Upload 14 files
Browse files- added_tokens.json +7 -0
- config.json +226 -0
- generation_config.json +7 -0
- merges.txt +0 -0
- model-00001-of-00004.safetensors +3 -0
- model-00002-of-00004.safetensors +3 -0
- model-00003-of-00004.safetensors +3 -0
- model-00004-of-00004.safetensors +3 -0
- model.safetensors.index.json +0 -0
- special_tokens_map.json +20 -0
- tokenizer_config.json +64 -0
- trainer_state.json +2283 -0
- training_args.bin +3 -0
- vocab.json +0 -0
added_tokens.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"<image>": 151646,
|
| 3 |
+
"<video>": 151647,
|
| 4 |
+
"<|endoftext|>": 151643,
|
| 5 |
+
"<|im_end|>": 151645,
|
| 6 |
+
"<|im_start|>": 151644
|
| 7 |
+
}
|
config.json
ADDED
|
@@ -0,0 +1,226 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "/raid/raushan/si-7b",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"LlavaInterleaveForConditionalGeneration"
|
| 5 |
+
],
|
| 6 |
+
"btadapter": false,
|
| 7 |
+
"btadapter_depth": 4,
|
| 8 |
+
"clip_post_pretrain": null,
|
| 9 |
+
"clip_weight": "/code/ppllava_v0/siglip-so400m-patch14-384",
|
| 10 |
+
"frame_shape": [
|
| 11 |
+
27,
|
| 12 |
+
27
|
| 13 |
+
],
|
| 14 |
+
"hidden_size": 3584,
|
| 15 |
+
"ignore_index": -100,
|
| 16 |
+
"image_grid_pinpoints": [
|
| 17 |
+
[
|
| 18 |
+
384,
|
| 19 |
+
384
|
| 20 |
+
],
|
| 21 |
+
[
|
| 22 |
+
384,
|
| 23 |
+
768
|
| 24 |
+
],
|
| 25 |
+
[
|
| 26 |
+
384,
|
| 27 |
+
1152
|
| 28 |
+
],
|
| 29 |
+
[
|
| 30 |
+
384,
|
| 31 |
+
1536
|
| 32 |
+
],
|
| 33 |
+
[
|
| 34 |
+
384,
|
| 35 |
+
1920
|
| 36 |
+
],
|
| 37 |
+
[
|
| 38 |
+
384,
|
| 39 |
+
2304
|
| 40 |
+
],
|
| 41 |
+
[
|
| 42 |
+
768,
|
| 43 |
+
384
|
| 44 |
+
],
|
| 45 |
+
[
|
| 46 |
+
768,
|
| 47 |
+
768
|
| 48 |
+
],
|
| 49 |
+
[
|
| 50 |
+
768,
|
| 51 |
+
1152
|
| 52 |
+
],
|
| 53 |
+
[
|
| 54 |
+
768,
|
| 55 |
+
1536
|
| 56 |
+
],
|
| 57 |
+
[
|
| 58 |
+
768,
|
| 59 |
+
1920
|
| 60 |
+
],
|
| 61 |
+
[
|
| 62 |
+
768,
|
| 63 |
+
2304
|
| 64 |
+
],
|
| 65 |
+
[
|
| 66 |
+
1152,
|
| 67 |
+
384
|
| 68 |
+
],
|
| 69 |
+
[
|
| 70 |
+
1152,
|
| 71 |
+
768
|
| 72 |
+
],
|
| 73 |
+
[
|
| 74 |
+
1152,
|
| 75 |
+
1152
|
| 76 |
+
],
|
| 77 |
+
[
|
| 78 |
+
1152,
|
| 79 |
+
1536
|
| 80 |
+
],
|
| 81 |
+
[
|
| 82 |
+
1152,
|
| 83 |
+
1920
|
| 84 |
+
],
|
| 85 |
+
[
|
| 86 |
+
1152,
|
| 87 |
+
2304
|
| 88 |
+
],
|
| 89 |
+
[
|
| 90 |
+
1536,
|
| 91 |
+
384
|
| 92 |
+
],
|
| 93 |
+
[
|
| 94 |
+
1536,
|
| 95 |
+
768
|
| 96 |
+
],
|
| 97 |
+
[
|
| 98 |
+
1536,
|
| 99 |
+
1152
|
| 100 |
+
],
|
| 101 |
+
[
|
| 102 |
+
1536,
|
| 103 |
+
1536
|
| 104 |
+
],
|
| 105 |
+
[
|
| 106 |
+
1536,
|
| 107 |
+
1920
|
| 108 |
+
],
|
| 109 |
+
[
|
| 110 |
+
1536,
|
| 111 |
+
2304
|
| 112 |
+
],
|
| 113 |
+
[
|
| 114 |
+
1920,
|
| 115 |
+
384
|
| 116 |
+
],
|
| 117 |
+
[
|
| 118 |
+
1920,
|
| 119 |
+
768
|
| 120 |
+
],
|
| 121 |
+
[
|
| 122 |
+
1920,
|
| 123 |
+
1152
|
| 124 |
+
],
|
| 125 |
+
[
|
| 126 |
+
1920,
|
| 127 |
+
1536
|
| 128 |
+
],
|
| 129 |
+
[
|
| 130 |
+
1920,
|
| 131 |
+
1920
|
| 132 |
+
],
|
| 133 |
+
[
|
| 134 |
+
1920,
|
| 135 |
+
2304
|
| 136 |
+
],
|
| 137 |
+
[
|
| 138 |
+
2304,
|
| 139 |
+
384
|
| 140 |
+
],
|
| 141 |
+
[
|
| 142 |
+
2304,
|
| 143 |
+
768
|
| 144 |
+
],
|
| 145 |
+
[
|
| 146 |
+
2304,
|
| 147 |
+
1152
|
| 148 |
+
],
|
| 149 |
+
[
|
| 150 |
+
2304,
|
| 151 |
+
1536
|
| 152 |
+
],
|
| 153 |
+
[
|
| 154 |
+
2304,
|
| 155 |
+
1920
|
| 156 |
+
],
|
| 157 |
+
[
|
| 158 |
+
2304,
|
| 159 |
+
2304
|
| 160 |
+
]
|
| 161 |
+
],
|
| 162 |
+
"image_pooling_kernel": [
|
| 163 |
+
1,
|
| 164 |
+
3,
|
| 165 |
+
3
|
| 166 |
+
],
|
| 167 |
+
"image_pooling_stride": [
|
| 168 |
+
1,
|
| 169 |
+
3,
|
| 170 |
+
3
|
| 171 |
+
],
|
| 172 |
+
"image_token_index": 151646,
|
| 173 |
+
"long_clip": true,
|
| 174 |
+
"max_T": 64,
|
| 175 |
+
"model_type": "llava_onevision",
|
| 176 |
+
"pad_token_id": 151643,
|
| 177 |
+
"pooling": "clipST_3d",
|
| 178 |
+
"pooling_kernel": [
|
| 179 |
+
1,
|
| 180 |
+
3,
|
| 181 |
+
3
|
| 182 |
+
],
|
| 183 |
+
"pooling_stride": [
|
| 184 |
+
1,
|
| 185 |
+
3,
|
| 186 |
+
3
|
| 187 |
+
],
|
| 188 |
+
"pooling_temp": 0.01,
|
| 189 |
+
"projector_hidden_act": "gelu",
|
| 190 |
+
"qwen": true,
|
| 191 |
+
"text_config": {
|
| 192 |
+
"_name_or_path": "Qwen/Qwen2-7B-Instruct",
|
| 193 |
+
"architectures": [
|
| 194 |
+
"Qwen2ForCausalLM"
|
| 195 |
+
],
|
| 196 |
+
"bos_token_id": 151643,
|
| 197 |
+
"eos_token_id": 151645,
|
| 198 |
+
"hidden_size": 3584,
|
| 199 |
+
"intermediate_size": 18944,
|
| 200 |
+
"model_type": "qwen2",
|
| 201 |
+
"num_attention_heads": 28,
|
| 202 |
+
"num_hidden_layers": 28,
|
| 203 |
+
"num_key_value_heads": 4,
|
| 204 |
+
"rope_theta": 1000000.0,
|
| 205 |
+
"torch_dtype": "bfloat16",
|
| 206 |
+
"vocab_size": 152128
|
| 207 |
+
},
|
| 208 |
+
"tie_word_embeddings": false,
|
| 209 |
+
"torch_dtype": "bfloat16",
|
| 210 |
+
"transformers_version": "4.45.2",
|
| 211 |
+
"use_image_newline_parameter": true,
|
| 212 |
+
"video_token_index": 151647,
|
| 213 |
+
"vision_aspect_ratio": "anyres_max_9",
|
| 214 |
+
"vision_config": {
|
| 215 |
+
"hidden_size": 1152,
|
| 216 |
+
"image_size": 384,
|
| 217 |
+
"intermediate_size": 4304,
|
| 218 |
+
"model_type": "siglip_vision_model",
|
| 219 |
+
"num_attention_heads": 16,
|
| 220 |
+
"num_hidden_layers": 26,
|
| 221 |
+
"patch_size": 14,
|
| 222 |
+
"vision_use_head": false
|
| 223 |
+
},
|
| 224 |
+
"vision_feature_layer": -1,
|
| 225 |
+
"vision_feature_select_strategy": "full"
|
| 226 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 151643,
|
| 4 |
+
"eos_token_id": 151645,
|
| 5 |
+
"pad_token_id": 151643,
|
| 6 |
+
"transformers_version": "4.45.2"
|
| 7 |
+
}
|
merges.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
model-00001-of-00004.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:28cfa9668f16fb4c453dfd5ed1bb6e382b02e1abca853041e505ea006dc81826
|
| 3 |
+
size 4909741188
|
model-00002-of-00004.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f417c115204369e06b12f8d38ebe95fd2ef6cce1fa5dd40ac25f96ac08973156
|
| 3 |
+
size 4991497768
|
model-00003-of-00004.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:338084512715be8f33e3c2b00d182dc0c5721f18222f8a88f4a3b9853afb62a2
|
| 3 |
+
size 4932752872
|
model-00004-of-00004.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f0f29190e06cadb4a0c1a3c3701af442c2374a09286004779e98d6d25caccdb6
|
| 3 |
+
size 2158500536
|
model.safetensors.index.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
special_tokens_map.json
ADDED
|
@@ -0,0 +1,20 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"additional_special_tokens": [
|
| 3 |
+
"<|im_start|>",
|
| 4 |
+
"<|im_end|>"
|
| 5 |
+
],
|
| 6 |
+
"eos_token": {
|
| 7 |
+
"content": "<|im_end|>",
|
| 8 |
+
"lstrip": false,
|
| 9 |
+
"normalized": false,
|
| 10 |
+
"rstrip": false,
|
| 11 |
+
"single_word": false
|
| 12 |
+
},
|
| 13 |
+
"pad_token": {
|
| 14 |
+
"content": "<|endoftext|>",
|
| 15 |
+
"lstrip": false,
|
| 16 |
+
"normalized": false,
|
| 17 |
+
"rstrip": false,
|
| 18 |
+
"single_word": false
|
| 19 |
+
}
|
| 20 |
+
}
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,64 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_prefix_space": false,
|
| 3 |
+
"added_tokens_decoder": {
|
| 4 |
+
"151643": {
|
| 5 |
+
"content": "<|endoftext|>",
|
| 6 |
+
"lstrip": false,
|
| 7 |
+
"normalized": false,
|
| 8 |
+
"rstrip": false,
|
| 9 |
+
"single_word": false,
|
| 10 |
+
"special": true
|
| 11 |
+
},
|
| 12 |
+
"151644": {
|
| 13 |
+
"content": "<|im_start|>",
|
| 14 |
+
"lstrip": false,
|
| 15 |
+
"normalized": false,
|
| 16 |
+
"rstrip": false,
|
| 17 |
+
"single_word": false,
|
| 18 |
+
"special": true
|
| 19 |
+
},
|
| 20 |
+
"151645": {
|
| 21 |
+
"content": "<|im_end|>",
|
| 22 |
+
"lstrip": false,
|
| 23 |
+
"normalized": false,
|
| 24 |
+
"rstrip": false,
|
| 25 |
+
"single_word": false,
|
| 26 |
+
"special": true
|
| 27 |
+
},
|
| 28 |
+
"151646": {
|
| 29 |
+
"content": "<image>",
|
| 30 |
+
"lstrip": false,
|
| 31 |
+
"normalized": false,
|
| 32 |
+
"rstrip": false,
|
| 33 |
+
"single_word": false,
|
| 34 |
+
"special": true
|
| 35 |
+
},
|
| 36 |
+
"151647": {
|
| 37 |
+
"content": "<video>",
|
| 38 |
+
"lstrip": false,
|
| 39 |
+
"normalized": false,
|
| 40 |
+
"rstrip": false,
|
| 41 |
+
"single_word": false,
|
| 42 |
+
"special": true
|
| 43 |
+
}
|
| 44 |
+
},
|
| 45 |
+
"additional_special_tokens": [
|
| 46 |
+
"<|im_start|>",
|
| 47 |
+
"<|im_end|>"
|
| 48 |
+
],
|
| 49 |
+
"bos_token": null,
|
| 50 |
+
"chat_template": "{% for message in messages %}{% if loop.first and messages[0]['role'] != 'system' %}{{ '<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n' }}{% endif %}{{'<|im_start|>' + message['role'] + '\n' + message['content'] + '<|im_end|>' + '\n'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant\n' }}{% endif %}",
|
| 51 |
+
"clean_up_tokenization_spaces": false,
|
| 52 |
+
"eos_token": "<|im_end|>",
|
| 53 |
+
"errors": "replace",
|
| 54 |
+
"max_length": null,
|
| 55 |
+
"model_max_length": 2048,
|
| 56 |
+
"pad_to_multiple_of": null,
|
| 57 |
+
"pad_token": "<|endoftext|>",
|
| 58 |
+
"pad_token_type_id": 0,
|
| 59 |
+
"padding_side": "right",
|
| 60 |
+
"processor_class": "LlavaOnevisionProcessor",
|
| 61 |
+
"split_special_tokens": false,
|
| 62 |
+
"tokenizer_class": "Qwen2Tokenizer",
|
| 63 |
+
"unk_token": null
|
| 64 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2283 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.23277467411545624,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 125,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.00186219739292365,
|
| 13 |
+
"grad_norm": 40.418399810791016,
|
| 14 |
+
"learning_rate": 9.259259259259259e-08,
|
| 15 |
+
"logps/chosen": -143.09214782714844,
|
| 16 |
+
"logps/rejected": -159.23831176757812,
|
| 17 |
+
"loss": 0.6931,
|
| 18 |
+
"losses/dpo": 0.6931471824645996,
|
| 19 |
+
"losses/sft": 1.9754583835601807,
|
| 20 |
+
"losses/total": 0.6931471824645996,
|
| 21 |
+
"ref_logps/chosen": -143.09214782714844,
|
| 22 |
+
"ref_logps/rejected": -159.23831176757812,
|
| 23 |
+
"rewards/accuracies": 0.0,
|
| 24 |
+
"rewards/chosen": 0.0,
|
| 25 |
+
"rewards/margins": 0.0,
|
| 26 |
+
"rewards/rejected": 0.0,
|
| 27 |
+
"step": 1
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.0037243947858473,
|
| 31 |
+
"grad_norm": 40.48415756225586,
|
| 32 |
+
"learning_rate": 1.8518518518518518e-07,
|
| 33 |
+
"logps/chosen": -139.04046630859375,
|
| 34 |
+
"logps/rejected": -167.1217041015625,
|
| 35 |
+
"loss": 0.6931,
|
| 36 |
+
"losses/dpo": 0.6931471824645996,
|
| 37 |
+
"losses/sft": 2.354458808898926,
|
| 38 |
+
"losses/total": 0.6931471824645996,
|
| 39 |
+
"ref_logps/chosen": -139.04046630859375,
|
| 40 |
+
"ref_logps/rejected": -167.1217041015625,
|
| 41 |
+
"rewards/accuracies": 0.0,
|
| 42 |
+
"rewards/chosen": 0.0,
|
| 43 |
+
"rewards/margins": 0.0,
|
| 44 |
+
"rewards/rejected": 0.0,
|
| 45 |
+
"step": 2
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.00558659217877095,
|
| 49 |
+
"grad_norm": 43.48381042480469,
|
| 50 |
+
"learning_rate": 2.7777777777777776e-07,
|
| 51 |
+
"logps/chosen": -145.57992553710938,
|
| 52 |
+
"logps/rejected": -169.97146606445312,
|
| 53 |
+
"loss": 0.6981,
|
| 54 |
+
"losses/dpo": 0.6891708374023438,
|
| 55 |
+
"losses/sft": 2.1739914417266846,
|
| 56 |
+
"losses/total": 0.6891708374023438,
|
| 57 |
+
"ref_logps/chosen": -145.53663635253906,
|
| 58 |
+
"ref_logps/rejected": -170.02041625976562,
|
| 59 |
+
"rewards/accuracies": 0.484375,
|
| 60 |
+
"rewards/chosen": -0.00432817917317152,
|
| 61 |
+
"rewards/margins": -0.009223854169249535,
|
| 62 |
+
"rewards/rejected": 0.004895674996078014,
|
| 63 |
+
"step": 3
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.0074487895716946,
|
| 67 |
+
"grad_norm": 40.21121597290039,
|
| 68 |
+
"learning_rate": 3.7037037037037036e-07,
|
| 69 |
+
"logps/chosen": -145.57131958007812,
|
| 70 |
+
"logps/rejected": -162.80123901367188,
|
| 71 |
+
"loss": 0.6915,
|
| 72 |
+
"losses/dpo": 0.6871472597122192,
|
| 73 |
+
"losses/sft": 2.114041805267334,
|
| 74 |
+
"losses/total": 0.6871472597122192,
|
| 75 |
+
"ref_logps/chosen": -145.52674865722656,
|
| 76 |
+
"ref_logps/rejected": -162.71473693847656,
|
| 77 |
+
"rewards/accuracies": 0.578125,
|
| 78 |
+
"rewards/chosen": -0.004457235801964998,
|
| 79 |
+
"rewards/margins": 0.004192234016954899,
|
| 80 |
+
"rewards/rejected": -0.00864946935325861,
|
| 81 |
+
"step": 4
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.00931098696461825,
|
| 85 |
+
"grad_norm": 43.5928955078125,
|
| 86 |
+
"learning_rate": 4.6296296296296297e-07,
|
| 87 |
+
"logps/chosen": -149.61972045898438,
|
| 88 |
+
"logps/rejected": -171.2332763671875,
|
| 89 |
+
"loss": 0.6897,
|
| 90 |
+
"losses/dpo": 0.6956955194473267,
|
| 91 |
+
"losses/sft": 1.9878240823745728,
|
| 92 |
+
"losses/total": 0.6956955194473267,
|
| 93 |
+
"ref_logps/chosen": -149.57879638671875,
|
| 94 |
+
"ref_logps/rejected": -171.1157684326172,
|
| 95 |
+
"rewards/accuracies": 0.5625,
|
| 96 |
+
"rewards/chosen": -0.004093348979949951,
|
| 97 |
+
"rewards/margins": 0.007657159119844437,
|
| 98 |
+
"rewards/rejected": -0.011750508099794388,
|
| 99 |
+
"step": 5
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.0111731843575419,
|
| 103 |
+
"grad_norm": 43.37643051147461,
|
| 104 |
+
"learning_rate": 5.555555555555555e-07,
|
| 105 |
+
"logps/chosen": -144.1446075439453,
|
| 106 |
+
"logps/rejected": -165.30984497070312,
|
| 107 |
+
"loss": 0.6839,
|
| 108 |
+
"losses/dpo": 0.7011494636535645,
|
| 109 |
+
"losses/sft": 2.2325987815856934,
|
| 110 |
+
"losses/total": 0.7011494636535645,
|
| 111 |
+
"ref_logps/chosen": -144.15261840820312,
|
| 112 |
+
"ref_logps/rejected": -165.12521362304688,
|
| 113 |
+
"rewards/accuracies": 0.609375,
|
| 114 |
+
"rewards/chosen": 0.0007997989887371659,
|
| 115 |
+
"rewards/margins": 0.019263559952378273,
|
| 116 |
+
"rewards/rejected": -0.01846376061439514,
|
| 117 |
+
"step": 6
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.01303538175046555,
|
| 121 |
+
"grad_norm": 41.759315490722656,
|
| 122 |
+
"learning_rate": 6.481481481481481e-07,
|
| 123 |
+
"logps/chosen": -140.94808959960938,
|
| 124 |
+
"logps/rejected": -161.61846923828125,
|
| 125 |
+
"loss": 0.6677,
|
| 126 |
+
"losses/dpo": 0.6434986591339111,
|
| 127 |
+
"losses/sft": 2.4416794776916504,
|
| 128 |
+
"losses/total": 0.6434986591339111,
|
| 129 |
+
"ref_logps/chosen": -140.9979248046875,
|
| 130 |
+
"ref_logps/rejected": -161.13681030273438,
|
| 131 |
+
"rewards/accuracies": 0.734375,
|
| 132 |
+
"rewards/chosen": 0.004983758553862572,
|
| 133 |
+
"rewards/margins": 0.05314992740750313,
|
| 134 |
+
"rewards/rejected": -0.048166170716285706,
|
| 135 |
+
"step": 7
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.0148975791433892,
|
| 139 |
+
"grad_norm": 40.88422393798828,
|
| 140 |
+
"learning_rate": 7.407407407407407e-07,
|
| 141 |
+
"logps/chosen": -150.63833618164062,
|
| 142 |
+
"logps/rejected": -174.64923095703125,
|
| 143 |
+
"loss": 0.6588,
|
| 144 |
+
"losses/dpo": 0.639857292175293,
|
| 145 |
+
"losses/sft": 2.3609848022460938,
|
| 146 |
+
"losses/total": 0.639857292175293,
|
| 147 |
+
"ref_logps/chosen": -150.88568115234375,
|
| 148 |
+
"ref_logps/rejected": -174.18106079101562,
|
| 149 |
+
"rewards/accuracies": 0.796875,
|
| 150 |
+
"rewards/chosen": 0.02473468706011772,
|
| 151 |
+
"rewards/margins": 0.0715511366724968,
|
| 152 |
+
"rewards/rejected": -0.046816445887088776,
|
| 153 |
+
"step": 8
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.01675977653631285,
|
| 157 |
+
"grad_norm": 36.576690673828125,
|
| 158 |
+
"learning_rate": 8.333333333333333e-07,
|
| 159 |
+
"logps/chosen": -146.96463012695312,
|
| 160 |
+
"logps/rejected": -174.3262939453125,
|
| 161 |
+
"loss": 0.6259,
|
| 162 |
+
"losses/dpo": 0.6055787801742554,
|
| 163 |
+
"losses/sft": 2.1100568771362305,
|
| 164 |
+
"losses/total": 0.6055787801742554,
|
| 165 |
+
"ref_logps/chosen": -147.0458221435547,
|
| 166 |
+
"ref_logps/rejected": -172.9729461669922,
|
| 167 |
+
"rewards/accuracies": 0.9375,
|
| 168 |
+
"rewards/chosen": 0.008118623867630959,
|
| 169 |
+
"rewards/margins": 0.14345335960388184,
|
| 170 |
+
"rewards/rejected": -0.13533474504947662,
|
| 171 |
+
"step": 9
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.0186219739292365,
|
| 175 |
+
"grad_norm": 41.190025329589844,
|
| 176 |
+
"learning_rate": 9.259259259259259e-07,
|
| 177 |
+
"logps/chosen": -154.90782165527344,
|
| 178 |
+
"logps/rejected": -160.24624633789062,
|
| 179 |
+
"loss": 0.5904,
|
| 180 |
+
"losses/dpo": 0.5785373449325562,
|
| 181 |
+
"losses/sft": 2.5960636138916016,
|
| 182 |
+
"losses/total": 0.5785373449325562,
|
| 183 |
+
"ref_logps/chosen": -155.20986938476562,
|
| 184 |
+
"ref_logps/rejected": -158.2913055419922,
|
| 185 |
+
"rewards/accuracies": 0.921875,
|
| 186 |
+
"rewards/chosen": 0.030205056071281433,
|
| 187 |
+
"rewards/margins": 0.2256987988948822,
|
| 188 |
+
"rewards/rejected": -0.19549374282360077,
|
| 189 |
+
"step": 10
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.020484171322160148,
|
| 193 |
+
"grad_norm": 32.080810546875,
|
| 194 |
+
"learning_rate": 1.0185185185185185e-06,
|
| 195 |
+
"logps/chosen": -138.2403106689453,
|
| 196 |
+
"logps/rejected": -162.02291870117188,
|
| 197 |
+
"loss": 0.5437,
|
| 198 |
+
"losses/dpo": 0.5474151968955994,
|
| 199 |
+
"losses/sft": 1.9134595394134521,
|
| 200 |
+
"losses/total": 0.5474151968955994,
|
| 201 |
+
"ref_logps/chosen": -138.54551696777344,
|
| 202 |
+
"ref_logps/rejected": -158.88693237304688,
|
| 203 |
+
"rewards/accuracies": 0.9375,
|
| 204 |
+
"rewards/chosen": 0.030520910397171974,
|
| 205 |
+
"rewards/margins": 0.34412044286727905,
|
| 206 |
+
"rewards/rejected": -0.31359952688217163,
|
| 207 |
+
"step": 11
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.0223463687150838,
|
| 211 |
+
"grad_norm": 29.643903732299805,
|
| 212 |
+
"learning_rate": 1.111111111111111e-06,
|
| 213 |
+
"logps/chosen": -154.15118408203125,
|
| 214 |
+
"logps/rejected": -180.0041961669922,
|
| 215 |
+
"loss": 0.4393,
|
| 216 |
+
"losses/dpo": 0.4857974946498871,
|
| 217 |
+
"losses/sft": 2.032329797744751,
|
| 218 |
+
"losses/total": 0.4857974946498871,
|
| 219 |
+
"ref_logps/chosen": -154.87046813964844,
|
| 220 |
+
"ref_logps/rejected": -174.29629516601562,
|
| 221 |
+
"rewards/accuracies": 0.96875,
|
| 222 |
+
"rewards/chosen": 0.0719279944896698,
|
| 223 |
+
"rewards/margins": 0.6427179574966431,
|
| 224 |
+
"rewards/rejected": -0.5707899332046509,
|
| 225 |
+
"step": 12
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 0.024208566108007448,
|
| 229 |
+
"grad_norm": 27.43463134765625,
|
| 230 |
+
"learning_rate": 1.2037037037037037e-06,
|
| 231 |
+
"logps/chosen": -146.56866455078125,
|
| 232 |
+
"logps/rejected": -173.71633911132812,
|
| 233 |
+
"loss": 0.4487,
|
| 234 |
+
"losses/dpo": 0.4681214988231659,
|
| 235 |
+
"losses/sft": 2.5795841217041016,
|
| 236 |
+
"losses/total": 0.4681214988231659,
|
| 237 |
+
"ref_logps/chosen": -146.011962890625,
|
| 238 |
+
"ref_logps/rejected": -166.83184814453125,
|
| 239 |
+
"rewards/accuracies": 0.9375,
|
| 240 |
+
"rewards/chosen": -0.055669985711574554,
|
| 241 |
+
"rewards/margins": 0.6327812075614929,
|
| 242 |
+
"rewards/rejected": -0.6884511709213257,
|
| 243 |
+
"step": 13
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.0260707635009311,
|
| 247 |
+
"grad_norm": 24.864574432373047,
|
| 248 |
+
"learning_rate": 1.2962962962962962e-06,
|
| 249 |
+
"logps/chosen": -151.02015686035156,
|
| 250 |
+
"logps/rejected": -181.14610290527344,
|
| 251 |
+
"loss": 0.3711,
|
| 252 |
+
"losses/dpo": 0.4647141695022583,
|
| 253 |
+
"losses/sft": 2.417238473892212,
|
| 254 |
+
"losses/total": 0.4647141695022583,
|
| 255 |
+
"ref_logps/chosen": -151.75128173828125,
|
| 256 |
+
"ref_logps/rejected": -172.87350463867188,
|
| 257 |
+
"rewards/accuracies": 0.96875,
|
| 258 |
+
"rewards/chosen": 0.07311299443244934,
|
| 259 |
+
"rewards/margins": 0.9003715515136719,
|
| 260 |
+
"rewards/rejected": -0.8272585272789001,
|
| 261 |
+
"step": 14
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.027932960893854747,
|
| 265 |
+
"grad_norm": 22.773723602294922,
|
| 266 |
+
"learning_rate": 1.3888888888888892e-06,
|
| 267 |
+
"logps/chosen": -146.68460083007812,
|
| 268 |
+
"logps/rejected": -180.87896728515625,
|
| 269 |
+
"loss": 0.3584,
|
| 270 |
+
"losses/dpo": 0.33828622102737427,
|
| 271 |
+
"losses/sft": 2.191209077835083,
|
| 272 |
+
"losses/total": 0.33828622102737427,
|
| 273 |
+
"ref_logps/chosen": -145.39768981933594,
|
| 274 |
+
"ref_logps/rejected": -170.00677490234375,
|
| 275 |
+
"rewards/accuracies": 0.984375,
|
| 276 |
+
"rewards/chosen": -0.12869027256965637,
|
| 277 |
+
"rewards/margins": 0.9585269689559937,
|
| 278 |
+
"rewards/rejected": -1.0872172117233276,
|
| 279 |
+
"step": 15
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.0297951582867784,
|
| 283 |
+
"grad_norm": 17.319433212280273,
|
| 284 |
+
"learning_rate": 1.4814814814814815e-06,
|
| 285 |
+
"logps/chosen": -157.9118194580078,
|
| 286 |
+
"logps/rejected": -180.584228515625,
|
| 287 |
+
"loss": 0.253,
|
| 288 |
+
"losses/dpo": 0.30005261301994324,
|
| 289 |
+
"losses/sft": 2.218407154083252,
|
| 290 |
+
"losses/total": 0.30005261301994324,
|
| 291 |
+
"ref_logps/chosen": -155.03277587890625,
|
| 292 |
+
"ref_logps/rejected": -161.18026733398438,
|
| 293 |
+
"rewards/accuracies": 0.984375,
|
| 294 |
+
"rewards/chosen": -0.2879039943218231,
|
| 295 |
+
"rewards/margins": 1.652491807937622,
|
| 296 |
+
"rewards/rejected": -1.940395712852478,
|
| 297 |
+
"step": 16
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.03165735567970205,
|
| 301 |
+
"grad_norm": 14.82257080078125,
|
| 302 |
+
"learning_rate": 1.5740740740740742e-06,
|
| 303 |
+
"logps/chosen": -151.82461547851562,
|
| 304 |
+
"logps/rejected": -182.52401733398438,
|
| 305 |
+
"loss": 0.2197,
|
| 306 |
+
"losses/dpo": 0.2500099837779999,
|
| 307 |
+
"losses/sft": 2.3733954429626465,
|
| 308 |
+
"losses/total": 0.2500099837779999,
|
| 309 |
+
"ref_logps/chosen": -147.43585205078125,
|
| 310 |
+
"ref_logps/rejected": -158.81658935546875,
|
| 311 |
+
"rewards/accuracies": 0.953125,
|
| 312 |
+
"rewards/chosen": -0.438876748085022,
|
| 313 |
+
"rewards/margins": 1.9318653345108032,
|
| 314 |
+
"rewards/rejected": -2.3707423210144043,
|
| 315 |
+
"step": 17
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 0.0335195530726257,
|
| 319 |
+
"grad_norm": 15.098052978515625,
|
| 320 |
+
"learning_rate": 1.6666666666666667e-06,
|
| 321 |
+
"logps/chosen": -146.62161254882812,
|
| 322 |
+
"logps/rejected": -192.061767578125,
|
| 323 |
+
"loss": 0.2086,
|
| 324 |
+
"losses/dpo": 0.087049201130867,
|
| 325 |
+
"losses/sft": 2.235438585281372,
|
| 326 |
+
"losses/total": 0.087049201130867,
|
| 327 |
+
"ref_logps/chosen": -142.35076904296875,
|
| 328 |
+
"ref_logps/rejected": -164.73597717285156,
|
| 329 |
+
"rewards/accuracies": 0.921875,
|
| 330 |
+
"rewards/chosen": -0.42708471417427063,
|
| 331 |
+
"rewards/margins": 2.305494785308838,
|
| 332 |
+
"rewards/rejected": -2.732579469680786,
|
| 333 |
+
"step": 18
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.035381750465549346,
|
| 337 |
+
"grad_norm": 14.139907836914062,
|
| 338 |
+
"learning_rate": 1.7592592592592594e-06,
|
| 339 |
+
"logps/chosen": -151.61148071289062,
|
| 340 |
+
"logps/rejected": -204.13427734375,
|
| 341 |
+
"loss": 0.1831,
|
| 342 |
+
"losses/dpo": 0.2179492712020874,
|
| 343 |
+
"losses/sft": 2.547309637069702,
|
| 344 |
+
"losses/total": 0.2179492712020874,
|
| 345 |
+
"ref_logps/chosen": -145.25982666015625,
|
| 346 |
+
"ref_logps/rejected": -172.7918701171875,
|
| 347 |
+
"rewards/accuracies": 0.984375,
|
| 348 |
+
"rewards/chosen": -0.6351636648178101,
|
| 349 |
+
"rewards/margins": 2.4990787506103516,
|
| 350 |
+
"rewards/rejected": -3.134242534637451,
|
| 351 |
+
"step": 19
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 0.037243947858473,
|
| 355 |
+
"grad_norm": 15.214645385742188,
|
| 356 |
+
"learning_rate": 1.8518518518518519e-06,
|
| 357 |
+
"logps/chosen": -149.88876342773438,
|
| 358 |
+
"logps/rejected": -218.4535675048828,
|
| 359 |
+
"loss": 0.181,
|
| 360 |
+
"losses/dpo": 0.18753334879875183,
|
| 361 |
+
"losses/sft": 2.394906759262085,
|
| 362 |
+
"losses/total": 0.18753334879875183,
|
| 363 |
+
"ref_logps/chosen": -142.5288543701172,
|
| 364 |
+
"ref_logps/rejected": -185.18878173828125,
|
| 365 |
+
"rewards/accuracies": 0.9375,
|
| 366 |
+
"rewards/chosen": -0.7359906435012817,
|
| 367 |
+
"rewards/margins": 2.590489387512207,
|
| 368 |
+
"rewards/rejected": -3.3264803886413574,
|
| 369 |
+
"step": 20
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.03910614525139665,
|
| 373 |
+
"grad_norm": 15.793328285217285,
|
| 374 |
+
"learning_rate": 1.944444444444445e-06,
|
| 375 |
+
"logps/chosen": -155.686279296875,
|
| 376 |
+
"logps/rejected": -189.35899353027344,
|
| 377 |
+
"loss": 0.1946,
|
| 378 |
+
"losses/dpo": 0.2072586715221405,
|
| 379 |
+
"losses/sft": 2.1763200759887695,
|
| 380 |
+
"losses/total": 0.2072586715221405,
|
| 381 |
+
"ref_logps/chosen": -148.760986328125,
|
| 382 |
+
"ref_logps/rejected": -153.56321716308594,
|
| 383 |
+
"rewards/accuracies": 0.96875,
|
| 384 |
+
"rewards/chosen": -0.6925297379493713,
|
| 385 |
+
"rewards/margins": 2.88704776763916,
|
| 386 |
+
"rewards/rejected": -3.5795774459838867,
|
| 387 |
+
"step": 21
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.040968342644320296,
|
| 391 |
+
"grad_norm": 15.932718276977539,
|
| 392 |
+
"learning_rate": 2.037037037037037e-06,
|
| 393 |
+
"logps/chosen": -147.47117614746094,
|
| 394 |
+
"logps/rejected": -206.65597534179688,
|
| 395 |
+
"loss": 0.1655,
|
| 396 |
+
"losses/dpo": 0.10909190773963928,
|
| 397 |
+
"losses/sft": 2.1695094108581543,
|
| 398 |
+
"losses/total": 0.10909190773963928,
|
| 399 |
+
"ref_logps/chosen": -137.73451232910156,
|
| 400 |
+
"ref_logps/rejected": -160.2368927001953,
|
| 401 |
+
"rewards/accuracies": 0.9375,
|
| 402 |
+
"rewards/chosen": -0.9736669063568115,
|
| 403 |
+
"rewards/margins": 3.6682400703430176,
|
| 404 |
+
"rewards/rejected": -4.641907691955566,
|
| 405 |
+
"step": 22
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 0.04283054003724395,
|
| 409 |
+
"grad_norm": 11.131953239440918,
|
| 410 |
+
"learning_rate": 2.1296296296296298e-06,
|
| 411 |
+
"logps/chosen": -172.5790252685547,
|
| 412 |
+
"logps/rejected": -234.46731567382812,
|
| 413 |
+
"loss": 0.0974,
|
| 414 |
+
"losses/dpo": 0.25966331362724304,
|
| 415 |
+
"losses/sft": 2.784914255142212,
|
| 416 |
+
"losses/total": 0.25966331362724304,
|
| 417 |
+
"ref_logps/chosen": -154.8829345703125,
|
| 418 |
+
"ref_logps/rejected": -172.43850708007812,
|
| 419 |
+
"rewards/accuracies": 0.96875,
|
| 420 |
+
"rewards/chosen": -1.7696071863174438,
|
| 421 |
+
"rewards/margins": 4.433272361755371,
|
| 422 |
+
"rewards/rejected": -6.202879905700684,
|
| 423 |
+
"step": 23
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.0446927374301676,
|
| 427 |
+
"grad_norm": 9.715975761413574,
|
| 428 |
+
"learning_rate": 2.222222222222222e-06,
|
| 429 |
+
"logps/chosen": -161.45799255371094,
|
| 430 |
+
"logps/rejected": -231.06072998046875,
|
| 431 |
+
"loss": 0.0859,
|
| 432 |
+
"losses/dpo": 0.04074842855334282,
|
| 433 |
+
"losses/sft": 2.738722324371338,
|
| 434 |
+
"losses/total": 0.04074842855334282,
|
| 435 |
+
"ref_logps/chosen": -142.22097778320312,
|
| 436 |
+
"ref_logps/rejected": -161.80909729003906,
|
| 437 |
+
"rewards/accuracies": 0.984375,
|
| 438 |
+
"rewards/chosen": -1.9237014055252075,
|
| 439 |
+
"rewards/margins": 5.001461029052734,
|
| 440 |
+
"rewards/rejected": -6.9251627922058105,
|
| 441 |
+
"step": 24
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 0.04655493482309125,
|
| 445 |
+
"grad_norm": 14.314465522766113,
|
| 446 |
+
"learning_rate": 2.314814814814815e-06,
|
| 447 |
+
"logps/chosen": -170.95504760742188,
|
| 448 |
+
"logps/rejected": -257.62115478515625,
|
| 449 |
+
"loss": 0.124,
|
| 450 |
+
"losses/dpo": 0.004803886171430349,
|
| 451 |
+
"losses/sft": 2.606781244277954,
|
| 452 |
+
"losses/total": 0.004803886171430349,
|
| 453 |
+
"ref_logps/chosen": -146.23434448242188,
|
| 454 |
+
"ref_logps/rejected": -173.50958251953125,
|
| 455 |
+
"rewards/accuracies": 0.953125,
|
| 456 |
+
"rewards/chosen": -2.472069263458252,
|
| 457 |
+
"rewards/margins": 5.939087867736816,
|
| 458 |
+
"rewards/rejected": -8.411157608032227,
|
| 459 |
+
"step": 25
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.048417132216014895,
|
| 463 |
+
"grad_norm": 16.382986068725586,
|
| 464 |
+
"learning_rate": 2.4074074074074075e-06,
|
| 465 |
+
"logps/chosen": -171.77096557617188,
|
| 466 |
+
"logps/rejected": -249.99264526367188,
|
| 467 |
+
"loss": 0.1927,
|
| 468 |
+
"losses/dpo": 0.023840777575969696,
|
| 469 |
+
"losses/sft": 2.574446201324463,
|
| 470 |
+
"losses/total": 0.023840777575969696,
|
| 471 |
+
"ref_logps/chosen": -146.96835327148438,
|
| 472 |
+
"ref_logps/rejected": -168.13232421875,
|
| 473 |
+
"rewards/accuracies": 0.96875,
|
| 474 |
+
"rewards/chosen": -2.4802613258361816,
|
| 475 |
+
"rewards/margins": 5.705770969390869,
|
| 476 |
+
"rewards/rejected": -8.18603229522705,
|
| 477 |
+
"step": 26
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 0.05027932960893855,
|
| 481 |
+
"grad_norm": 9.626073837280273,
|
| 482 |
+
"learning_rate": 2.5e-06,
|
| 483 |
+
"logps/chosen": -169.91448974609375,
|
| 484 |
+
"logps/rejected": -251.7610626220703,
|
| 485 |
+
"loss": 0.058,
|
| 486 |
+
"losses/dpo": 0.0014569438062608242,
|
| 487 |
+
"losses/sft": 2.5121726989746094,
|
| 488 |
+
"losses/total": 0.0014569438062608242,
|
| 489 |
+
"ref_logps/chosen": -141.93600463867188,
|
| 490 |
+
"ref_logps/rejected": -160.8459014892578,
|
| 491 |
+
"rewards/accuracies": 0.984375,
|
| 492 |
+
"rewards/chosen": -2.7978479862213135,
|
| 493 |
+
"rewards/margins": 6.293667793273926,
|
| 494 |
+
"rewards/rejected": -9.091516494750977,
|
| 495 |
+
"step": 27
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 0.0521415270018622,
|
| 499 |
+
"grad_norm": 9.746648788452148,
|
| 500 |
+
"learning_rate": 2.5925925925925925e-06,
|
| 501 |
+
"logps/chosen": -173.24583435058594,
|
| 502 |
+
"logps/rejected": -267.49298095703125,
|
| 503 |
+
"loss": 0.0456,
|
| 504 |
+
"losses/dpo": 0.010660952888429165,
|
| 505 |
+
"losses/sft": 2.7988812923431396,
|
| 506 |
+
"losses/total": 0.010660952888429165,
|
| 507 |
+
"ref_logps/chosen": -147.74620056152344,
|
| 508 |
+
"ref_logps/rejected": -170.100830078125,
|
| 509 |
+
"rewards/accuracies": 0.984375,
|
| 510 |
+
"rewards/chosen": -2.5499653816223145,
|
| 511 |
+
"rewards/margins": 7.189251899719238,
|
| 512 |
+
"rewards/rejected": -9.739216804504395,
|
| 513 |
+
"step": 28
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.054003724394785846,
|
| 517 |
+
"grad_norm": 19.729585647583008,
|
| 518 |
+
"learning_rate": 2.6851851851851856e-06,
|
| 519 |
+
"logps/chosen": -173.5810546875,
|
| 520 |
+
"logps/rejected": -260.90985107421875,
|
| 521 |
+
"loss": 0.1598,
|
| 522 |
+
"losses/dpo": 0.11473169922828674,
|
| 523 |
+
"losses/sft": 2.465557098388672,
|
| 524 |
+
"losses/total": 0.11473169922828674,
|
| 525 |
+
"ref_logps/chosen": -144.18157958984375,
|
| 526 |
+
"ref_logps/rejected": -164.2210693359375,
|
| 527 |
+
"rewards/accuracies": 0.890625,
|
| 528 |
+
"rewards/chosen": -2.939948081970215,
|
| 529 |
+
"rewards/margins": 6.72892951965332,
|
| 530 |
+
"rewards/rejected": -9.668878555297852,
|
| 531 |
+
"step": 29
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 0.055865921787709494,
|
| 535 |
+
"grad_norm": 15.30510425567627,
|
| 536 |
+
"learning_rate": 2.7777777777777783e-06,
|
| 537 |
+
"logps/chosen": -170.92425537109375,
|
| 538 |
+
"logps/rejected": -277.4226379394531,
|
| 539 |
+
"loss": 0.1033,
|
| 540 |
+
"losses/dpo": 0.13019727170467377,
|
| 541 |
+
"losses/sft": 2.4969606399536133,
|
| 542 |
+
"losses/total": 0.13019727170467377,
|
| 543 |
+
"ref_logps/chosen": -142.30502319335938,
|
| 544 |
+
"ref_logps/rejected": -169.90554809570312,
|
| 545 |
+
"rewards/accuracies": 0.9375,
|
| 546 |
+
"rewards/chosen": -2.8619208335876465,
|
| 547 |
+
"rewards/margins": 7.889788627624512,
|
| 548 |
+
"rewards/rejected": -10.751708984375,
|
| 549 |
+
"step": 30
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.05772811918063315,
|
| 553 |
+
"grad_norm": 6.729362487792969,
|
| 554 |
+
"learning_rate": 2.8703703703703706e-06,
|
| 555 |
+
"logps/chosen": -177.1944122314453,
|
| 556 |
+
"logps/rejected": -277.68206787109375,
|
| 557 |
+
"loss": 0.0363,
|
| 558 |
+
"losses/dpo": 0.03610475733876228,
|
| 559 |
+
"losses/sft": 2.655313491821289,
|
| 560 |
+
"losses/total": 0.03610475733876228,
|
| 561 |
+
"ref_logps/chosen": -140.73678588867188,
|
| 562 |
+
"ref_logps/rejected": -164.96824645996094,
|
| 563 |
+
"rewards/accuracies": 1.0,
|
| 564 |
+
"rewards/chosen": -3.6457643508911133,
|
| 565 |
+
"rewards/margins": 7.625619888305664,
|
| 566 |
+
"rewards/rejected": -11.271383285522461,
|
| 567 |
+
"step": 31
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 0.0595903165735568,
|
| 571 |
+
"grad_norm": 15.916095733642578,
|
| 572 |
+
"learning_rate": 2.962962962962963e-06,
|
| 573 |
+
"logps/chosen": -168.07888793945312,
|
| 574 |
+
"logps/rejected": -266.97467041015625,
|
| 575 |
+
"loss": 0.1384,
|
| 576 |
+
"losses/dpo": 0.003958229906857014,
|
| 577 |
+
"losses/sft": 3.21946120262146,
|
| 578 |
+
"losses/total": 0.003958229906857014,
|
| 579 |
+
"ref_logps/chosen": -135.16700744628906,
|
| 580 |
+
"ref_logps/rejected": -166.81027221679688,
|
| 581 |
+
"rewards/accuracies": 0.9375,
|
| 582 |
+
"rewards/chosen": -3.2911877632141113,
|
| 583 |
+
"rewards/margins": 6.725253582000732,
|
| 584 |
+
"rewards/rejected": -10.016440391540527,
|
| 585 |
+
"step": 32
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 0.061452513966480445,
|
| 589 |
+
"grad_norm": 13.613255500793457,
|
| 590 |
+
"learning_rate": 3.055555555555556e-06,
|
| 591 |
+
"logps/chosen": -163.11419677734375,
|
| 592 |
+
"logps/rejected": -267.9139099121094,
|
| 593 |
+
"loss": 0.1,
|
| 594 |
+
"losses/dpo": 0.0983101949095726,
|
| 595 |
+
"losses/sft": 2.5492136478424072,
|
| 596 |
+
"losses/total": 0.0983101949095726,
|
| 597 |
+
"ref_logps/chosen": -140.14205932617188,
|
| 598 |
+
"ref_logps/rejected": -160.53872680664062,
|
| 599 |
+
"rewards/accuracies": 0.9375,
|
| 600 |
+
"rewards/chosen": -2.297213554382324,
|
| 601 |
+
"rewards/margins": 8.44030475616455,
|
| 602 |
+
"rewards/rejected": -10.737519264221191,
|
| 603 |
+
"step": 33
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 0.0633147113594041,
|
| 607 |
+
"grad_norm": 15.18605899810791,
|
| 608 |
+
"learning_rate": 3.1481481481481483e-06,
|
| 609 |
+
"logps/chosen": -167.43917846679688,
|
| 610 |
+
"logps/rejected": -285.1209411621094,
|
| 611 |
+
"loss": 0.1385,
|
| 612 |
+
"losses/dpo": 0.010567452758550644,
|
| 613 |
+
"losses/sft": 2.848440170288086,
|
| 614 |
+
"losses/total": 0.010567452758550644,
|
| 615 |
+
"ref_logps/chosen": -137.24496459960938,
|
| 616 |
+
"ref_logps/rejected": -172.29844665527344,
|
| 617 |
+
"rewards/accuracies": 0.96875,
|
| 618 |
+
"rewards/chosen": -3.019423723220825,
|
| 619 |
+
"rewards/margins": 8.262825965881348,
|
| 620 |
+
"rewards/rejected": -11.282249450683594,
|
| 621 |
+
"step": 34
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 0.06517690875232775,
|
| 625 |
+
"grad_norm": 14.179464340209961,
|
| 626 |
+
"learning_rate": 3.240740740740741e-06,
|
| 627 |
+
"logps/chosen": -163.87579345703125,
|
| 628 |
+
"logps/rejected": -273.77777099609375,
|
| 629 |
+
"loss": 0.1047,
|
| 630 |
+
"losses/dpo": 0.010419070720672607,
|
| 631 |
+
"losses/sft": 2.3946168422698975,
|
| 632 |
+
"losses/total": 0.010419070720672607,
|
| 633 |
+
"ref_logps/chosen": -144.74697875976562,
|
| 634 |
+
"ref_logps/rejected": -167.2154541015625,
|
| 635 |
+
"rewards/accuracies": 0.9375,
|
| 636 |
+
"rewards/chosen": -1.9128817319869995,
|
| 637 |
+
"rewards/margins": 8.743349075317383,
|
| 638 |
+
"rewards/rejected": -10.656229972839355,
|
| 639 |
+
"step": 35
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.0670391061452514,
|
| 643 |
+
"grad_norm": 7.81901741027832,
|
| 644 |
+
"learning_rate": 3.3333333333333333e-06,
|
| 645 |
+
"logps/chosen": -170.99649047851562,
|
| 646 |
+
"logps/rejected": -260.6686096191406,
|
| 647 |
+
"loss": 0.0408,
|
| 648 |
+
"losses/dpo": 0.03330932930111885,
|
| 649 |
+
"losses/sft": 2.9487311840057373,
|
| 650 |
+
"losses/total": 0.03330932930111885,
|
| 651 |
+
"ref_logps/chosen": -150.3196563720703,
|
| 652 |
+
"ref_logps/rejected": -156.96363830566406,
|
| 653 |
+
"rewards/accuracies": 0.984375,
|
| 654 |
+
"rewards/chosen": -2.0676848888397217,
|
| 655 |
+
"rewards/margins": 8.302813529968262,
|
| 656 |
+
"rewards/rejected": -10.370498657226562,
|
| 657 |
+
"step": 36
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 0.06890130353817504,
|
| 661 |
+
"grad_norm": 14.016197204589844,
|
| 662 |
+
"learning_rate": 3.4259259259259265e-06,
|
| 663 |
+
"logps/chosen": -161.38865661621094,
|
| 664 |
+
"logps/rejected": -281.09271240234375,
|
| 665 |
+
"loss": 0.1047,
|
| 666 |
+
"losses/dpo": 0.5629028677940369,
|
| 667 |
+
"losses/sft": 2.4373741149902344,
|
| 668 |
+
"losses/total": 0.5629028677940369,
|
| 669 |
+
"ref_logps/chosen": -143.14517211914062,
|
| 670 |
+
"ref_logps/rejected": -172.78611755371094,
|
| 671 |
+
"rewards/accuracies": 0.953125,
|
| 672 |
+
"rewards/chosen": -1.8243476152420044,
|
| 673 |
+
"rewards/margins": 9.006311416625977,
|
| 674 |
+
"rewards/rejected": -10.830659866333008,
|
| 675 |
+
"step": 37
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.07076350093109869,
|
| 679 |
+
"grad_norm": 15.18514347076416,
|
| 680 |
+
"learning_rate": 3.5185185185185187e-06,
|
| 681 |
+
"logps/chosen": -160.60586547851562,
|
| 682 |
+
"logps/rejected": -263.134765625,
|
| 683 |
+
"loss": 0.1468,
|
| 684 |
+
"losses/dpo": 0.018482662737369537,
|
| 685 |
+
"losses/sft": 2.2324092388153076,
|
| 686 |
+
"losses/total": 0.018482662737369537,
|
| 687 |
+
"ref_logps/chosen": -141.36175537109375,
|
| 688 |
+
"ref_logps/rejected": -164.61480712890625,
|
| 689 |
+
"rewards/accuracies": 0.953125,
|
| 690 |
+
"rewards/chosen": -1.9244120121002197,
|
| 691 |
+
"rewards/margins": 7.92758321762085,
|
| 692 |
+
"rewards/rejected": -9.851995468139648,
|
| 693 |
+
"step": 38
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 0.07262569832402235,
|
| 697 |
+
"grad_norm": 8.755450248718262,
|
| 698 |
+
"learning_rate": 3.6111111111111115e-06,
|
| 699 |
+
"logps/chosen": -170.1685791015625,
|
| 700 |
+
"logps/rejected": -288.41119384765625,
|
| 701 |
+
"loss": 0.0421,
|
| 702 |
+
"losses/dpo": 9.010350913740695e-05,
|
| 703 |
+
"losses/sft": 2.3953020572662354,
|
| 704 |
+
"losses/total": 9.010350913740695e-05,
|
| 705 |
+
"ref_logps/chosen": -152.65139770507812,
|
| 706 |
+
"ref_logps/rejected": -172.203857421875,
|
| 707 |
+
"rewards/accuracies": 0.984375,
|
| 708 |
+
"rewards/chosen": -1.751717448234558,
|
| 709 |
+
"rewards/margins": 9.86901569366455,
|
| 710 |
+
"rewards/rejected": -11.620733261108398,
|
| 711 |
+
"step": 39
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 0.074487895716946,
|
| 715 |
+
"grad_norm": 2.3320696353912354,
|
| 716 |
+
"learning_rate": 3.7037037037037037e-06,
|
| 717 |
+
"logps/chosen": -164.16981506347656,
|
| 718 |
+
"logps/rejected": -263.49163818359375,
|
| 719 |
+
"loss": 0.0103,
|
| 720 |
+
"losses/dpo": 0.013380440883338451,
|
| 721 |
+
"losses/sft": 2.408010482788086,
|
| 722 |
+
"losses/total": 0.013380440883338451,
|
| 723 |
+
"ref_logps/chosen": -153.0699462890625,
|
| 724 |
+
"ref_logps/rejected": -166.4036407470703,
|
| 725 |
+
"rewards/accuracies": 1.0,
|
| 726 |
+
"rewards/chosen": -1.1099885702133179,
|
| 727 |
+
"rewards/margins": 8.598810195922852,
|
| 728 |
+
"rewards/rejected": -9.708799362182617,
|
| 729 |
+
"step": 40
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 0.07635009310986965,
|
| 733 |
+
"grad_norm": 13.349935531616211,
|
| 734 |
+
"learning_rate": 3.796296296296297e-06,
|
| 735 |
+
"logps/chosen": -162.9290771484375,
|
| 736 |
+
"logps/rejected": -275.8412780761719,
|
| 737 |
+
"loss": 0.0746,
|
| 738 |
+
"losses/dpo": 0.0015751949977129698,
|
| 739 |
+
"losses/sft": 2.3437771797180176,
|
| 740 |
+
"losses/total": 0.0015751949977129698,
|
| 741 |
+
"ref_logps/chosen": -151.49713134765625,
|
| 742 |
+
"ref_logps/rejected": -168.54776000976562,
|
| 743 |
+
"rewards/accuracies": 0.953125,
|
| 744 |
+
"rewards/chosen": -1.1431937217712402,
|
| 745 |
+
"rewards/margins": 9.586158752441406,
|
| 746 |
+
"rewards/rejected": -10.729352951049805,
|
| 747 |
+
"step": 41
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 0.0782122905027933,
|
| 751 |
+
"grad_norm": 7.708398818969727,
|
| 752 |
+
"learning_rate": 3.88888888888889e-06,
|
| 753 |
+
"logps/chosen": -170.53414916992188,
|
| 754 |
+
"logps/rejected": -267.343017578125,
|
| 755 |
+
"loss": 0.0415,
|
| 756 |
+
"losses/dpo": 0.16293640434741974,
|
| 757 |
+
"losses/sft": 2.4445619583129883,
|
| 758 |
+
"losses/total": 0.16293640434741974,
|
| 759 |
+
"ref_logps/chosen": -158.01504516601562,
|
| 760 |
+
"ref_logps/rejected": -162.6872100830078,
|
| 761 |
+
"rewards/accuracies": 0.984375,
|
| 762 |
+
"rewards/chosen": -1.2519094944000244,
|
| 763 |
+
"rewards/margins": 9.213672637939453,
|
| 764 |
+
"rewards/rejected": -10.465582847595215,
|
| 765 |
+
"step": 42
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 0.08007448789571694,
|
| 769 |
+
"grad_norm": 8.482429504394531,
|
| 770 |
+
"learning_rate": 3.9814814814814814e-06,
|
| 771 |
+
"logps/chosen": -152.68186950683594,
|
| 772 |
+
"logps/rejected": -256.3250427246094,
|
| 773 |
+
"loss": 0.0489,
|
| 774 |
+
"losses/dpo": 0.005869926419109106,
|
| 775 |
+
"losses/sft": 2.2688705921173096,
|
| 776 |
+
"losses/total": 0.005869926419109106,
|
| 777 |
+
"ref_logps/chosen": -143.47671508789062,
|
| 778 |
+
"ref_logps/rejected": -158.04229736328125,
|
| 779 |
+
"rewards/accuracies": 0.96875,
|
| 780 |
+
"rewards/chosen": -0.9205170273780823,
|
| 781 |
+
"rewards/margins": 8.907758712768555,
|
| 782 |
+
"rewards/rejected": -9.828275680541992,
|
| 783 |
+
"step": 43
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 0.08193668528864059,
|
| 787 |
+
"grad_norm": 4.836800575256348,
|
| 788 |
+
"learning_rate": 4.074074074074074e-06,
|
| 789 |
+
"logps/chosen": -161.54238891601562,
|
| 790 |
+
"logps/rejected": -249.66888427734375,
|
| 791 |
+
"loss": 0.0293,
|
| 792 |
+
"losses/dpo": 0.007885000668466091,
|
| 793 |
+
"losses/sft": 2.5034093856811523,
|
| 794 |
+
"losses/total": 0.007885000668466091,
|
| 795 |
+
"ref_logps/chosen": -150.9593963623047,
|
| 796 |
+
"ref_logps/rejected": -160.7205810546875,
|
| 797 |
+
"rewards/accuracies": 0.984375,
|
| 798 |
+
"rewards/chosen": -1.05829918384552,
|
| 799 |
+
"rewards/margins": 7.8365325927734375,
|
| 800 |
+
"rewards/rejected": -8.894831657409668,
|
| 801 |
+
"step": 44
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 0.08379888268156424,
|
| 805 |
+
"grad_norm": 7.306879997253418,
|
| 806 |
+
"learning_rate": 4.166666666666667e-06,
|
| 807 |
+
"logps/chosen": -168.57354736328125,
|
| 808 |
+
"logps/rejected": -273.81622314453125,
|
| 809 |
+
"loss": 0.0409,
|
| 810 |
+
"losses/dpo": 0.044236376881599426,
|
| 811 |
+
"losses/sft": 2.285489559173584,
|
| 812 |
+
"losses/total": 0.044236376881599426,
|
| 813 |
+
"ref_logps/chosen": -153.68539428710938,
|
| 814 |
+
"ref_logps/rejected": -173.37750244140625,
|
| 815 |
+
"rewards/accuracies": 1.0,
|
| 816 |
+
"rewards/chosen": -1.4888172149658203,
|
| 817 |
+
"rewards/margins": 8.555052757263184,
|
| 818 |
+
"rewards/rejected": -10.043869972229004,
|
| 819 |
+
"step": 45
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 0.0856610800744879,
|
| 823 |
+
"grad_norm": 8.716976165771484,
|
| 824 |
+
"learning_rate": 4.2592592592592596e-06,
|
| 825 |
+
"logps/chosen": -157.63235473632812,
|
| 826 |
+
"logps/rejected": -256.28546142578125,
|
| 827 |
+
"loss": 0.0498,
|
| 828 |
+
"losses/dpo": 0.047989651560783386,
|
| 829 |
+
"losses/sft": 2.594020366668701,
|
| 830 |
+
"losses/total": 0.047989651560783386,
|
| 831 |
+
"ref_logps/chosen": -149.37562561035156,
|
| 832 |
+
"ref_logps/rejected": -163.25059509277344,
|
| 833 |
+
"rewards/accuracies": 0.96875,
|
| 834 |
+
"rewards/chosen": -0.8256728053092957,
|
| 835 |
+
"rewards/margins": 8.477815628051758,
|
| 836 |
+
"rewards/rejected": -9.303489685058594,
|
| 837 |
+
"step": 46
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 0.08752327746741155,
|
| 841 |
+
"grad_norm": 14.776224136352539,
|
| 842 |
+
"learning_rate": 4.351851851851852e-06,
|
| 843 |
+
"logps/chosen": -159.35926818847656,
|
| 844 |
+
"logps/rejected": -254.82000732421875,
|
| 845 |
+
"loss": 0.0947,
|
| 846 |
+
"losses/dpo": 0.0015063448809087276,
|
| 847 |
+
"losses/sft": 2.186452627182007,
|
| 848 |
+
"losses/total": 0.0015063448809087276,
|
| 849 |
+
"ref_logps/chosen": -153.80572509765625,
|
| 850 |
+
"ref_logps/rejected": -160.63690185546875,
|
| 851 |
+
"rewards/accuracies": 0.953125,
|
| 852 |
+
"rewards/chosen": -0.555354654788971,
|
| 853 |
+
"rewards/margins": 8.862955093383789,
|
| 854 |
+
"rewards/rejected": -9.41831111907959,
|
| 855 |
+
"step": 47
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 0.0893854748603352,
|
| 859 |
+
"grad_norm": 6.18978214263916,
|
| 860 |
+
"learning_rate": 4.444444444444444e-06,
|
| 861 |
+
"logps/chosen": -146.09950256347656,
|
| 862 |
+
"logps/rejected": -269.2020568847656,
|
| 863 |
+
"loss": 0.0442,
|
| 864 |
+
"losses/dpo": 0.010637683793902397,
|
| 865 |
+
"losses/sft": 2.3249499797821045,
|
| 866 |
+
"losses/total": 0.010637683793902397,
|
| 867 |
+
"ref_logps/chosen": -140.32867431640625,
|
| 868 |
+
"ref_logps/rejected": -161.42984008789062,
|
| 869 |
+
"rewards/accuracies": 0.984375,
|
| 870 |
+
"rewards/chosen": -0.5770829319953918,
|
| 871 |
+
"rewards/margins": 10.200138092041016,
|
| 872 |
+
"rewards/rejected": -10.7772216796875,
|
| 873 |
+
"step": 48
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 0.09124767225325885,
|
| 877 |
+
"grad_norm": 8.178156852722168,
|
| 878 |
+
"learning_rate": 4.537037037037038e-06,
|
| 879 |
+
"logps/chosen": -150.715087890625,
|
| 880 |
+
"logps/rejected": -288.4056396484375,
|
| 881 |
+
"loss": 0.0446,
|
| 882 |
+
"losses/dpo": 0.2217281013727188,
|
| 883 |
+
"losses/sft": 2.4317214488983154,
|
| 884 |
+
"losses/total": 0.2217281013727188,
|
| 885 |
+
"ref_logps/chosen": -143.90017700195312,
|
| 886 |
+
"ref_logps/rejected": -177.4407958984375,
|
| 887 |
+
"rewards/accuracies": 0.984375,
|
| 888 |
+
"rewards/chosen": -0.6814901828765869,
|
| 889 |
+
"rewards/margins": 10.414993286132812,
|
| 890 |
+
"rewards/rejected": -11.09648323059082,
|
| 891 |
+
"step": 49
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 0.0931098696461825,
|
| 895 |
+
"grad_norm": 11.958553314208984,
|
| 896 |
+
"learning_rate": 4.62962962962963e-06,
|
| 897 |
+
"logps/chosen": -156.93243408203125,
|
| 898 |
+
"logps/rejected": -276.21051025390625,
|
| 899 |
+
"loss": 0.0786,
|
| 900 |
+
"losses/dpo": 0.020680470392107964,
|
| 901 |
+
"losses/sft": 2.4808478355407715,
|
| 902 |
+
"losses/total": 0.020680470392107964,
|
| 903 |
+
"ref_logps/chosen": -146.65036010742188,
|
| 904 |
+
"ref_logps/rejected": -166.49574279785156,
|
| 905 |
+
"rewards/accuracies": 0.96875,
|
| 906 |
+
"rewards/chosen": -1.0282070636749268,
|
| 907 |
+
"rewards/margins": 9.943269729614258,
|
| 908 |
+
"rewards/rejected": -10.971476554870605,
|
| 909 |
+
"step": 50
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 0.09497206703910614,
|
| 913 |
+
"grad_norm": 2.3756279945373535,
|
| 914 |
+
"learning_rate": 4.722222222222222e-06,
|
| 915 |
+
"logps/chosen": -146.2957763671875,
|
| 916 |
+
"logps/rejected": -290.5218505859375,
|
| 917 |
+
"loss": 0.0115,
|
| 918 |
+
"losses/dpo": 0.021228570491075516,
|
| 919 |
+
"losses/sft": 2.0093941688537598,
|
| 920 |
+
"losses/total": 0.021228570491075516,
|
| 921 |
+
"ref_logps/chosen": -144.70184326171875,
|
| 922 |
+
"ref_logps/rejected": -169.62815856933594,
|
| 923 |
+
"rewards/accuracies": 1.0,
|
| 924 |
+
"rewards/chosen": -0.15939302742481232,
|
| 925 |
+
"rewards/margins": 11.929975509643555,
|
| 926 |
+
"rewards/rejected": -12.08936882019043,
|
| 927 |
+
"step": 51
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 0.09683426443202979,
|
| 931 |
+
"grad_norm": 4.650004863739014,
|
| 932 |
+
"learning_rate": 4.814814814814815e-06,
|
| 933 |
+
"logps/chosen": -151.42132568359375,
|
| 934 |
+
"logps/rejected": -283.79669189453125,
|
| 935 |
+
"loss": 0.0183,
|
| 936 |
+
"losses/dpo": 0.005594337359070778,
|
| 937 |
+
"losses/sft": 2.4590542316436768,
|
| 938 |
+
"losses/total": 0.005594337359070778,
|
| 939 |
+
"ref_logps/chosen": -148.097412109375,
|
| 940 |
+
"ref_logps/rejected": -164.08462524414062,
|
| 941 |
+
"rewards/accuracies": 1.0,
|
| 942 |
+
"rewards/chosen": -0.33239230513572693,
|
| 943 |
+
"rewards/margins": 11.638812065124512,
|
| 944 |
+
"rewards/rejected": -11.97120475769043,
|
| 945 |
+
"step": 52
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 0.09869646182495345,
|
| 949 |
+
"grad_norm": 7.309722423553467,
|
| 950 |
+
"learning_rate": 4.907407407407408e-06,
|
| 951 |
+
"logps/chosen": -163.71875,
|
| 952 |
+
"logps/rejected": -280.90655517578125,
|
| 953 |
+
"loss": 0.0367,
|
| 954 |
+
"losses/dpo": 0.05281204357743263,
|
| 955 |
+
"losses/sft": 2.374516725540161,
|
| 956 |
+
"losses/total": 0.05281204357743263,
|
| 957 |
+
"ref_logps/chosen": -151.98492431640625,
|
| 958 |
+
"ref_logps/rejected": -164.55319213867188,
|
| 959 |
+
"rewards/accuracies": 1.0,
|
| 960 |
+
"rewards/chosen": -1.173383116722107,
|
| 961 |
+
"rewards/margins": 10.461955070495605,
|
| 962 |
+
"rewards/rejected": -11.635337829589844,
|
| 963 |
+
"step": 53
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 0.1005586592178771,
|
| 967 |
+
"grad_norm": 6.878481388092041,
|
| 968 |
+
"learning_rate": 5e-06,
|
| 969 |
+
"logps/chosen": -158.5335693359375,
|
| 970 |
+
"logps/rejected": -270.8926086425781,
|
| 971 |
+
"loss": 0.0462,
|
| 972 |
+
"losses/dpo": 0.008242600597441196,
|
| 973 |
+
"losses/sft": 2.6928701400756836,
|
| 974 |
+
"losses/total": 0.008242600597441196,
|
| 975 |
+
"ref_logps/chosen": -154.32550048828125,
|
| 976 |
+
"ref_logps/rejected": -162.00128173828125,
|
| 977 |
+
"rewards/accuracies": 0.96875,
|
| 978 |
+
"rewards/chosen": -0.4208071529865265,
|
| 979 |
+
"rewards/margins": 10.468324661254883,
|
| 980 |
+
"rewards/rejected": -10.889131546020508,
|
| 981 |
+
"step": 54
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 0.10242085661080075,
|
| 985 |
+
"grad_norm": 5.541274070739746,
|
| 986 |
+
"learning_rate": 4.989648033126294e-06,
|
| 987 |
+
"logps/chosen": -150.91574096679688,
|
| 988 |
+
"logps/rejected": -279.86248779296875,
|
| 989 |
+
"loss": 0.0329,
|
| 990 |
+
"losses/dpo": 0.0699252262711525,
|
| 991 |
+
"losses/sft": 2.2539589405059814,
|
| 992 |
+
"losses/total": 0.0699252262711525,
|
| 993 |
+
"ref_logps/chosen": -147.29766845703125,
|
| 994 |
+
"ref_logps/rejected": -170.07183837890625,
|
| 995 |
+
"rewards/accuracies": 0.984375,
|
| 996 |
+
"rewards/chosen": -0.36180704832077026,
|
| 997 |
+
"rewards/margins": 10.61725902557373,
|
| 998 |
+
"rewards/rejected": -10.97906494140625,
|
| 999 |
+
"step": 55
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 0.1042830540037244,
|
| 1003 |
+
"grad_norm": 6.928717613220215,
|
| 1004 |
+
"learning_rate": 4.9792960662525884e-06,
|
| 1005 |
+
"logps/chosen": -157.90420532226562,
|
| 1006 |
+
"logps/rejected": -274.0709228515625,
|
| 1007 |
+
"loss": 0.0401,
|
| 1008 |
+
"losses/dpo": 0.0023821014910936356,
|
| 1009 |
+
"losses/sft": 2.440980911254883,
|
| 1010 |
+
"losses/total": 0.0023821014910936356,
|
| 1011 |
+
"ref_logps/chosen": -145.0316925048828,
|
| 1012 |
+
"ref_logps/rejected": -164.71649169921875,
|
| 1013 |
+
"rewards/accuracies": 0.984375,
|
| 1014 |
+
"rewards/chosen": -1.2872496843338013,
|
| 1015 |
+
"rewards/margins": 9.648194313049316,
|
| 1016 |
+
"rewards/rejected": -10.935443878173828,
|
| 1017 |
+
"step": 56
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 0.10614525139664804,
|
| 1021 |
+
"grad_norm": 12.181221961975098,
|
| 1022 |
+
"learning_rate": 4.968944099378882e-06,
|
| 1023 |
+
"logps/chosen": -164.98733520507812,
|
| 1024 |
+
"logps/rejected": -292.568115234375,
|
| 1025 |
+
"loss": 0.0608,
|
| 1026 |
+
"losses/dpo": 0.2203754484653473,
|
| 1027 |
+
"losses/sft": 2.4729645252227783,
|
| 1028 |
+
"losses/total": 0.2203754484653473,
|
| 1029 |
+
"ref_logps/chosen": -145.78073120117188,
|
| 1030 |
+
"ref_logps/rejected": -170.54788208007812,
|
| 1031 |
+
"rewards/accuracies": 0.984375,
|
| 1032 |
+
"rewards/chosen": -1.9206593036651611,
|
| 1033 |
+
"rewards/margins": 10.28136157989502,
|
| 1034 |
+
"rewards/rejected": -12.202020645141602,
|
| 1035 |
+
"step": 57
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 0.10800744878957169,
|
| 1039 |
+
"grad_norm": 3.31730580329895,
|
| 1040 |
+
"learning_rate": 4.9585921325051765e-06,
|
| 1041 |
+
"logps/chosen": -144.76385498046875,
|
| 1042 |
+
"logps/rejected": -279.8059997558594,
|
| 1043 |
+
"loss": 0.0189,
|
| 1044 |
+
"losses/dpo": 0.023249687626957893,
|
| 1045 |
+
"losses/sft": 2.241105079650879,
|
| 1046 |
+
"losses/total": 0.023249687626957893,
|
| 1047 |
+
"ref_logps/chosen": -135.7710723876953,
|
| 1048 |
+
