0.1351, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 9.42977369195286e-07, | |
| "loss": 0.1288, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 9.15874942234024e-07, | |
| "loss": 0.1268, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 8.891489950959254e-07, | |
| "loss": 0.1196, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 8.628006354033291e-07, | |
| "loss": 0.1106, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 8.368309551299536e-07, | |
| "loss": 0.1241, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 8.112410305556307e-07, | |
| "loss": 0.1426, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 7.860319222217206e-07, | |
| "loss": 0.1108, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 7.612046748871327e-07, | |
| "loss": 0.1264, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 7.367603174850502e-07, | |
| "loss": 0.1374, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 7.126998630802772e-07, | |
| "loss": 0.1306, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 6.890243088272453e-07, | |
| "loss": 0.1361, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 6.65734635928711e-07, | |
| "loss": 0.1253, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 6.428318095950648e-07, | |
| "loss": 0.1078, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 6.203167790043462e-07, | |
| "loss": 0.1098, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 5.98190477262901e-07, | |
| "loss": 0.1222, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 5.764538213667103e-07, | |
| "loss": 0.1405, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 5.551077121633875e-07, | |
| "loss": 0.1105, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 5.341530343148415e-07, | |
| "loss": 0.1339, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 5.135906562606118e-07, | |
| "loss": 0.1254, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 4.934214301818873e-07, | |
| "loss": 0.129, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 4.73646191966175e-07, | |
| "loss": 0.1211, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 4.5426576117266643e-07, | |
| "loss": 0.1328, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 4.3528094099827056e-07, | |
| "loss": 0.1205, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 4.1669251824432267e-07, | |
| "loss": 0.1294, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 3.985012632839824e-07, | |
| "loss": 0.1098, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 3.8070793003030296e-07, | |
| "loss": 0.133, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 3.6331325590498344e-07, | |
| "loss": 0.1229, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 3.4631796180781496e-07, | |
| "loss": 0.1338, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 3.2972275208679625e-07, | |
| "loss": 0.1415, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 3.135283145089485e-07, | |
| "loss": 0.1141, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 2.9773532023180897e-07, | |
| "loss": 0.1359, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 2.823444237756123e-07, | |
| "loss": 0.1313, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 2.6735626299617456e-07, | |
| "loss": 0.1341, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 2.5277145905844957e-07, | |
| "loss": 0.1271, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 2.385906164107854e-07, | |
| "loss": 0.1196, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 2.248143227598809e-07, | |
| "loss": 0.1371, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 2.1144314904642194e-07, | |
| "loss": 0.133, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1.9847764942142246e-07, | |
| "loss": 0.1271, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1.8591836122326068e-07, | |
| "loss": 0.1497, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1.7376580495540142e-07, | |
| "loss": 0.119, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.6202048426483652e-07, | |
| "loss": 0.1354, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.5068288592120284e-07, | |
| "loss": 0.1172, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1.3975347979661046e-07, | |
| "loss": 0.1363, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1.2923271884617282e-07, | |
| "loss": 0.1168, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1.1912103908922945e-07, | |
| "loss": 0.0889, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 1.094188595912804e-07, | |
| "loss": 0.1144, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 1.0012658244661799e-07, | |
| "loss": 0.1407, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 9.124459276165787e-08, | |
| "loss": 0.1372, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 8.27732586389851e-08, | |
| "loss": 0.1334, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 7.471293116209421e-08, | |
| "loss": 0.1193, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 6.706394438083962e-08, | |
| "loss": 0.1377, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 5.982661529759459e-08, | |
| "loss": 0.1228, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 5.300124385410943e-08, | |
| "loss": 0.1348, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 4.6588112919082605e-08, | |
| "loss": 0.1328, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 4.058748827643899e-08, | |
| "loss": 0.1462, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 3.4999618614309784e-08, | |
| "loss": 0.1062, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 2.982473551473297e-08, | |
| "loss": 0.1318, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 2.5063053444052133e-08, | |
| "loss": 0.1102, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 2.0714769744023533e-08, | |
| "loss": 0.1217, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 1.678006462364823e-08, | |
| "loss": 0.1309, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.325910115169471e-08, | |
| "loss": 0.1239, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.0152025249943187e-08, | |
| "loss": 0.1392, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 7.45896568713933e-09, | |
| "loss": 0.1274, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 5.1800340736563034e-09, | |
| "loss": 0.103, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 3.315324856869584e-09, | |
| "loss": 0.1315, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 1.8649153172423106e-09, | |
| "loss": 0.1313, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 8.288655651234045e-10, | |
| "loss": 0.1236, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 2.0721853825733662e-10, | |
| "loss": 0.1198, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.1186, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "step": 504, | |
| "total_flos": 6.677125143849861e+17, | |
| "train_loss": 0.14185471720402204, | |
| "train_runtime": 5536.4545, | |
| "train_samples_per_second": 11.652, | |
| "train_steps_per_second": 0.091 | |
| } | |
| ], | |
| "max_steps": 504, | |
| "num_train_epochs": 2, | |
| "total_flos": 6.677125143849861e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |