Upload continuous evaluation checkpoints, batch 32
Browse files- continuous/smollm2/mc1_b16_224_2.5b/pretrain/trainer_state.json +421 -0
- continuous/smollm2/mc1_b16_224_400m/finetune/chat_template.jinja +6 -0
- continuous/smollm2/mc1_b16_224_400m/finetune/config.json +70 -0
- continuous/smollm2/mc1_b16_224_400m/finetune/generation_config.json +9 -0
- continuous/smollm2/mc1_b16_224_400m/finetune/merges.txt +0 -0
- continuous/smollm2/mc1_b16_224_400m/finetune/model.safetensors +3 -0
- continuous/smollm2/mc1_b16_224_400m/finetune/special_tokens_map.json +28 -0
- continuous/smollm2/mc1_b16_224_400m/finetune/tokenizer.json +0 -0
- continuous/smollm2/mc1_b16_224_400m/finetune/tokenizer_config.json +155 -0
- continuous/smollm2/mc1_b16_224_400m/finetune/trainer_state.json +1856 -0
- continuous/smollm2/mc1_b16_224_400m/finetune/training_args.bin +3 -0
- continuous/smollm2/mc1_b16_224_400m/finetune/vocab.json +0 -0
- continuous/smollm2/mc1_b16_224_400m/pretrain/config.json +70 -0
- continuous/smollm2/mc1_b16_224_400m/pretrain/mm_projector.bin +3 -0
- continuous/smollm2/mc1_b16_224_400m/pretrain/trainer_state.json +421 -0
- continuous/smollm2/mc1_b32_224_2.5b/finetune/chat_template.jinja +6 -0
- continuous/smollm2/mc1_b32_224_2.5b/finetune/config.json +70 -0
- continuous/smollm2/mc1_b32_224_2.5b/finetune/generation_config.json +9 -0
- continuous/smollm2/mc1_b32_224_2.5b/finetune/merges.txt +0 -0
- continuous/smollm2/mc1_b32_224_2.5b/finetune/model.safetensors +3 -0
- continuous/smollm2/mc1_b32_224_2.5b/finetune/special_tokens_map.json +28 -0
- continuous/smollm2/mc1_b32_224_2.5b/finetune/tokenizer.json +0 -0
- continuous/smollm2/mc1_b32_224_2.5b/finetune/tokenizer_config.json +155 -0
- continuous/smollm2/mc1_b32_224_2.5b/finetune/trainer_state.json +1856 -0
- continuous/smollm2/mc1_b32_224_2.5b/finetune/training_args.bin +3 -0
- continuous/smollm2/mc1_b32_224_2.5b/finetune/vocab.json +0 -0
- continuous/smollm2/mc1_b32_224_2.5b/pretrain/config.json +70 -0
- continuous/smollm2/mc1_b32_224_2.5b/pretrain/mm_projector.bin +3 -0
- continuous/smollm2/mc1_b32_224_2.5b/pretrain/trainer_state.json +421 -0
- continuous/smollm2/mc1_b32_224_400m/finetune/chat_template.jinja +6 -0
- continuous/smollm2/mc1_b32_224_400m/finetune/config.json +70 -0
- continuous/smollm2/mc1_b32_224_400m/finetune/generation_config.json +9 -0
- continuous/smollm2/mc1_b32_224_400m/finetune/merges.txt +0 -0
- continuous/smollm2/mc1_b32_224_400m/finetune/model.safetensors +3 -0
- continuous/smollm2/mc1_b32_224_400m/finetune/special_tokens_map.json +28 -0
- continuous/smollm2/mc1_b32_224_400m/finetune/tokenizer.json +0 -0
- continuous/smollm2/mc1_b32_224_400m/finetune/tokenizer_config.json +155 -0
- continuous/smollm2/mc1_b32_224_400m/finetune/trainer_state.json +1856 -0
- continuous/smollm2/mc1_b32_224_400m/finetune/training_args.bin +3 -0
- continuous/smollm2/mc1_b32_224_400m/finetune/vocab.json +0 -0
- continuous/smollm2/mc1_b32_224_400m/pretrain/config.json +70 -0
- continuous/smollm2/mc1_b32_224_400m/pretrain/mm_projector.bin +3 -0
- continuous/smollm2/mc1_b32_224_400m/pretrain/trainer_state.json +421 -0
- continuous/smollm2/mc1_g14_224_2.5b/finetune/chat_template.jinja +6 -0
- continuous/smollm2/mc1_g14_224_2.5b/finetune/config.json +70 -0
- continuous/smollm2/mc1_g14_224_2.5b/finetune/generation_config.json +9 -0
- continuous/smollm2/mc1_g14_224_2.5b/finetune/merges.txt +0 -0
- continuous/smollm2/mc1_g14_224_2.5b/finetune/model-00001-of-00002.safetensors +3 -0
continuous/smollm2/mc1_b16_224_2.5b/pretrain/trainer_state.json
ADDED
|
@@ -0,0 +1,421 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 1.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 545,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.01834862385321101,
|
| 14 |
+
"grad_norm": 1.706994891166687,
|
| 15 |
+
"learning_rate": 0.00032142857142857147,
|
| 16 |
+
"loss": 3.5928,
|
| 17 |
+
"step": 10
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.03669724770642202,
|
| 21 |
+
"grad_norm": 0.32562363147735596,
|
| 22 |
+
"learning_rate": 0.0006785714285714287,
|
| 23 |
+
"loss": 2.9263,
|
| 24 |
+
"step": 20
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.05504587155963303,
|
| 28 |
+
"grad_norm": 0.12782445549964905,
|
| 29 |
+
"learning_rate": 0.0009999907688176173,
|
| 30 |
+
"loss": 2.8126,
|
| 31 |
+
"step": 30
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.07339449541284404,
|
| 35 |
+
"grad_norm": 0.08279376477003098,
|
| 36 |
+
"learning_rate": 0.0009988834393115766,
|
| 37 |
+
"loss": 2.7509,
|
| 38 |
+
"step": 40
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.09174311926605505,
|
| 42 |
+
"grad_norm": 0.10916870832443237,
|
| 43 |
+
"learning_rate": 0.0009959345572940772,
|
| 44 |
+
"loss": 2.7258,
|
| 45 |
+
"step": 50
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.11009174311926606,
|
| 49 |
+
"grad_norm": 0.14105260372161865,
|
| 50 |
+
"learning_rate": 0.0009911550081155983,
|
| 51 |
+
"loss": 2.6767,
|
| 52 |
+
"step": 60
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.12844036697247707,
|
| 56 |
+
"grad_norm": 0.12183316797018051,
|
| 57 |
+
"learning_rate": 0.0009845624347566433,
|
| 58 |
+
"loss": 2.6279,
|
| 59 |
+
"step": 70
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.14678899082568808,
|
| 63 |
+
"grad_norm": 0.10005690157413483,
|
| 64 |
+
"learning_rate": 0.0009761811727013547,
|
| 65 |
+
"loss": 2.5774,
|
| 66 |
+
"step": 80
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.1651376146788991,
|
| 70 |
+
"grad_norm": 0.08537808060646057,
|
| 71 |
+
"learning_rate": 0.0009660421601067666,
|
| 72 |
+
"loss": 2.5354,
|
| 73 |
+
"step": 90
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.1834862385321101,
|
| 77 |
+
"grad_norm": 0.17895951867103577,
|
| 78 |
+
"learning_rate": 0.0009541828235992926,
|
| 79 |
+
"loss": 2.5078,
|
| 80 |
+
"step": 100
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.2018348623853211,
|
| 84 |
+
"grad_norm": 0.12853018939495087,
|
| 85 |
+
"learning_rate": 0.0009406469401200151,
|
| 86 |
+
"loss": 2.4713,
|
| 87 |
+
"step": 110
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.22018348623853212,
|
| 91 |
+
"grad_norm": 0.13017429411411285,
|
| 92 |
+
"learning_rate": 0.0009254844753287492,
|
| 93 |
+
"loss": 2.44,
|
| 94 |
+
"step": 120
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.23853211009174313,
|
| 98 |
+
"grad_norm": 0.15240995585918427,
|
| 99 |
+
"learning_rate": 0.0009087513991633924,
|
| 100 |
+
"loss": 2.3948,
|
| 101 |
+
"step": 130
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.25688073394495414,
|
| 105 |
+
"grad_norm": 0.12597467005252838,
|
| 106 |
+
"learning_rate": 0.0008905094792353917,
|
| 107 |
+
"loss": 2.3593,
|
| 108 |
+
"step": 140
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.27522935779816515,
|
| 112 |
+
"grad_norm": 0.13725033402442932,
|
| 113 |
+
"learning_rate": 0.0008708260528239789,
|
| 114 |
+
"loss": 2.3478,
|
| 115 |
+
"step": 150
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.29357798165137616,
|
| 119 |
+
"grad_norm": 0.10004206001758575,
|
| 120 |
+
"learning_rate": 0.0008497737783108237,
|
| 121 |
+
"loss": 2.3206,
|
| 122 |
+
"step": 160
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.3119266055045872,
|
| 126 |
+
"grad_norm": 0.12458819895982742,
|
| 127 |
+
"learning_rate": 0.0008274303669726426,
|
| 128 |
+
"loss": 2.3113,
|
| 129 |
+
"step": 170
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.3302752293577982,
|
| 133 |
+
"grad_norm": 0.17835143208503723,
|
| 134 |
+
"learning_rate": 0.0008038782961218137,
|
| 135 |
+
"loss": 2.304,
|
| 136 |
+
"step": 180
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.3486238532110092,
|
| 140 |
+
"grad_norm": 0.10673367232084274,
|
| 141 |
+
"learning_rate": 0.000779204504653894,
|
| 142 |
+
"loss": 2.293,
|
| 143 |
+
"step": 190
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.3669724770642202,
|
| 147 |
+
"grad_norm": 0.1339321732521057,
|
| 148 |
+
"learning_rate": 0.0007535000721258791,
|
| 149 |
+
"loss": 2.2665,
|
| 150 |
+
"step": 200
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.3853211009174312,
|
| 154 |
+
"grad_norm": 0.1266634464263916,
|
| 155 |
+
"learning_rate": 0.00072685988254984,
|
| 156 |
+
"loss": 2.2647,
|
| 157 |
+
"step": 210
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.4036697247706422,
|
| 161 |
+
"grad_norm": 0.10726504027843475,
|
| 162 |
+
"learning_rate": 0.0006993822741429906,
|
| 163 |
+
"loss": 2.2709,
|
| 164 |
+
"step": 220
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.42201834862385323,
|
| 168 |
+
"grad_norm": 0.10359807312488556,
|
| 169 |
+
"learning_rate": 0.0006711686763270818,
|
| 170 |
+
"loss": 2.2602,
|
| 171 |
+
"step": 230
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.44036697247706424,
|
| 175 |
+
"grad_norm": 0.10765370726585388,
|
| 176 |
+
"learning_rate": 0.0006423232353170831,
|
| 177 |
+
"loss": 2.2507,
|
| 178 |
+
"step": 240
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.45871559633027525,
|
| 182 |
+
"grad_norm": 0.11113200336694717,
|
| 183 |
+
"learning_rate": 0.0006129524296812335,
|
| 184 |
+
"loss": 2.2522,
|
| 185 |
+
"step": 250
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.47706422018348627,
|
| 189 |
+
"grad_norm": 0.10349487513303757,
|
| 190 |
+
"learning_rate": 0.0005831646772915651,
|
| 191 |
+
"loss": 2.2564,
|
| 192 |
+
"step": 260
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.4954128440366973,
|
| 196 |
+
"grad_norm": 0.10266552120447159,
|
| 197 |
+
"learning_rate": 0.0005530699351157809,
|
| 198 |
+
"loss": 2.2345,
|
| 199 |
+
"step": 270
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.5137614678899083,
|
| 203 |
+
"grad_norm": 0.09799429774284363,
|
| 204 |
+
"learning_rate": 0.0005227792933277943,
|
| 205 |
+
"loss": 2.238,
|
| 206 |
+
"step": 280
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.5321100917431193,
|
| 210 |
+
"grad_norm": 0.0955471321940422,
|
| 211 |
+
"learning_rate": 0.0004924045652352092,
|
| 212 |
+
"loss": 2.2257,
|
| 213 |
+
"step": 290
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.5504587155963303,
|
| 217 |
+
"grad_norm": 0.08558019250631332,
|
| 218 |
+
"learning_rate": 0.0004620578745374633,
|
| 219 |
+
"loss": 2.2279,
|
| 220 |
+
"step": 300
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.5688073394495413,
|
| 224 |
+
"grad_norm": 0.12679877877235413,
|
| 225 |
+
"learning_rate": 0.0004318512414382059,
|
| 226 |
+
"loss": 2.2301,
|
| 227 |
+
"step": 310
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.5871559633027523,
|
| 231 |
+
"grad_norm": 0.0962977483868599,
|
| 232 |
+
"learning_rate": 0.0004018961691397148,
|
| 233 |
+
"loss": 2.2303,
|
| 234 |
+
"step": 320
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.6055045871559633,
|
| 238 |
+
"grad_norm": 0.09710817039012909,
|
| 239 |
+
"learning_rate": 0.0003723032322457458,
|
| 240 |
+
"loss": 2.2176,
|
| 241 |
+
"step": 330
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.6238532110091743,
|
| 245 |
+
"grad_norm": 0.12647433578968048,
|
| 246 |
+
"learning_rate": 0.0003431816685921625,
|
| 247 |
+
"loss": 2.2199,
|
| 248 |
+
"step": 340
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.6422018348623854,
|
| 252 |
+
"grad_norm": 0.11681321263313293,
|
| 253 |
+
"learning_rate": 0.0003146389760120416,
|
| 254 |
+
"loss": 2.2196,
|
| 255 |
+
"step": 350
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.6605504587155964,
|
| 259 |
+
"grad_norm": 0.10799563676118851,
|
| 260 |
+
"learning_rate": 0.00028678051552373485,
|
| 261 |
+
"loss": 2.2199,
|
| 262 |
+
"step": 360
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.6788990825688074,
|
| 266 |
+
"grad_norm": 0.09262187778949738,
|
| 267 |
+
"learning_rate": 0.0002597091224066581,
|
| 268 |
+
"loss": 2.2153,
|
| 269 |
+
"step": 370
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.6972477064220184,
|
| 273 |
+
"grad_norm": 0.09784838557243347,
|
| 274 |
+
"learning_rate": 0.0002335247266004629,
|
| 275 |
+
"loss": 2.2203,
|
| 276 |
+
"step": 380
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.7155963302752294,
|
| 280 |
+
"grad_norm": 0.0965687707066536,
|
| 281 |
+
"learning_rate": 0.000208323983828828,
|
| 282 |
+
"loss": 2.2107,
|
| 283 |
+
"step": 390
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.7339449541284404,
|
| 287 |
+
"grad_norm": 0.09708311408758163,
|
| 288 |
+
"learning_rate": 0.0001841999188095224,
|
| 289 |
+
"loss": 2.2056,
|
| 290 |
+
"step": 400
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.7522935779816514,
|
| 294 |
+
"grad_norm": 0.10050690919160843,
|
| 295 |
+
"learning_rate": 0.00016124158186777676,
|
| 296 |
+
"loss": 2.2084,
|
| 297 |
+
"step": 410
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.7706422018348624,
|
| 301 |
+
"grad_norm": 0.09254024922847748,
|
| 302 |
+
"learning_rate": 0.0001395337202205211,
|
| 303 |
+
"loss": 2.2059,
|
| 304 |
+
"step": 420
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 0.7889908256880734,
|
| 308 |
+
"grad_norm": 0.09977176040410995,
|
| 309 |
+
"learning_rate": 0.00011915646514489292,
|
| 310 |
+
"loss": 2.2044,
|
| 311 |
+
"step": 430
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.8073394495412844,
|
| 315 |
+
"grad_norm": 0.10293921828269958,
|
| 316 |
+
"learning_rate": 0.00010018503618578818,
|
| 317 |
+
"loss": 2.2126,
|
| 318 |
+
"step": 440
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.8256880733944955,
|
| 322 |
+
"grad_norm": 0.08376018702983856,
|
| 323 |
+
"learning_rate": 8.268946349432583e-05,
|
| 324 |
+
"loss": 2.2055,
|
| 325 |
+
"step": 450
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.8440366972477065,
|
| 329 |
+
"grad_norm": 0.09050320088863373,
|
| 330 |
+
"learning_rate": 6.673432932217117e-05,
|
| 331 |
+
"loss": 2.1972,
|
| 332 |
+
"step": 460
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.8623853211009175,
|
| 336 |
+
"grad_norm": 0.08690407127141953,
|
| 337 |
+
"learning_rate": 5.2378529625954684e-05,
|
| 338 |
+
"loss": 2.1919,
|
| 339 |
+
"step": 470
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.8807339449541285,
|
| 343 |
+
"grad_norm": 0.08345736563205719,
|
| 344 |
+
"learning_rate": 3.9675056661785556e-05,
|
| 345 |
+
"loss": 2.195,
|
| 346 |
+
"step": 480
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.8990825688073395,
|
| 350 |
+
"grad_norm": 0.0854126363992691,
|
| 351 |
+
"learning_rate": 2.8670803372379538e-05,
|
| 352 |
+
"loss": 2.2011,
|
| 353 |
+
"step": 490
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.9174311926605505,
|
| 357 |
+
"grad_norm": 0.08284084498882294,
|
| 358 |
+
"learning_rate": 1.9406390288876586e-05,
|
| 359 |
+
"loss": 2.2051,
|
| 360 |
+
"step": 500
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.9357798165137615,
|
| 364 |
+
"grad_norm": 0.08117683231830597,
|
| 365 |
+
"learning_rate": 1.191601558631461e-05,
|
| 366 |
+
"loss": 2.1969,
|
| 367 |
+
"step": 510
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.9541284403669725,
|
| 371 |
+
"grad_norm": 0.08256425708532333,
|
| 372 |
+
"learning_rate": 6.2273288462559305e-06,
|
| 373 |
+
"loss": 2.1946,
|
| 374 |
+
"step": 520
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.9724770642201835,
|
| 378 |
+
"grad_norm": 0.08296385407447815,
|
| 379 |
+
"learning_rate": 2.361328992552314e-06,
|
| 380 |
+
"loss": 2.1971,
|
| 381 |
+
"step": 530
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.9908256880733946,
|
| 385 |
+
"grad_norm": 0.0795731395483017,
|
| 386 |
+
"learning_rate": 3.322867770044202e-07,
|
| 387 |
+
"loss": 2.1946,
|
| 388 |
+
"step": 540
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 1.0,
|
| 392 |
+
"step": 545,
|
| 393 |
+
"total_flos": 3.540700722154701e+17,
|
| 394 |
+
"train_loss": 2.3456867305510634,
|
| 395 |
+
"train_runtime": 1390.6602,
|
| 396 |
+
"train_samples_per_second": 401.34,
|
| 397 |
+
"train_steps_per_second": 0.392
|
| 398 |
+
}
|
| 399 |
+
],
|
| 400 |
+
"logging_steps": 10,
|
| 401 |
+
"max_steps": 545,
|
| 402 |
+
"num_input_tokens_seen": 34658176,
|
| 403 |
+
"num_train_epochs": 1,
|
| 404 |
+
"save_steps": 500,
|
| 405 |
+
"stateful_callbacks": {
|
| 406 |
+
"TrainerControl": {
|
| 407 |
+
"args": {
|
| 408 |
+
"should_epoch_stop": false,
|
| 409 |
+
"should_evaluate": false,
|
| 410 |
+
"should_log": false,
|
| 411 |
+
"should_save": false,
|
| 412 |
+
"should_training_stop": false
|
| 413 |
+
},
|
| 414 |
+
"attributes": {}
|
| 415 |
+
}
|
| 416 |
+
},
|
| 417 |
+
"total_flos": 3.540700722154701e+17,
|
| 418 |
+
"train_batch_size": 64,
|
| 419 |
+
"trial_name": null,
|
| 420 |
+
"trial_params": null
|
| 421 |
+
}
|
continuous/smollm2/mc1_b16_224_400m/finetune/chat_template.jinja