"ref_logps/rejected": -163.89041137695312,
|
| 1049 |
+
"rewards/accuracies": 1.0,
|
| 1050 |
+
"rewards/chosen": -0.8992785215377808,
|
| 1051 |
+
"rewards/margins": 10.692280769348145,
|
| 1052 |
+
"rewards/rejected": -11.591558456420898,
|
| 1053 |
+
"step": 58
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 0.10986964618249534,
|
| 1057 |
+
"grad_norm": 5.754706382751465,
|
| 1058 |
+
"learning_rate": 4.94824016563147e-06,
|
| 1059 |
+
"logps/chosen": -159.64266967773438,
|
| 1060 |
+
"logps/rejected": -272.6561584472656,
|
| 1061 |
+
"loss": 0.0327,
|
| 1062 |
+
"losses/dpo": 0.0525113008916378,
|
| 1063 |
+
"losses/sft": 2.5197577476501465,
|
| 1064 |
+
"losses/total": 0.0525113008916378,
|
| 1065 |
+
"ref_logps/chosen": -142.01556396484375,
|
| 1066 |
+
"ref_logps/rejected": -161.62533569335938,
|
| 1067 |
+
"rewards/accuracies": 1.0,
|
| 1068 |
+
"rewards/chosen": -1.7627112865447998,
|
| 1069 |
+
"rewards/margins": 9.340368270874023,
|
| 1070 |
+
"rewards/rejected": -11.103079795837402,
|
| 1071 |
+
"step": 59
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 0.11173184357541899,
|
| 1075 |
+
"grad_norm": 2.9009838104248047,
|
| 1076 |
+
"learning_rate": 4.9378881987577645e-06,
|
| 1077 |
+
"logps/chosen": -153.46998596191406,
|
| 1078 |
+
"logps/rejected": -300.505615234375,
|
| 1079 |
+
"loss": 0.0109,
|
| 1080 |
+
"losses/dpo": 0.05081873759627342,
|
| 1081 |
+
"losses/sft": 2.3817033767700195,
|
| 1082 |
+
"losses/total": 0.05081873759627342,
|
| 1083 |
+
"ref_logps/chosen": -137.58828735351562,
|
| 1084 |
+
"ref_logps/rejected": -164.65621948242188,
|
| 1085 |
+
"rewards/accuracies": 1.0,
|
| 1086 |
+
"rewards/chosen": -1.5881686210632324,
|
| 1087 |
+
"rewards/margins": 11.996772766113281,
|
| 1088 |
+
"rewards/rejected": -13.584940910339355,
|
| 1089 |
+
"step": 60
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 0.11359404096834265,
|
| 1093 |
+
"grad_norm": 10.7985258102417,
|
| 1094 |
+
"learning_rate": 4.927536231884059e-06,
|
| 1095 |
+
"logps/chosen": -161.22463989257812,
|
| 1096 |
+
"logps/rejected": -312.06597900390625,
|
| 1097 |
+
"loss": 0.0594,
|
| 1098 |
+
"losses/dpo": 0.02082718536257744,
|
| 1099 |
+
"losses/sft": 2.446563720703125,
|
| 1100 |
+
"losses/total": 0.02082718536257744,
|
| 1101 |
+
"ref_logps/chosen": -150.21656799316406,
|
| 1102 |
+
"ref_logps/rejected": -171.88088989257812,
|
| 1103 |
+
"rewards/accuracies": 0.984375,
|
| 1104 |
+
"rewards/chosen": -1.1008076667785645,
|
| 1105 |
+
"rewards/margins": 12.917702674865723,
|
| 1106 |
+
"rewards/rejected": -14.018510818481445,
|
| 1107 |
+
"step": 61
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 0.1154562383612663,
|
| 1111 |
+
"grad_norm": 16.700149536132812,
|
| 1112 |
+
"learning_rate": 4.9171842650103525e-06,
|
| 1113 |
+
"logps/chosen": -157.88522338867188,
|
| 1114 |
+
"logps/rejected": -296.93817138671875,
|
| 1115 |
+
"loss": 0.0929,
|
| 1116 |
+
"losses/dpo": 0.0005755929159931839,
|
| 1117 |
+
"losses/sft": 2.514739751815796,
|
| 1118 |
+
"losses/total": 0.0005755929159931839,
|
| 1119 |
+
"ref_logps/chosen": -140.2955780029297,
|
| 1120 |
+
"ref_logps/rejected": -165.69281005859375,
|
| 1121 |
+
"rewards/accuracies": 0.984375,
|
| 1122 |
+
"rewards/chosen": -1.7589635848999023,
|
| 1123 |
+
"rewards/margins": 11.365575790405273,
|
| 1124 |
+
"rewards/rejected": -13.124539375305176,
|
| 1125 |
+
"step": 62
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 0.11731843575418995,
|
| 1129 |
+
"grad_norm": 10.798019409179688,
|
| 1130 |
+
"learning_rate": 4.906832298136646e-06,
|
| 1131 |
+
"logps/chosen": -160.2606201171875,
|
| 1132 |
+
"logps/rejected": -316.3315734863281,
|
| 1133 |
+
"loss": 0.0485,
|
| 1134 |
+
"losses/dpo": 0.0018289771396666765,
|
| 1135 |
+
"losses/sft": 3.0345890522003174,
|
| 1136 |
+
"losses/total": 0.0018289771396666765,
|
| 1137 |
+
"ref_logps/chosen": -144.05250549316406,
|
| 1138 |
+
"ref_logps/rejected": -170.66879272460938,
|
| 1139 |
+
"rewards/accuracies": 0.984375,
|
| 1140 |
+
"rewards/chosen": -1.6208107471466064,
|
| 1141 |
+
"rewards/margins": 12.945466041564941,
|
| 1142 |
+
"rewards/rejected": -14.566276550292969,
|
| 1143 |
+
"step": 63
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 0.1191806331471136,
|
| 1147 |
+
"grad_norm": 8.646073341369629,
|
| 1148 |
+
"learning_rate": 4.896480331262941e-06,
|
| 1149 |
+
"logps/chosen": -163.75540161132812,
|
| 1150 |
+
"logps/rejected": -315.20458984375,
|
| 1151 |
+
"loss": 0.0576,
|
| 1152 |
+
"losses/dpo": 0.014824606478214264,
|
| 1153 |
+
"losses/sft": 3.261763572692871,
|
| 1154 |
+
"losses/total": 0.014824606478214264,
|
| 1155 |
+
"ref_logps/chosen": -149.48907470703125,
|
| 1156 |
+
"ref_logps/rejected": -173.37435913085938,
|
| 1157 |
+
"rewards/accuracies": 0.984375,
|
| 1158 |
+
"rewards/chosen": -1.4266325235366821,
|
| 1159 |
+
"rewards/margins": 12.756391525268555,
|
| 1160 |
+
"rewards/rejected": -14.183023452758789,
|
| 1161 |
+
"step": 64
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 0.12104283054003724,
|
| 1165 |
+
"grad_norm": 5.649742603302002,
|
| 1166 |
+
"learning_rate": 4.886128364389234e-06,
|
| 1167 |
+
"logps/chosen": -174.45712280273438,
|
| 1168 |
+
"logps/rejected": -324.653076171875,
|
| 1169 |
+
"loss": 0.018,
|
| 1170 |
+
"losses/dpo": 0.03694448992609978,
|
| 1171 |
+
"losses/sft": 2.400181293487549,
|
| 1172 |
+
"losses/total": 0.03694448992609978,
|
| 1173 |
+
"ref_logps/chosen": -155.43272399902344,
|
| 1174 |
+
"ref_logps/rejected": -175.6329345703125,
|
| 1175 |
+
"rewards/accuracies": 1.0,
|
| 1176 |
+
"rewards/chosen": -1.902438759803772,
|
| 1177 |
+
"rewards/margins": 12.999577522277832,
|
| 1178 |
+
"rewards/rejected": -14.902015686035156,
|
| 1179 |
+
"step": 65
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 0.12290502793296089,
|
| 1183 |
+
"grad_norm": 4.2772345542907715,
|
| 1184 |
+
"learning_rate": 4.875776397515528e-06,
|
| 1185 |
+
"logps/chosen": -160.1391143798828,
|
| 1186 |
+
"logps/rejected": -298.70098876953125,
|
| 1187 |
+
"loss": 0.0198,
|
| 1188 |
+
"losses/dpo": 0.009300258941948414,
|
| 1189 |
+
"losses/sft": 2.3741018772125244,
|
| 1190 |
+
"losses/total": 0.009300258941948414,
|
| 1191 |
+
"ref_logps/chosen": -143.62374877929688,
|
| 1192 |
+
"ref_logps/rejected": -162.72705078125,
|
| 1193 |
+
"rewards/accuracies": 1.0,
|
| 1194 |
+
"rewards/chosen": -1.6515378952026367,
|
| 1195 |
+
"rewards/margins": 11.945855140686035,
|
| 1196 |
+
"rewards/rejected": -13.597392082214355,
|
| 1197 |
+
"step": 66
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 0.12476722532588454,
|
| 1201 |
+
"grad_norm": 12.112936019897461,
|
| 1202 |
+
"learning_rate": 4.865424430641822e-06,
|
| 1203 |
+
"logps/chosen": -171.04205322265625,
|
| 1204 |
+
"logps/rejected": -303.92144775390625,
|
| 1205 |
+
"loss": 0.0626,
|
| 1206 |
+
"losses/dpo": 0.36290252208709717,
|
| 1207 |
+
"losses/sft": 2.4964911937713623,
|
| 1208 |
+
"losses/total": 0.36290252208709717,
|
| 1209 |
+
"ref_logps/chosen": -152.29478454589844,
|
| 1210 |
+
"ref_logps/rejected": -161.23959350585938,
|
| 1211 |
+
"rewards/accuracies": 0.984375,
|
| 1212 |
+
"rewards/chosen": -1.8747248649597168,
|
| 1213 |
+
"rewards/margins": 12.393461227416992,
|
| 1214 |
+
"rewards/rejected": -14.268186569213867,
|
| 1215 |
+
"step": 67
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 0.1266294227188082,
|
| 1219 |
+
"grad_norm": 5.256114959716797,
|
| 1220 |
+
"learning_rate": 4.855072463768117e-06,
|
| 1221 |
+
"logps/chosen": -174.6351776123047,
|
| 1222 |
+
"logps/rejected": -348.933837890625,
|
| 1223 |
+
"loss": 0.0251,
|
| 1224 |
+
"losses/dpo": 0.09207179397344589,
|
| 1225 |
+
"losses/sft": 2.219914436340332,
|
| 1226 |
+
"losses/total": 0.09207179397344589,
|
| 1227 |
+
"ref_logps/chosen": -151.7803955078125,
|
| 1228 |
+
"ref_logps/rejected": -185.33236694335938,
|
| 1229 |
+
"rewards/accuracies": 1.0,
|
| 1230 |
+
"rewards/chosen": -2.285478353500366,
|
| 1231 |
+
"rewards/margins": 14.074670791625977,
|
| 1232 |
+
"rewards/rejected": -16.36014747619629,
|
| 1233 |
+
"step": 68
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 0.12849162011173185,
|
| 1237 |
+
"grad_norm": 5.340096950531006,
|
| 1238 |
+
"learning_rate": 4.84472049689441e-06,
|
| 1239 |
+
"logps/chosen": -161.56605529785156,
|
| 1240 |
+
"logps/rejected": -303.75201416015625,
|
| 1241 |
+
"loss": 0.027,
|
| 1242 |
+
"losses/dpo": 0.013381706550717354,
|
| 1243 |
+
"losses/sft": 2.1893093585968018,
|
| 1244 |
+
"losses/total": 0.013381706550717354,
|
| 1245 |
+
"ref_logps/chosen": -150.4385986328125,
|
| 1246 |
+
"ref_logps/rejected": -168.53863525390625,
|
| 1247 |
+
"rewards/accuracies": 1.0,
|
| 1248 |
+
"rewards/chosen": -1.1127451658248901,
|
| 1249 |
+
"rewards/margins": 12.40859317779541,
|
| 1250 |
+
"rewards/rejected": -13.521337509155273,
|
| 1251 |
+
"step": 69
|
| 1252 |
+
},
|
| 1253 |
+
{
|
| 1254 |
+
"epoch": 0.1303538175046555,
|
| 1255 |
+
"grad_norm": 13.288958549499512,
|
| 1256 |
+
"learning_rate": 4.834368530020705e-06,
|
| 1257 |
+
"logps/chosen": -172.6832275390625,
|
| 1258 |
+
"logps/rejected": -334.8218078613281,
|
| 1259 |
+
"loss": 0.1264,
|
| 1260 |
+
"losses/dpo": 0.0077868797816336155,
|
| 1261 |
+
"losses/sft": 2.6989755630493164,
|
| 1262 |
+
"losses/total": 0.0077868797816336155,
|
| 1263 |
+
"ref_logps/chosen": -152.52114868164062,
|
| 1264 |
+
"ref_logps/rejected": -174.5692901611328,
|
| 1265 |
+
"rewards/accuracies": 0.96875,
|
| 1266 |
+
"rewards/chosen": -2.0162088871002197,
|
| 1267 |
+
"rewards/margins": 14.009044647216797,
|
| 1268 |
+
"rewards/rejected": -16.02525520324707,
|
| 1269 |
+
"step": 70
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 0.13221601489757914,
|
| 1273 |
+
"grad_norm": 6.7856059074401855,
|
| 1274 |
+
"learning_rate": 4.824016563146998e-06,
|
| 1275 |
+
"logps/chosen": -166.0758056640625,
|
| 1276 |
+
"logps/rejected": -302.7711181640625,
|
| 1277 |
+
"loss": 0.0375,
|
| 1278 |
+
"losses/dpo": 0.008599307388067245,
|
| 1279 |
+
"losses/sft": 2.780581474304199,
|
| 1280 |
+
"losses/total": 0.008599307388067245,
|
| 1281 |
+
"ref_logps/chosen": -144.70916748046875,
|
| 1282 |
+
"ref_logps/rejected": -158.5432586669922,
|
| 1283 |
+
"rewards/accuracies": 0.984375,
|
| 1284 |
+
"rewards/chosen": -2.136662244796753,
|
| 1285 |
+
"rewards/margins": 12.286121368408203,
|
| 1286 |
+
"rewards/rejected": -14.422782897949219,
|
| 1287 |
+
"step": 71
|
| 1288 |
+
},
|
| 1289 |
+
{
|
| 1290 |
+
"epoch": 0.1340782122905028,
|
| 1291 |
+
"grad_norm": 7.3059282302856445,
|
| 1292 |
+
"learning_rate": 4.813664596273293e-06,
|
| 1293 |
+
"logps/chosen": -175.05950927734375,
|
| 1294 |
+
"logps/rejected": -346.2865905761719,
|
| 1295 |
+
"loss": 0.0272,
|
| 1296 |
+
"losses/dpo": 0.06259185075759888,
|
| 1297 |
+
"losses/sft": 2.080183982849121,
|
| 1298 |
+
"losses/total": 0.06259185075759888,
|
| 1299 |
+
"ref_logps/chosen": -153.53565979003906,
|
| 1300 |
+
"ref_logps/rejected": -175.8729248046875,
|
| 1301 |
+
"rewards/accuracies": 0.984375,
|
| 1302 |
+
"rewards/chosen": -2.152385950088501,
|
| 1303 |
+
"rewards/margins": 14.888980865478516,
|
| 1304 |
+
"rewards/rejected": -17.041364669799805,
|
| 1305 |
+
"step": 72
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 0.13594040968342644,
|
| 1309 |
+
"grad_norm": 14.326214790344238,
|
| 1310 |
+
"learning_rate": 4.803312629399586e-06,
|
| 1311 |
+
"logps/chosen": -159.51559448242188,
|
| 1312 |
+
"logps/rejected": -303.34576416015625,
|
| 1313 |
+
"loss": 0.1862,
|
| 1314 |
+
"losses/dpo": 0.1816004067659378,
|
| 1315 |
+
"losses/sft": 2.4103786945343018,
|
| 1316 |
+
"losses/total": 0.1816004067659378,
|
| 1317 |
+
"ref_logps/chosen": -137.89926147460938,
|
| 1318 |
+
"ref_logps/rejected": -151.60308837890625,
|
| 1319 |
+
"rewards/accuracies": 0.9375,
|
| 1320 |
+
"rewards/chosen": -2.1616339683532715,
|
| 1321 |
+
"rewards/margins": 13.0126371383667,
|
| 1322 |
+
"rewards/rejected": -15.174270629882812,
|
| 1323 |
+
"step": 73
|
| 1324 |
+
},
|
| 1325 |
+
{
|
| 1326 |
+
"epoch": 0.1378026070763501,
|
| 1327 |
+
"grad_norm": 11.816011428833008,
|
| 1328 |
+
"learning_rate": 4.79296066252588e-06,
|
| 1329 |
+
"logps/chosen": -163.58856201171875,
|
| 1330 |
+
"logps/rejected": -307.9797058105469,
|
| 1331 |
+
"loss": 0.076,
|
| 1332 |
+
"losses/dpo": 0.04001971334218979,
|
| 1333 |
+
"losses/sft": 2.3666553497314453,
|
| 1334 |
+
"losses/total": 0.04001971334218979,
|
| 1335 |
+
"ref_logps/chosen": -140.401123046875,
|
| 1336 |
+
"ref_logps/rejected": -169.53005981445312,
|
| 1337 |
+
"rewards/accuracies": 0.96875,
|
| 1338 |
+
"rewards/chosen": -2.3187437057495117,
|
| 1339 |
+
"rewards/margins": 11.52622127532959,
|
| 1340 |
+
"rewards/rejected": -13.844964981079102,
|
| 1341 |
+
"step": 74
|
| 1342 |
+
},
|
| 1343 |
+
{
|
| 1344 |
+
"epoch": 0.13966480446927373,
|
| 1345 |
+
"grad_norm": 13.491276741027832,
|
| 1346 |
+
"learning_rate": 4.782608695652174e-06,
|
| 1347 |
+
"logps/chosen": -163.01483154296875,
|
| 1348 |
+
"logps/rejected": -305.1317443847656,
|
| 1349 |
+
"loss": 0.1175,
|
| 1350 |
+
"losses/dpo": 0.021199261769652367,
|
| 1351 |
+
"losses/sft": 2.5555906295776367,
|
| 1352 |
+
"losses/total": 0.021199261769652367,
|
| 1353 |
+
"ref_logps/chosen": -135.76666259765625,
|
| 1354 |
+
"ref_logps/rejected": -159.01651000976562,
|
| 1355 |
+
"rewards/accuracies": 0.96875,
|
| 1356 |
+
"rewards/chosen": -2.72481632232666,
|
| 1357 |
+
"rewards/margins": 11.88670825958252,
|
| 1358 |
+
"rewards/rejected": -14.61152458190918,
|
| 1359 |
+
"step": 75
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 0.14152700186219738,
|
| 1363 |
+
"grad_norm": 13.310988426208496,
|
| 1364 |
+
"learning_rate": 4.772256728778468e-06,
|
| 1365 |
+
"logps/chosen": -166.88494873046875,
|
| 1366 |
+
"logps/rejected": -300.84918212890625,
|
| 1367 |
+
"loss": 0.0695,
|
| 1368 |
+
"losses/dpo": 0.23336642980575562,
|
| 1369 |
+
"losses/sft": 2.7684144973754883,
|
| 1370 |
+
"losses/total": 0.23336642980575562,
|
| 1371 |
+
"ref_logps/chosen": -145.76715087890625,
|
| 1372 |
+
"ref_logps/rejected": -162.08209228515625,
|
| 1373 |
+
"rewards/accuracies": 0.953125,
|
| 1374 |
+
"rewards/chosen": -2.1117796897888184,
|
| 1375 |
+
"rewards/margins": 11.764933586120605,
|
| 1376 |
+
"rewards/rejected": -13.876713752746582,
|
| 1377 |
+
"step": 76
|
| 1378 |
+
},
|
| 1379 |
+
{
|
| 1380 |
+
"epoch": 0.14338919925512103,
|
| 1381 |
+
"grad_norm": 5.915053844451904,
|
| 1382 |
+
"learning_rate": 4.761904761904762e-06,
|
| 1383 |
+
"logps/chosen": -166.23074340820312,
|
| 1384 |
+
"logps/rejected": -321.7993469238281,
|
| 1385 |
+
"loss": 0.0245,
|
| 1386 |
+
"losses/dpo": 0.00781274028122425,
|
| 1387 |
+
"losses/sft": 2.6997854709625244,
|
| 1388 |
+
"losses/total": 0.00781274028122425,
|
| 1389 |
+
"ref_logps/chosen": -146.89723205566406,
|
| 1390 |
+
"ref_logps/rejected": -168.7777099609375,
|
| 1391 |
+
"rewards/accuracies": 1.0,
|
| 1392 |
+
"rewards/chosen": -1.9333515167236328,
|
| 1393 |
+
"rewards/margins": 13.368810653686523,
|
| 1394 |
+
"rewards/rejected": -15.302162170410156,
|
| 1395 |
+
"step": 77
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 0.1452513966480447,
|
| 1399 |
+
"grad_norm": 8.788707733154297,
|
| 1400 |
+
"learning_rate": 4.751552795031056e-06,
|
| 1401 |
+
"logps/chosen": -175.02999877929688,
|
| 1402 |
+
"logps/rejected": -321.0180969238281,
|
| 1403 |
+
"loss": 0.0271,
|
| 1404 |
+
"losses/dpo": 0.14333294332027435,
|
| 1405 |
+
"losses/sft": 2.538876533508301,
|
| 1406 |
+
"losses/total": 0.14333294332027435,
|
| 1407 |
+
"ref_logps/chosen": -157.10501098632812,
|
| 1408 |
+
"ref_logps/rejected": -181.61178588867188,
|
| 1409 |
+
"rewards/accuracies": 0.984375,
|
| 1410 |
+
"rewards/chosen": -1.7924991846084595,
|
| 1411 |
+
"rewards/margins": 12.148136138916016,
|
| 1412 |
+
"rewards/rejected": -13.940634727478027,
|
| 1413 |
+
"step": 78
|
| 1414 |
+
},
|
| 1415 |
+
{
|
| 1416 |
+
"epoch": 0.14711359404096835,
|
| 1417 |
+
"grad_norm": 3.836193323135376,
|
| 1418 |
+
"learning_rate": 4.74120082815735e-06,
|
| 1419 |
+
"logps/chosen": -161.6851806640625,
|
| 1420 |
+
"logps/rejected": -307.22308349609375,
|
| 1421 |
+
"loss": 0.018,
|
| 1422 |
+
"losses/dpo": 0.005441778339445591,
|
| 1423 |
+
"losses/sft": 2.4244680404663086,
|
| 1424 |
+
"losses/total": 0.005441778339445591,
|
| 1425 |
+
"ref_logps/chosen": -141.81394958496094,
|
| 1426 |
+
"ref_logps/rejected": -166.87547302246094,
|
| 1427 |
+
"rewards/accuracies": 1.0,
|
| 1428 |
+
"rewards/chosen": -1.9871225357055664,
|
| 1429 |
+
"rewards/margins": 12.047636985778809,
|
| 1430 |
+
"rewards/rejected": -14.034758567810059,
|
| 1431 |
+
"step": 79
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 0.148975791433892,
|
| 1435 |
+
"grad_norm": 1.0960124731063843,
|
| 1436 |
+
"learning_rate": 4.730848861283645e-06,
|
| 1437 |
+
"logps/chosen": -164.8990020751953,
|
| 1438 |
+
"logps/rejected": -306.0439453125,
|
| 1439 |
+
"loss": 0.0046,
|
| 1440 |
+
"losses/dpo": 0.00012883917952422053,
|
| 1441 |
+
"losses/sft": 2.337103843688965,
|
| 1442 |
+
"losses/total": 0.00012883917952422053,
|
| 1443 |
+
"ref_logps/chosen": -150.0802001953125,
|
| 1444 |
+
"ref_logps/rejected": -169.78543090820312,
|
| 1445 |
+
"rewards/accuracies": 1.0,
|
| 1446 |
+
"rewards/chosen": -1.48188054561615,
|
| 1447 |
+
"rewards/margins": 12.143974304199219,
|
| 1448 |
+
"rewards/rejected": -13.6258544921875,
|
| 1449 |
+
"step": 80
|
| 1450 |
+
},
|
| 1451 |
+
{
|
| 1452 |
+
"epoch": 0.15083798882681565,
|
| 1453 |
+
"grad_norm": 10.833060264587402,
|
| 1454 |
+
"learning_rate": 4.7204968944099384e-06,
|
| 1455 |
+
"logps/chosen": -166.50360107421875,
|
| 1456 |
+
"logps/rejected": -291.5777893066406,
|
| 1457 |
+
"loss": 0.0699,
|
| 1458 |
+
"losses/dpo": 0.3678208291530609,
|
| 1459 |
+
"losses/sft": 2.360893726348877,
|
| 1460 |
+
"losses/total": 0.3678208291530609,
|
| 1461 |
+
"ref_logps/chosen": -139.50567626953125,
|
| 1462 |
+
"ref_logps/rejected": -159.87924194335938,
|
| 1463 |
+
"rewards/accuracies": 0.984375,
|
| 1464 |
+
"rewards/chosen": -2.699791193008423,
|
| 1465 |
+
"rewards/margins": 10.470064163208008,
|
| 1466 |
+
"rewards/rejected": -13.169855117797852,
|
| 1467 |
+
"step": 81
|
| 1468 |
+
},
|
| 1469 |
+
{
|
| 1470 |
+
"epoch": 0.1527001862197393,
|
| 1471 |
+
"grad_norm": 10.486247062683105,
|
| 1472 |
+
"learning_rate": 4.710144927536232e-06,
|
| 1473 |
+
"logps/chosen": -158.02520751953125,
|
| 1474 |
+
"logps/rejected": -298.325927734375,
|
| 1475 |
+
"loss": 0.0641,
|
| 1476 |
+
"losses/dpo": 0.04021751880645752,
|
| 1477 |
+
"losses/sft": 2.221280574798584,
|
| 1478 |
+
"losses/total": 0.04021751880645752,
|
| 1479 |
+
"ref_logps/chosen": -147.9010009765625,
|
| 1480 |
+
"ref_logps/rejected": -168.83761596679688,
|
| 1481 |
+
"rewards/accuracies": 0.96875,
|
| 1482 |
+
"rewards/chosen": -1.0124225616455078,
|
| 1483 |
+
"rewards/margins": 11.936410903930664,
|
| 1484 |
+
"rewards/rejected": -12.948833465576172,
|
| 1485 |
+
"step": 82
|
| 1486 |
+
},
|
| 1487 |
+
{
|
| 1488 |
+
"epoch": 0.15456238361266295,
|
| 1489 |
+
"grad_norm": 4.4983320236206055,
|
| 1490 |
+
"learning_rate": 4.6997929606625265e-06,
|
| 1491 |
+
"logps/chosen": -157.27609252929688,
|
| 1492 |
+
"logps/rejected": -284.30670166015625,
|
| 1493 |
+
"loss": 0.0134,
|
| 1494 |
+
"losses/dpo": 0.07455439865589142,
|
| 1495 |
+
"losses/sft": 2.680065631866455,
|
| 1496 |
+
"losses/total": 0.07455439865589142,
|
| 1497 |
+
"ref_logps/chosen": -143.0989990234375,
|
| 1498 |
+
"ref_logps/rejected": -166.63612365722656,
|
| 1499 |
+
"rewards/accuracies": 1.0,
|
| 1500 |
+
"rewards/chosen": -1.4177087545394897,
|
| 1501 |
+
"rewards/margins": 10.349348068237305,
|
| 1502 |
+
"rewards/rejected": -11.767056465148926,
|
| 1503 |
+
"step": 83
|
| 1504 |
+
},
|
| 1505 |
+
{
|
| 1506 |
+
"epoch": 0.1564245810055866,
|
| 1507 |
+
"grad_norm": 5.925443172454834,
|
| 1508 |
+
"learning_rate": 4.68944099378882e-06,
|
| 1509 |
+
"logps/chosen": -166.75436401367188,
|
| 1510 |
+
"logps/rejected": -307.4884338378906,
|
| 1511 |
+
"loss": 0.0294,
|
| 1512 |
+
"losses/dpo": 0.018268397077918053,
|
| 1513 |
+
"losses/sft": 2.3180205821990967,
|
| 1514 |
+
"losses/total": 0.018268397077918053,
|
| 1515 |
+
"ref_logps/chosen": -146.59835815429688,
|
| 1516 |
+
"ref_logps/rejected": -171.87164306640625,
|
| 1517 |
+
"rewards/accuracies": 0.984375,
|
| 1518 |
+
"rewards/chosen": -2.015601634979248,
|
| 1519 |
+
"rewards/margins": 11.546079635620117,
|
| 1520 |
+
"rewards/rejected": -13.561680793762207,
|
| 1521 |
+
"step": 84
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 0.15828677839851024,
|
| 1525 |
+
"grad_norm": 10.823968887329102,
|
| 1526 |
+
"learning_rate": 4.679089026915114e-06,
|
| 1527 |
+
"logps/chosen": -164.76190185546875,
|
| 1528 |
+
"logps/rejected": -291.744873046875,
|
| 1529 |
+
"loss": 0.0787,
|
| 1530 |
+
"losses/dpo": 0.0033458764664828777,
|
| 1531 |
+
"losses/sft": 2.6443734169006348,
|
| 1532 |
+
"losses/total": 0.0033458764664828777,
|
| 1533 |
+
"ref_logps/chosen": -147.3948516845703,
|
| 1534 |
+
"ref_logps/rejected": -165.80938720703125,
|
| 1535 |
+
"rewards/accuracies": 0.96875,
|
| 1536 |
+
"rewards/chosen": -1.73670494556427,
|
| 1537 |
+
"rewards/margins": 10.856843948364258,
|
| 1538 |
+
"rewards/rejected": -12.593548774719238,
|
| 1539 |
+
"step": 85
|
| 1540 |
+
},
|
| 1541 |
+
{
|
| 1542 |
+
"epoch": 0.1601489757914339,
|
| 1543 |
+
"grad_norm": 10.07721996307373,
|
| 1544 |
+
"learning_rate": 4.668737060041408e-06,
|
| 1545 |
+
"logps/chosen": -170.19273376464844,
|
| 1546 |
+
"logps/rejected": -289.0329284667969,
|
| 1547 |
+
"loss": 0.0578,
|
| 1548 |
+
"losses/dpo": 0.22292722761631012,
|
| 1549 |
+
"losses/sft": 2.8669309616088867,
|
| 1550 |
+
"losses/total": 0.22292722761631012,
|
| 1551 |
+
"ref_logps/chosen": -155.04888916015625,
|
| 1552 |
+
"ref_logps/rejected": -169.506103515625,
|
| 1553 |
+
"rewards/accuracies": 0.984375,
|
| 1554 |
+
"rewards/chosen": -1.5143851041793823,
|
| 1555 |
+
"rewards/margins": 10.438299179077148,
|
| 1556 |
+
"rewards/rejected": -11.95268440246582,
|
| 1557 |
+
"step": 86
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 0.16201117318435754,
|
| 1561 |
+
"grad_norm": 7.202996730804443,
|
| 1562 |
+
"learning_rate": 4.6583850931677025e-06,
|
| 1563 |
+
"logps/chosen": -152.85894775390625,
|
| 1564 |
+
"logps/rejected": -276.5400390625,
|
| 1565 |
+
"loss": 0.0464,
|
| 1566 |
+
"losses/dpo": 0.0008024298003874719,
|
| 1567 |
+
"losses/sft": 2.105069637298584,
|
| 1568 |
+
"losses/total": 0.0008024298003874719,
|
| 1569 |
+
"ref_logps/chosen": -135.02166748046875,
|
| 1570 |
+
"ref_logps/rejected": -162.55917358398438,
|
| 1571 |
+
"rewards/accuracies": 0.96875,
|
| 1572 |
+
"rewards/chosen": -1.783726453781128,
|
| 1573 |
+
"rewards/margins": 9.614358901977539,
|
| 1574 |
+
"rewards/rejected": -11.39808464050293,
|
| 1575 |
+
"step": 87
|
| 1576 |
+
},
|
| 1577 |
+
{
|
| 1578 |
+
"epoch": 0.16387337057728119,
|
| 1579 |
+
"grad_norm": 10.529645919799805,
|
| 1580 |
+
"learning_rate": 4.648033126293996e-06,
|
| 1581 |
+
"logps/chosen": -167.41766357421875,
|
| 1582 |
+
"logps/rejected": -312.9851989746094,
|
| 1583 |
+
"loss": 0.1059,
|
| 1584 |
+
"losses/dpo": 1.8586399164632894e-05,
|
| 1585 |
+
"losses/sft": 2.3417227268218994,
|
| 1586 |
+
"losses/total": 1.8586399164632894e-05,
|
| 1587 |
+
"ref_logps/chosen": -150.24447631835938,
|
| 1588 |
+
"ref_logps/rejected": -171.88534545898438,
|
| 1589 |
+
"rewards/accuracies": 0.984375,
|
| 1590 |
+
"rewards/chosen": -1.7173188924789429,
|
| 1591 |
+
"rewards/margins": 12.392667770385742,
|
| 1592 |
+
"rewards/rejected": -14.1099853515625,
|
| 1593 |
+
"step": 88
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 0.16573556797020483,
|
| 1597 |
+
"grad_norm": 5.379014492034912,
|
| 1598 |
+
"learning_rate": 4.637681159420291e-06,
|
| 1599 |
+
"logps/chosen": -149.58767700195312,
|
| 1600 |
+
"logps/rejected": -283.6622009277344,
|
| 1601 |
+
"loss": 0.0265,
|
| 1602 |
+
"losses/dpo": 0.0076700253412127495,
|
| 1603 |
+
"losses/sft": 2.0566470623016357,
|
| 1604 |
+
"losses/total": 0.0076700253412127495,
|
| 1605 |
+
"ref_logps/chosen": -142.82745361328125,
|
| 1606 |
+
"ref_logps/rejected": -166.57040405273438,
|
| 1607 |
+
"rewards/accuracies": 1.0,
|
| 1608 |
+
"rewards/chosen": -0.6760219931602478,
|
| 1609 |
+
"rewards/margins": 11.03315544128418,
|
| 1610 |
+
"rewards/rejected": -11.709177017211914,
|
| 1611 |
+
"step": 89
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 0.16759776536312848,
|
| 1615 |
+
"grad_norm": 7.1271820068359375,
|
| 1616 |
+
"learning_rate": 4.627329192546584e-06,
|
| 1617 |
+
"logps/chosen": -166.07994079589844,
|
| 1618 |
+
"logps/rejected": -288.1270446777344,
|
| 1619 |
+
"loss": 0.0383,
|
| 1620 |
+
"losses/dpo": 0.006637689657509327,
|
| 1621 |
+
"losses/sft": 2.695986270904541,
|
| 1622 |
+
"losses/total": 0.006637689657509327,
|
| 1623 |
+
"ref_logps/chosen": -152.90316772460938,
|
| 1624 |
+
"ref_logps/rejected": -167.97433471679688,
|
| 1625 |
+
"rewards/accuracies": 0.984375,
|
| 1626 |
+
"rewards/chosen": -1.3176772594451904,
|
| 1627 |
+
"rewards/margins": 10.697591781616211,
|
| 1628 |
+
"rewards/rejected": -12.01526927947998,
|
| 1629 |
+
"step": 90
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 0.16945996275605213,
|
| 1633 |
+
"grad_norm": 3.960089683532715,
|
| 1634 |
+
"learning_rate": 4.616977225672879e-06,
|
| 1635 |
+
"logps/chosen": -155.234619140625,
|
| 1636 |
+
"logps/rejected": -274.2052917480469,
|
| 1637 |
+
"loss": 0.0283,
|
| 1638 |
+
"losses/dpo": 0.029921425506472588,
|
| 1639 |
+
"losses/sft": 2.448662757873535,
|
| 1640 |
+
"losses/total": 0.029921425506472588,
|
| 1641 |
+
"ref_logps/chosen": -146.57998657226562,
|
| 1642 |
+
"ref_logps/rejected": -161.9693603515625,
|
| 1643 |
+
"rewards/accuracies": 1.0,
|
| 1644 |
+
"rewards/chosen": -0.8654617071151733,
|
| 1645 |
+
"rewards/margins": 10.35813045501709,
|
| 1646 |
+
"rewards/rejected": -11.223592758178711,
|
| 1647 |
+
"step": 91
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 0.1713221601489758,
|
| 1651 |
+
"grad_norm": 8.884647369384766,
|
| 1652 |
+
"learning_rate": 4.606625258799172e-06,
|
| 1653 |
+
"logps/chosen": -156.59239196777344,
|
| 1654 |
+
"logps/rejected": -269.66265869140625,
|
| 1655 |
+
"loss": 0.0549,
|
| 1656 |
+
"losses/dpo": 0.000325945409713313,
|
| 1657 |
+
"losses/sft": 2.27170991897583,
|
| 1658 |
+
"losses/total": 0.000325945409713313,
|
| 1659 |
+
"ref_logps/chosen": -143.44590759277344,
|
| 1660 |
+
"ref_logps/rejected": -160.02975463867188,
|
| 1661 |
+
"rewards/accuracies": 0.96875,
|
| 1662 |
+
"rewards/chosen": -1.3146480321884155,
|
| 1663 |
+
"rewards/margins": 9.648642539978027,
|
| 1664 |
+
"rewards/rejected": -10.963289260864258,
|
| 1665 |
+
"step": 92
|
| 1666 |
+
},
|
| 1667 |
+
{
|
| 1668 |
+
"epoch": 0.17318435754189945,
|
| 1669 |
+
"grad_norm": 8.483807563781738,
|
| 1670 |
+
"learning_rate": 4.596273291925466e-06,
|
| 1671 |
+
"logps/chosen": -157.52755737304688,
|
| 1672 |
+
"logps/rejected": -302.79736328125,
|
| 1673 |
+
"loss": 0.0525,
|
| 1674 |
+
"losses/dpo": 0.007826759479939938,
|
| 1675 |
+
"losses/sft": 2.404571533203125,
|
| 1676 |
+
"losses/total": 0.007826759479939938,
|
| 1677 |
+
"ref_logps/chosen": -151.92408752441406,
|
| 1678 |
+
"ref_logps/rejected": -174.8727569580078,
|
| 1679 |
+
"rewards/accuracies": 0.96875,
|
| 1680 |
+
"rewards/chosen": -0.5603480339050293,
|
| 1681 |
+
"rewards/margins": 12.232114791870117,
|
| 1682 |
+
"rewards/rejected": -12.792463302612305,
|
| 1683 |
+
"step": 93
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 0.1750465549348231,
|
| 1687 |
+
"grad_norm": 2.803177833557129,
|
| 1688 |
+
"learning_rate": 4.58592132505176e-06,
|
| 1689 |
+
"logps/chosen": -154.108154296875,
|
| 1690 |
+
"logps/rejected": -316.1098937988281,
|
| 1691 |
+
"loss": 0.0115,
|
| 1692 |
+
"losses/dpo": 0.014974968507885933,
|
| 1693 |
+
"losses/sft": 2.5976779460906982,
|
| 1694 |
+
"losses/total": 0.014974968507885933,
|
| 1695 |
+
"ref_logps/chosen": -139.6196746826172,
|
| 1696 |
+
"ref_logps/rejected": -177.22061157226562,
|
| 1697 |
+
"rewards/accuracies": 1.0,
|
| 1698 |
+
"rewards/chosen": -1.4488475322723389,
|
| 1699 |
+
"rewards/margins": 12.440080642700195,
|
| 1700 |
+
"rewards/rejected": -13.88892936706543,
|
| 1701 |
+
"step": 94
|
| 1702 |
+
},
|
| 1703 |
+
{
|
| 1704 |
+
"epoch": 0.17690875232774675,
|
| 1705 |
+
"grad_norm": 10.011286735534668,
|
| 1706 |
+
"learning_rate": 4.575569358178054e-06,
|
| 1707 |
+
"logps/chosen": -161.99615478515625,
|
| 1708 |
+
"logps/rejected": -290.70928955078125,
|
| 1709 |
+
"loss": 0.0521,
|
| 1710 |
+
"losses/dpo": 0.004248100332915783,
|
| 1711 |
+
"losses/sft": 2.6869282722473145,
|
| 1712 |
+
"losses/total": 0.004248100332915783,
|
| 1713 |
+
"ref_logps/chosen": -150.13644409179688,
|
| 1714 |
+
"ref_logps/rejected": -170.98199462890625,
|
| 1715 |
+
"rewards/accuracies": 0.96875,
|
| 1716 |
+
"rewards/chosen": -1.1859712600708008,
|
| 1717 |
+
"rewards/margins": 10.786758422851562,
|
| 1718 |
+
"rewards/rejected": -11.972728729248047,
|
| 1719 |
+
"step": 95
|
| 1720 |
+
},
|
| 1721 |
+
{
|
| 1722 |
+
"epoch": 0.1787709497206704,
|
| 1723 |
+
"grad_norm": 13.34592342376709,
|
| 1724 |
+
"learning_rate": 4.565217391304348e-06,
|
| 1725 |
+
"logps/chosen": -157.46038818359375,
|
| 1726 |
+
"logps/rejected": -293.8796691894531,
|
| 1727 |
+
"loss": 0.0613,
|
| 1728 |
+
"losses/dpo": 0.0029001892544329166,
|
| 1729 |
+
"losses/sft": 2.191887378692627,
|
| 1730 |
+
"losses/total": 0.0029001892544329166,
|
| 1731 |
+
"ref_logps/chosen": -147.4150390625,
|
| 1732 |
+
"ref_logps/rejected": -175.098876953125,
|
| 1733 |
+
"rewards/accuracies": 0.96875,
|
| 1734 |
+
"rewards/chosen": -1.004532814025879,
|
| 1735 |
+
"rewards/margins": 10.873546600341797,
|
| 1736 |
+
"rewards/rejected": -11.878079414367676,
|
| 1737 |
+
"step": 96
|
| 1738 |
+
},
|
| 1739 |
+
{
|
| 1740 |
+
"epoch": 0.18063314711359404,
|
| 1741 |
+
"grad_norm": 13.375811576843262,
|
| 1742 |
+
"learning_rate": 4.554865424430642e-06,
|
| 1743 |
+
"logps/chosen": -166.24191284179688,
|
| 1744 |
+
"logps/rejected": -305.58819580078125,
|
| 1745 |
+
"loss": 0.0981,
|
| 1746 |
+
"losses/dpo": 0.0007697511464357376,
|
| 1747 |
+
"losses/sft": 2.3007588386535645,
|
| 1748 |
+
"losses/total": 0.0007697511464357376,
|
| 1749 |
+
"ref_logps/chosen": -144.6219482421875,
|
| 1750 |
+
"ref_logps/rejected": -174.1811981201172,
|
| 1751 |
+
"rewards/accuracies": 0.953125,
|
| 1752 |
+
"rewards/chosen": -2.1619954109191895,
|
| 1753 |
+
"rewards/margins": 10.978706359863281,
|
| 1754 |
+
"rewards/rejected": -13.140701293945312,
|
| 1755 |
+
"step": 97
|
| 1756 |
+
},
|
| 1757 |
+
{
|
| 1758 |
+
"epoch": 0.1824953445065177,
|
| 1759 |
+
"grad_norm": 11.116385459899902,
|
| 1760 |
+
"learning_rate": 4.544513457556936e-06,
|
| 1761 |
+
"logps/chosen": -151.30795288085938,
|
| 1762 |
+
"logps/rejected": -303.7850341796875,
|
| 1763 |
+
"loss": 0.0838,
|
| 1764 |
+
"losses/dpo": 0.006321331951767206,
|
| 1765 |
+
"losses/sft": 2.45919132232666,
|
| 1766 |
+
"losses/total": 0.006321331951767206,
|
| 1767 |
+
"ref_logps/chosen": -141.2546844482422,
|
| 1768 |
+
"ref_logps/rejected": -166.2907257080078,
|
| 1769 |
+
"rewards/accuracies": 0.96875,
|
| 1770 |
+
"rewards/chosen": -1.0053260326385498,
|
| 1771 |
+
"rewards/margins": 12.744101524353027,
|
| 1772 |
+
"rewards/rejected": -13.749427795410156,
|
| 1773 |
+
"step": 98
|
| 1774 |
+
},
|
| 1775 |
+
{
|
| 1776 |
+
"epoch": 0.18435754189944134,
|
| 1777 |
+
"grad_norm": 8.1275053024292,
|
| 1778 |
+
"learning_rate": 4.534161490683231e-06,
|
| 1779 |
+
"logps/chosen": -169.12844848632812,
|
| 1780 |
+
"logps/rejected": -291.58355712890625,
|
| 1781 |
+
"loss": 0.0379,
|
| 1782 |
+
"losses/dpo": 0.04353442043066025,
|
| 1783 |
+
"losses/sft": 2.9029669761657715,
|
| 1784 |
+
"losses/total": 0.04353442043066025,
|
| 1785 |
+
"ref_logps/chosen": -148.775390625,
|
| 1786 |
+
"ref_logps/rejected": -162.24786376953125,
|
| 1787 |
+
"rewards/accuracies": 0.984375,
|
| 1788 |
+
"rewards/chosen": -2.0353074073791504,
|
| 1789 |
+
"rewards/margins": 10.898263931274414,
|
| 1790 |
+
"rewards/rejected": -12.933570861816406,
|
| 1791 |
+
"step": 99
|
| 1792 |
+
},
|
| 1793 |
+
{
|
| 1794 |
+
"epoch": 0.186219739292365,
|
| 1795 |
+
"grad_norm": 3.8263306617736816,
|
| 1796 |
+
"learning_rate": 4.523809523809524e-06,
|
| 1797 |
+
"logps/chosen": -160.37161254882812,
|
| 1798 |
+
"logps/rejected": -291.48577880859375,
|
| 1799 |
+
"loss": 0.0159,
|
| 1800 |
+
"losses/dpo": 0.06080714985728264,
|
| 1801 |
+
"losses/sft": 2.056828498840332,
|
| 1802 |
+
"losses/total": 0.06080714985728264,
|
| 1803 |
+
"ref_logps/chosen": -143.4698486328125,
|
| 1804 |
+
"ref_logps/rejected": -165.9519500732422,
|
| 1805 |
+
"rewards/accuracies": 1.0,
|
| 1806 |
+
"rewards/chosen": -1.6901748180389404,
|
| 1807 |
+
"rewards/margins": 10.863210678100586,
|
| 1808 |
+
"rewards/rejected": -12.553384780883789,
|
| 1809 |
+
"step": 100
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 0.18808193668528864,
|
| 1813 |
+
"grad_norm": 9.816635131835938,
|
| 1814 |
+
"learning_rate": 4.513457556935818e-06,
|
| 1815 |
+
"logps/chosen": -177.78240966796875,
|
| 1816 |
+
"logps/rejected": -330.1160888671875,
|
| 1817 |
+
"loss": 0.0489,
|
| 1818 |
+
"losses/dpo": 0.2933081388473511,
|
| 1819 |
+
"losses/sft": 2.7717084884643555,
|
| 1820 |
+
"losses/total": 0.2933081388473511,
|
| 1821 |
+
"ref_logps/chosen": -155.60540771484375,
|
| 1822 |
+
"ref_logps/rejected": -177.44467163085938,
|
| 1823 |
+
"rewards/accuracies": 0.984375,
|
| 1824 |
+
"rewards/chosen": -2.2176995277404785,
|
| 1825 |
+
"rewards/margins": 13.04944133758545,
|
| 1826 |
+
"rewards/rejected": -15.267141342163086,
|
| 1827 |
+
"step": 101
|
| 1828 |
+
},
|
| 1829 |
+
{
|
| 1830 |
+
"epoch": 0.18994413407821228,
|
| 1831 |
+
"grad_norm": 1.3710126876831055,
|
| 1832 |
+
"learning_rate": 4.503105590062112e-06,
|
| 1833 |
+
"logps/chosen": -166.66590881347656,
|
| 1834 |
+
"logps/rejected": -290.761962890625,
|
| 1835 |
+
"loss": 0.0054,
|
| 1836 |
+
"losses/dpo": 0.02758178301155567,
|
| 1837 |
+
"losses/sft": 3.162290096282959,
|
| 1838 |
+
"losses/total": 0.02758178301155567,
|
| 1839 |
+
"ref_logps/chosen": -147.08917236328125,
|
| 1840 |
+
"ref_logps/rejected": -151.07664489746094,
|
| 1841 |
+
"rewards/accuracies": 1.0,
|
| 1842 |
+
"rewards/chosen": -1.9576739072799683,
|
| 1843 |
+
"rewards/margins": 12.010858535766602,
|
| 1844 |
+
"rewards/rejected": -13.96853256225586,
|
| 1845 |
+
"step": 102
|
| 1846 |
+
},
|
| 1847 |
+
{
|
| 1848 |
+
"epoch": 0.19180633147113593,
|
| 1849 |
+
"grad_norm": 7.93828010559082,
|
| 1850 |
+
"learning_rate": 4.492753623188406e-06,
|
| 1851 |
+
"logps/chosen": -146.8543701171875,
|
| 1852 |
+
"logps/rejected": -312.2845764160156,
|
| 1853 |
+
"loss": 0.0425,
|
| 1854 |
+
"losses/dpo": 0.2460930347442627,
|
| 1855 |
+
"losses/sft": 2.2462611198425293,
|
| 1856 |
+
"losses/total": 0.2460930347442627,
|
| 1857 |
+
"ref_logps/chosen": -135.14158630371094,
|
| 1858 |
+
"ref_logps/rejected": -168.27618408203125,
|
| 1859 |
+
"rewards/accuracies": 0.984375,
|
| 1860 |
+
"rewards/chosen": -1.1712779998779297,
|
| 1861 |
+
"rewards/margins": 13.229562759399414,
|
| 1862 |
+
"rewards/rejected": -14.400840759277344,
|
| 1863 |
+
"step": 103
|
| 1864 |
+
},
|
| 1865 |
+
{
|
| 1866 |
+
"epoch": 0.19366852886405958,
|
| 1867 |
+
"grad_norm": 13.03959846496582,
|
| 1868 |
+
"learning_rate": 4.4824016563146996e-06,
|
| 1869 |
+
"logps/chosen": -163.9506378173828,
|
| 1870 |
+
"logps/rejected": -338.9526062011719,
|
| 1871 |
+
"loss": 0.105,
|
| 1872 |
+
"losses/dpo": 1.4241080862120725e-05,
|
| 1873 |
+
"losses/sft": 2.1639623641967773,
|
| 1874 |
+
"losses/total": 1.4241080862120725e-05,
|
| 1875 |
+
"ref_logps/chosen": -144.27719116210938,
|
| 1876 |
+
"ref_logps/rejected": -172.1199951171875,
|
| 1877 |
+
"rewards/accuracies": 0.984375,
|
| 1878 |
+
"rewards/chosen": -1.9673457145690918,
|
| 1879 |
+
"rewards/margins": 14.71591567993164,
|
| 1880 |
+
"rewards/rejected": -16.683259963989258,
|
| 1881 |
+
"step": 104
|
| 1882 |
+
},
|
| 1883 |
+
{
|
| 1884 |
+
"epoch": 0.19553072625698323,
|
| 1885 |
+
"grad_norm": 12.940238952636719,
|
| 1886 |
+
"learning_rate": 4.472049689440994e-06,
|
| 1887 |
+
"logps/chosen": -167.60537719726562,
|
| 1888 |
+
"logps/rejected": -305.2105407714844,
|
| 1889 |
+
"loss": 0.1266,
|
| 1890 |
+
"losses/dpo": 0.016943011432886124,
|
| 1891 |
+
"losses/sft": 2.568720817565918,
|
| 1892 |
+
"losses/total": 0.016943011432886124,
|
| 1893 |
+
"ref_logps/chosen": -144.21282958984375,
|
| 1894 |
+
"ref_logps/rejected": -158.3541259765625,
|
| 1895 |
+
"rewards/accuracies": 0.953125,
|
| 1896 |
+
"rewards/chosen": -2.339254140853882,
|
| 1897 |
+
"rewards/margins": 12.34638786315918,
|
| 1898 |
+
"rewards/rejected": -14.685641288757324,
|
| 1899 |
+
"step": 105
|
| 1900 |
+
},
|
| 1901 |
+
{
|
| 1902 |
+
"epoch": 0.1973929236499069,
|
| 1903 |
+
"grad_norm": 9.413414001464844,
|
| 1904 |
+
"learning_rate": 4.4616977225672884e-06,
|
| 1905 |
+
"logps/chosen": -171.18016052246094,
|
| 1906 |
+
"logps/rejected": -302.7274475097656,
|
| 1907 |
+
"loss": 0.0475,
|
| 1908 |
+
"losses/dpo": 0.0004924152162857354,
|
| 1909 |
+
"losses/sft": 2.5999226570129395,
|
| 1910 |
+
"losses/total": 0.0004924152162857354,
|
| 1911 |
+
"ref_logps/chosen": -144.5438690185547,
|
| 1912 |
+
"ref_logps/rejected": -162.2921905517578,
|
| 1913 |
+
"rewards/accuracies": 0.984375,
|
| 1914 |
+
"rewards/chosen": -2.6636300086975098,
|
| 1915 |
+
"rewards/margins": 11.37989616394043,
|
| 1916 |
+
"rewards/rejected": -14.043526649475098,
|
| 1917 |
+
"step": 106
|
| 1918 |
+
},
|
| 1919 |
+
{
|
| 1920 |
+
"epoch": 0.19925512104283055,
|
| 1921 |
+
"grad_norm": 1.439731240272522,
|
| 1922 |
+
"learning_rate": 4.451345755693582e-06,
|
| 1923 |
+
"logps/chosen": -169.63963317871094,
|
| 1924 |
+
"logps/rejected": -335.1871337890625,
|
| 1925 |
+
"loss": 0.0046,
|
| 1926 |
+
"losses/dpo": 0.01760055683553219,
|
| 1927 |
+
"losses/sft": 2.191661834716797,
|
| 1928 |
+
"losses/total": 0.01760055683553219,
|
| 1929 |
+
"ref_logps/chosen": -148.11293029785156,
|
| 1930 |
+
"ref_logps/rejected": -171.40560913085938,
|
| 1931 |
+
"rewards/accuracies": 1.0,
|
| 1932 |
+
"rewards/chosen": -2.1526713371276855,
|
| 1933 |
+
"rewards/margins": 14.225481986999512,
|
| 1934 |
+
"rewards/rejected": -16.37815284729004,
|
| 1935 |
+
"step": 107
|
| 1936 |
+
},
|
| 1937 |
+
{
|
| 1938 |
+
"epoch": 0.2011173184357542,
|
| 1939 |
+
"grad_norm": 12.107815742492676,
|
| 1940 |
+
"learning_rate": 4.4409937888198765e-06,
|
| 1941 |
+
"logps/chosen": -174.36929321289062,
|
| 1942 |
+
"logps/rejected": -322.1456604003906,
|
| 1943 |
+
"loss": 0.1067,
|
| 1944 |
+
"losses/dpo": 0.0005443975096568465,
|
| 1945 |
+
"losses/sft": 2.864640712738037,
|
| 1946 |
+
"losses/total": 0.0005443975096568465,
|
| 1947 |
+
"ref_logps/chosen": -149.93984985351562,
|
| 1948 |
+
"ref_logps/rejected": -166.4761962890625,
|
| 1949 |
+
"rewards/accuracies": 0.953125,
|
| 1950 |
+
"rewards/chosen": -2.442944288253784,
|
| 1951 |
+
"rewards/margins": 13.124004364013672,
|
| 1952 |
+
"rewards/rejected": -15.566947937011719,
|
| 1953 |
+
"step": 108
|
| 1954 |
+
},
|
| 1955 |
+
{
|
| 1956 |
+
"epoch": 0.20297951582867785,
|
| 1957 |
+
"grad_norm": 7.122522354125977,
|
| 1958 |
+
"learning_rate": 4.43064182194617e-06,
|
| 1959 |
+
"logps/chosen": -155.75596618652344,
|
| 1960 |
+
"logps/rejected": -314.00762939453125,
|
| 1961 |
+
"loss": 0.0363,
|
| 1962 |
+
"losses/dpo": 0.1575506329536438,
|
| 1963 |
+
"losses/sft": 2.296907424926758,
|
| 1964 |
+
"losses/total": 0.1575506329536438,
|
| 1965 |
+
"ref_logps/chosen": -143.9917755126953,
|
| 1966 |
+
"ref_logps/rejected": -175.87164306640625,
|
| 1967 |
+
"rewards/accuracies": 0.984375,
|
| 1968 |
+
"rewards/chosen": -1.1764178276062012,
|
| 1969 |
+
"rewards/margins": 12.637179374694824,
|
| 1970 |
+
"rewards/rejected": -13.813596725463867,
|
| 1971 |
+
"step": 109
|
| 1972 |
+
},
|
| 1973 |
+
{
|
| 1974 |
+
"epoch": 0.2048417132216015,
|
| 1975 |
+
"grad_norm": 1.012307047843933,
|
| 1976 |
+
"learning_rate": 4.4202898550724645e-06,
|
| 1977 |
+
"logps/chosen": -164.65115356445312,
|
| 1978 |
+
"logps/rejected": -318.34326171875,
|
| 1979 |
+
"loss": 0.0034,
|
| 1980 |
+
"losses/dpo": 0.011834606528282166,
|
| 1981 |
+
"losses/sft": 2.363070249557495,
|
| 1982 |
+
"losses/total": 0.011834606528282166,
|
| 1983 |
+
"ref_logps/chosen": -149.04367065429688,
|
| 1984 |
+
"ref_logps/rejected": -166.6707000732422,
|
| 1985 |
+
"rewards/accuracies": 1.0,
|
| 1986 |
+
"rewards/chosen": -1.5607484579086304,
|
| 1987 |
+
"rewards/margins": 13.606508255004883,
|
| 1988 |
+
"rewards/rejected": -15.167256355285645,
|
| 1989 |
+
"step": 110
|
| 1990 |
+
},
|
| 1991 |
+
{
|
| 1992 |
+
"epoch": 0.20670391061452514,
|
| 1993 |
+
"grad_norm": 5.557791233062744,
|
| 1994 |
+
"learning_rate": 4.409937888198758e-06,
|
| 1995 |
+
"logps/chosen": -150.85084533691406,
|
| 1996 |
+
"logps/rejected": -309.1524658203125,
|
| 1997 |
+
"loss": 0.0232,
|
| 1998 |
+
"losses/dpo": 0.014252948574721813,
|
| 1999 |
+
"losses/sft": 2.4658865928649902,
|
| 2000 |
+
"losses/total": 0.014252948574721813,
|
| 2001 |
+
"ref_logps/chosen": -137.03976440429688,
|
| 2002 |
+
"ref_logps/rejected": -161.1869659423828,
|
| 2003 |
+
"rewards/accuracies": 0.984375,
|
| 2004 |
+
"rewards/chosen": -1.381108045578003,
|
| 2005 |
+
"rewards/margins": 13.415441513061523,
|
| 2006 |
+
"rewards/rejected": -14.796548843383789,
|
| 2007 |
+
"step": 111
|
| 2008 |
+
},
|
| 2009 |
+
{
|
| 2010 |
+
"epoch": 0.2085661080074488,
|
| 2011 |
+
"grad_norm": 9.207598686218262,
|
| 2012 |
+
"learning_rate": 4.399585921325052e-06,
|
| 2013 |
+
"logps/chosen": -167.4930419921875,
|
| 2014 |
+
"logps/rejected": -303.5692138671875,
|
| 2015 |
+
"loss": 0.0453,
|
| 2016 |
+
"losses/dpo": 0.0007799149607308209,
|
| 2017 |
+
"losses/sft": 2.518667459487915,
|
| 2018 |
+
"losses/total": 0.0007799149607308209,
|
| 2019 |
+
"ref_logps/chosen": -149.21981811523438,
|
| 2020 |
+
"ref_logps/rejected": -170.2943115234375,
|
| 2021 |
+
"rewards/accuracies": 0.984375,
|
| 2022 |
+
"rewards/chosen": -1.8273215293884277,
|
| 2023 |
+
"rewards/margins": 11.500171661376953,
|
| 2024 |
+
"rewards/rejected": -13.327491760253906,
|
| 2025 |
+
"step": 112
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 0.21042830540037244,
|
| 2029 |
+
"grad_norm": 7.050368309020996,
|
| 2030 |
+
"learning_rate": 4.389233954451346e-06,
|
| 2031 |
+
"logps/chosen": -159.51071166992188,
|
| 2032 |
+
"logps/rejected": -307.4903869628906,
|
| 2033 |
+
"loss": 0.0301,
|
| 2034 |
+
"losses/dpo": 0.0039618429727852345,
|
| 2035 |
+
"losses/sft": 2.5831570625305176,
|
| 2036 |
+
"losses/total": 0.0039618429727852345,
|
| 2037 |
+
"ref_logps/chosen": -137.70912170410156,
|
| 2038 |
+
"ref_logps/rejected": -167.49789428710938,
|
| 2039 |
+
"rewards/accuracies": 0.984375,
|
| 2040 |
+
"rewards/chosen": -2.180156946182251,
|
| 2041 |
+
"rewards/margins": 11.819091796875,
|
| 2042 |
+
"rewards/rejected": -13.999248504638672,
|
| 2043 |
+
"step": 113
|
| 2044 |
+
},
|
| 2045 |
+
{
|
| 2046 |
+
"epoch": 0.2122905027932961,
|
| 2047 |
+
"grad_norm": 11.241766929626465,
|
| 2048 |
+
"learning_rate": 4.37888198757764e-06,
|
| 2049 |
+
"logps/chosen": -164.54208374023438,
|
| 2050 |
+
"logps/rejected": -306.73822021484375,
|
| 2051 |
+
"loss": 0.0491,
|
| 2052 |
+
"losses/dpo": 0.17672446370124817,
|
| 2053 |
+
"losses/sft": 2.2986695766448975,
|
| 2054 |
+
"losses/total": 0.17672446370124817,
|
| 2055 |
+
"ref_logps/chosen": -145.33531188964844,
|
| 2056 |
+
"ref_logps/rejected": -164.6031951904297,
|
| 2057 |
+
"rewards/accuracies": 0.984375,
|
| 2058 |
+
"rewards/chosen": -1.92067551612854,
|
| 2059 |
+
"rewards/margins": 12.292827606201172,
|
| 2060 |
+
"rewards/rejected": -14.213502883911133,
|
| 2061 |
+
"step": 114
|
| 2062 |
+
},
|
| 2063 |
+
{
|
| 2064 |
+
"epoch": 0.21415270018621974,
|
| 2065 |
+
"grad_norm": 4.248304843902588,
|
| 2066 |
+
"learning_rate": 4.368530020703934e-06,
|
| 2067 |
+
"logps/chosen": -160.98765563964844,
|
| 2068 |
+
"logps/rejected": -324.66845703125,
|
| 2069 |
+
"loss": 0.0157,
|
| 2070 |
+
"losses/dpo": 0.028906403109431267,
|
| 2071 |
+
"losses/sft": 2.51163387298584,
|
| 2072 |
+
"losses/total": 0.028906403109431267,
|
| 2073 |
+
"ref_logps/chosen": -142.23577880859375,
|
| 2074 |
+
"ref_logps/rejected": -178.46255493164062,
|
| 2075 |
+
"rewards/accuracies": 1.0,
|
| 2076 |
+
"rewards/chosen": -1.875186800956726,
|
| 2077 |
+
"rewards/margins": 12.745403289794922,
|
| 2078 |
+
"rewards/rejected": -14.620590209960938,
|
| 2079 |
+
"step": 115
|
| 2080 |
+
},
|
| 2081 |
+
{
|
| 2082 |
+
"epoch": 0.21601489757914338,
|
| 2083 |
+
"grad_norm": 11.712361335754395,
|
| 2084 |
+
"learning_rate": 4.358178053830228e-06,
|
| 2085 |
+
"logps/chosen": -176.92311096191406,
|
| 2086 |
+
"logps/rejected": -309.1515808105469,
|
| 2087 |
+
"loss": 0.0532,
|
| 2088 |
+
"losses/dpo": 0.0009635839378461242,
|
| 2089 |
+
"losses/sft": 2.519484758377075,
|
| 2090 |
+
"losses/total": 0.0009635839378461242,
|
| 2091 |
+
"ref_logps/chosen": -155.59686279296875,
|
| 2092 |
+
"ref_logps/rejected": -167.6163787841797,
|
| 2093 |
+
"rewards/accuracies": 0.984375,
|
| 2094 |
+
"rewards/chosen": -2.132624626159668,
|
| 2095 |
+
"rewards/margins": 12.020895004272461,
|
| 2096 |
+
"rewards/rejected": -14.153520584106445,
|
| 2097 |
+
"step": 116
|
| 2098 |
+
},
|
| 2099 |
+
{
|
| 2100 |
+
"epoch": 0.21787709497206703,
|
| 2101 |
+
"grad_norm": 12.869205474853516,
|
| 2102 |
+
"learning_rate": 4.347826086956522e-06,
|
| 2103 |
+
"logps/chosen": -164.499755859375,
|
| 2104 |
+
"logps/rejected": -322.2271728515625,
|
| 2105 |
+
"loss": 0.0481,
|
| 2106 |
+
"losses/dpo": 0.026883596554398537,
|
| 2107 |
+
"losses/sft": 2.4711999893188477,
|
| 2108 |
+
"losses/total": 0.026883596554398537,
|
| 2109 |
+
"ref_logps/chosen": -141.9335174560547,
|
| 2110 |
+
"ref_logps/rejected": -174.70437622070312,
|
| 2111 |
+
"rewards/accuracies": 0.984375,
|
| 2112 |
+
"rewards/chosen": -2.256622791290283,
|
| 2113 |
+
"rewards/margins": 12.495659828186035,
|
| 2114 |
+
"rewards/rejected": -14.75228214263916,
|
| 2115 |
+
"step": 117
|
| 2116 |
+
},
|
| 2117 |
+
{
|
| 2118 |
+
"epoch": 0.21973929236499068,
|
| 2119 |
+
"grad_norm": 7.2970499992370605,
|
| 2120 |
+
"learning_rate": 4.337474120082817e-06,
|
| 2121 |
+
"logps/chosen": -154.8804931640625,
|
| 2122 |
+
"logps/rejected": -307.47796630859375,
|
| 2123 |
+
"loss": 0.0217,
|
| 2124 |
+
"losses/dpo": 0.005061456002295017,
|
| 2125 |
+
"losses/sft": 2.7621004581451416,
|
| 2126 |
+
"losses/total": 0.005061456002295017,
|
| 2127 |
+
"ref_logps/chosen": -144.57373046875,
|
| 2128 |
+
"ref_logps/rejected": -164.53269958496094,
|
| 2129 |
+
"rewards/accuracies": 0.984375,
|
| 2130 |
+
"rewards/chosen": -1.0306763648986816,
|
| 2131 |
+
"rewards/margins": 13.263851165771484,
|
| 2132 |
+
"rewards/rejected": -14.294527053833008,
|
| 2133 |
+
"step": 118
|
| 2134 |
+
},
|
| 2135 |
+
{
|
| 2136 |
+
"epoch": 0.22160148975791433,
|
| 2137 |
+
"grad_norm": 12.324630737304688,
|
| 2138 |
+
"learning_rate": 4.32712215320911e-06,
|
| 2139 |
+
"logps/chosen": -169.20716857910156,
|
| 2140 |
+
"logps/rejected": -327.8011474609375,
|
| 2141 |
+
"loss": 0.0692,
|
| 2142 |
+
"losses/dpo": 0.05636470764875412,
|
| 2143 |
+
"losses/sft": 2.663618326187134,
|
| 2144 |
+
"losses/total": 0.05636470764875412,
|
| 2145 |
+
"ref_logps/chosen": -141.14559936523438,
|
| 2146 |
+
"ref_logps/rejected": -169.51641845703125,
|
| 2147 |
+
"rewards/accuracies": 0.96875,
|
| 2148 |
+
"rewards/chosen": -2.806156635284424,
|
| 2149 |
+
"rewards/margins": 13.022315979003906,
|
| 2150 |
+
"rewards/rejected": -15.828473091125488,
|
| 2151 |
+
"step": 119
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 0.22346368715083798,
|
| 2155 |
+
"grad_norm": 7.082119464874268,
|
| 2156 |
+
"learning_rate": 4.316770186335404e-06,
|
| 2157 |
+
"logps/chosen": -162.06655883789062,
|
| 2158 |
+
"logps/rejected": -312.4548645019531,
|
| 2159 |
+
"loss": 0.0211,
|
| 2160 |
+
"losses/dpo": 0.14828312397003174,
|
| 2161 |
+
"losses/sft": 2.6526529788970947,
|
| 2162 |
+
"losses/total": 0.14828312397003174,
|
| 2163 |
+
"ref_logps/chosen": -147.12899780273438,
|
| 2164 |
+
"ref_logps/rejected": -163.53634643554688,
|
| 2165 |
+
"rewards/accuracies": 0.984375,
|
| 2166 |
+
"rewards/chosen": -1.4937548637390137,
|
| 2167 |
+
"rewards/margins": 13.39809799194336,
|
| 2168 |
+
"rewards/rejected": -14.891853332519531,
|
| 2169 |
+
"step": 120
|
| 2170 |
+
},
|
| 2171 |
+
{
|
| 2172 |
+
"epoch": 0.22532588454376165,
|
| 2173 |
+
"grad_norm": 7.707842826843262,
|
| 2174 |
+
"learning_rate": 4.306418219461698e-06,
|
| 2175 |
+
"logps/chosen": -169.38192749023438,
|
| 2176 |
+
"logps/rejected": -323.21514892578125,
|
| 2177 |
+
"loss": 0.0198,
|
| 2178 |
+
"losses/dpo": 0.1473950445652008,
|
| 2179 |
+
"losses/sft": 2.6331777572631836,
|
| 2180 |
+
"losses/total": 0.1473950445652008,
|
| 2181 |
+
"ref_logps/chosen": -149.7891845703125,
|
| 2182 |
+
"ref_logps/rejected": -174.02752685546875,
|
| 2183 |
+
"rewards/accuracies": 0.984375,
|
| 2184 |
+
"rewards/chosen": -1.9592747688293457,
|
| 2185 |
+
"rewards/margins": 12.959487915039062,
|
| 2186 |
+
"rewards/rejected": -14.918761253356934,
|
| 2187 |
+
"step": 121
|
| 2188 |
+
},
|
| 2189 |
+
{
|
| 2190 |
+
"epoch": 0.2271880819366853,
|
| 2191 |
+
"grad_norm": 6.156428337097168,
|
| 2192 |
+
"learning_rate": 4.296066252587992e-06,
|
| 2193 |
+
"logps/chosen": -177.64190673828125,
|
| 2194 |
+
"logps/rejected": -320.0240478515625,
|
| 2195 |
+
"loss": 0.0313,
|
| 2196 |
+
"losses/dpo": 0.0023224614560604095,
|
| 2197 |
+
"losses/sft": 2.282505512237549,
|
| 2198 |
+
"losses/total": 0.0023224614560604095,
|
| 2199 |
+
"ref_logps/chosen": -150.99745178222656,
|
| 2200 |
+
"ref_logps/rejected": -166.0125732421875,
|
| 2201 |
+
"rewards/accuracies": 0.984375,
|
| 2202 |
+
"rewards/chosen": -2.664444923400879,
|
| 2203 |
+
"rewards/margins": 12.736701011657715,
|
| 2204 |
+
"rewards/rejected": -15.40114688873291,
|
| 2205 |
+
"step": 122
|
| 2206 |
+
},
|
| 2207 |
+
{
|
| 2208 |
+
"epoch": 0.22905027932960895,
|
| 2209 |
+
"grad_norm": 12.972006797790527,
|
| 2210 |
+
"learning_rate": 4.2857142857142855e-06,
|
| 2211 |
+
"logps/chosen": -171.61099243164062,
|
| 2212 |
+
"logps/rejected": -314.13824462890625,
|
| 2213 |
+
"loss": 0.204,
|
| 2214 |
+
"losses/dpo": 0.11684274673461914,
|
| 2215 |
+
"losses/sft": 2.5414156913757324,
|
| 2216 |
+
"losses/total": 0.11684274673461914,
|
| 2217 |
+
"ref_logps/chosen": -152.2470703125,
|
| 2218 |
+
"ref_logps/rejected": -169.70111083984375,
|
| 2219 |
+
"rewards/accuracies": 0.953125,
|
| 2220 |
+
"rewards/chosen": -1.936392903327942,
|
| 2221 |
+
"rewards/margins": 12.507322311401367,
|
| 2222 |
+
"rewards/rejected": -14.443716049194336,
|
| 2223 |
+
"step": 123
|
| 2224 |
+
},
|
| 2225 |
+
{
|
| 2226 |
+
"epoch": 0.2309124767225326,
|
| 2227 |
+
"grad_norm": 8.865942001342773,
|
| 2228 |
+
"learning_rate": 4.27536231884058e-06,
|
| 2229 |
+
"logps/chosen": -178.2666015625,
|
| 2230 |
+
"logps/rejected": -294.51971435546875,
|
| 2231 |
+
"loss": 0.0887,
|
| 2232 |
+
"losses/dpo": 0.04358779639005661,
|
| 2233 |
+
"losses/sft": 2.905651569366455,
|
| 2234 |
+
"losses/total": 0.04358779639005661,
|
| 2235 |
+
"ref_logps/chosen": -146.95541381835938,
|
| 2236 |
+
"ref_logps/rejected": -152.81776428222656,
|
| 2237 |
+
"rewards/accuracies": 0.984375,
|
| 2238 |
+
"rewards/chosen": -3.1311190128326416,
|
| 2239 |
+
"rewards/margins": 11.039077758789062,
|
| 2240 |
+
"rewards/rejected": -14.170197486877441,
|
| 2241 |
+
"step": 124
|
| 2242 |
+
},
|
| 2243 |
+
{
|
| 2244 |
+
"epoch": 0.23277467411545624,
|
| 2245 |
+
"grad_norm": 14.640778541564941,
|
| 2246 |
+
"learning_rate": 4.265010351966874e-06,
|
| 2247 |
+
"logps/chosen": -176.9116668701172,
|
| 2248 |
+
"logps/rejected": -332.41241455078125,
|
| 2249 |
+
"loss": 0.1211,
|
| 2250 |
+
"losses/dpo": 0.053478606045246124,
|
| 2251 |
+
"losses/sft": 2.7518341541290283,
|
| 2252 |
+
"losses/total": 0.053478606045246124,
|
| 2253 |
+
"ref_logps/chosen": -150.49269104003906,
|
| 2254 |
+
"ref_logps/rejected": -171.48587036132812,
|
| 2255 |
+
"rewards/accuracies": 0.984375,
|
| 2256 |
+
"rewards/chosen": -2.641897678375244,
|
| 2257 |
+
"rewards/margins": 13.450759887695312,
|
| 2258 |
+
"rewards/rejected": -16.0926570892334,
|
| 2259 |
+
"step": 125
|
| 2260 |
+
}
|
| 2261 |
+
],
|
| 2262 |
+
"logging_steps": 1.0,
|
| 2263 |
+
"max_steps": 537,
|
| 2264 |
+
"num_input_tokens_seen": 0,
|
| 2265 |
+
"num_train_epochs": 1,
|
| 2266 |
+
"save_steps": 25,
|
| 2267 |
+
"stateful_callbacks": {
|
| 2268 |
+
"TrainerControl": {
|
| 2269 |
+
"args": {
|
| 2270 |
+
"should_epoch_stop": false,
|
| 2271 |
+
"should_evaluate": false,
|
| 2272 |
+
"should_log": false,
|
| 2273 |
+
"should_save": true,
|
| 2274 |
+
"should_training_stop": false
|
| 2275 |
+
},
|
| 2276 |
+
"attributes": {}
|
| 2277 |
+
}
|
| 2278 |
+
},
|
| 2279 |
+
"total_flos": 0.0,
|
| 2280 |
+
"train_batch_size": 2,
|
| 2281 |
+
"trial_name": null,
|
| 2282 |
+
"trial_params": null
|
| 2283 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:494f542b170a80f31e9196cf525539951715f4482c3206039742a9432cebf311
|
| 3 |
+
size 6904
|
vocab.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|