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{% for message in messages %}{% if loop.first and messages[0]['role'] != 'system' %}{{ '<|im_start|>system
|
| 2 |
+
You are a helpful AI assistant named SmolLM, trained by Hugging Face<|im_end|>
|
| 3 |
+
' }}{% endif %}{{'<|im_start|>' + message['role'] + '
|
| 4 |
+
' + message['content'] + '<|im_end|>' + '
|
| 5 |
+
'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant
|
| 6 |
+
' }}{% endif %}
|
continuous/smollm2/mc1_b16_224_400m/finetune/config.json
ADDED
|
@@ -0,0 +1,70 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_faster_video": false,
|
| 3 |
+
"add_time_instruction": false,
|
| 4 |
+
"architectures": [
|
| 5 |
+
"LlavaLlamaForCausalLM"
|
| 6 |
+
],
|
| 7 |
+
"attention_bias": false,
|
| 8 |
+
"attention_dropout": 0.0,
|
| 9 |
+
"bos_token_id": 1,
|
| 10 |
+
"dtype": "bfloat16",
|
| 11 |
+
"eos_token_id": 2,
|
| 12 |
+
"faster_token_stride": 10,
|
| 13 |
+
"force_sample": false,
|
| 14 |
+
"head_dim": 64,
|
| 15 |
+
"hidden_act": "silu",
|
| 16 |
+
"hidden_size": 2048,
|
| 17 |
+
"image_aspect_ratio": "square",
|
| 18 |
+
"image_crop_resolution": null,
|
| 19 |
+
"image_grid_pinpoints": null,
|
| 20 |
+
"image_split_resolution": null,
|
| 21 |
+
"initializer_range": 0.02,
|
| 22 |
+
"intermediate_size": 8192,
|
| 23 |
+
"max_position_embeddings": 8192,
|
| 24 |
+
"mlp_bias": false,
|
| 25 |
+
"mm_hidden_size": 768,
|
| 26 |
+
"mm_newline_position": "grid",
|
| 27 |
+
"mm_patch_merge_type": "flat",
|
| 28 |
+
"mm_projector_lr": 0.0001,
|
| 29 |
+
"mm_projector_type": "mlp2x_gelu",
|
| 30 |
+
"mm_resampler_type": null,
|
| 31 |
+
"mm_spatial_pool_mode": "bilinear",
|
| 32 |
+
"mm_spatial_pool_stride": null,
|
| 33 |
+
"mm_tunable_parts": "mm_mlp_adapter,mm_language_model",
|
| 34 |
+
"mm_use_im_patch_token": false,
|
| 35 |
+
"mm_use_im_start_end": false,
|
| 36 |
+
"mm_vision_select_feature": "patch",
|
| 37 |
+
"mm_vision_select_layer": -2,
|
| 38 |
+
"mm_vision_tower": "open_clip_hub:ViT-B-16",
|
| 39 |
+
"mm_vision_tower_lr": null,
|
| 40 |
+
"model_type": "llama",
|
| 41 |
+
"num_attention_heads": 32,
|
| 42 |
+
"num_hidden_layers": 24,
|
| 43 |
+
"num_key_value_heads": 32,
|
| 44 |
+
"pad_token_id": 0,
|
| 45 |
+
"pos_skipping_range": 4096,
|
| 46 |
+
"pretraining_tp": 1,
|
| 47 |
+
"rms_norm_eps": 1e-05,
|
| 48 |
+
"rope_scaling": null,
|
| 49 |
+
"rope_theta": 130000,
|
| 50 |
+
"tie_word_embeddings": true,
|
| 51 |
+
"tokenizer_model_max_length": 8192,
|
| 52 |
+
"tokenizer_padding_side": "right",
|
| 53 |
+
"transformers.js_config": {
|
| 54 |
+
"dtype": "q4",
|
| 55 |
+
"kv_cache_dtype": {
|
| 56 |
+
"fp16": "float16",
|
| 57 |
+
"q4f16": "float16"
|
| 58 |
+
},
|
| 59 |
+
"use_external_data_format": {
|
| 60 |
+
"model.onnx": true,
|
| 61 |
+
"model_fp16.onnx": true
|
| 62 |
+
}
|
| 63 |
+
},
|
| 64 |
+
"transformers_version": "4.57.6",
|
| 65 |
+
"use_cache": true,
|
| 66 |
+
"use_mm_proj": true,
|
| 67 |
+
"use_pos_skipping": false,
|
| 68 |
+
"vision_tower_pretrained": "/cache/ckpt/download/tokenizer/continuous/mc1/mc1_b16_224_400m.pt",
|
| 69 |
+
"vocab_size": 49152
|
| 70 |
+
}
|
continuous/smollm2/mc1_b16_224_400m/finetune/generation_config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 1,
|
| 4 |
+
"eos_token_id": [
|
| 5 |
+
2
|
| 6 |
+
],
|
| 7 |
+
"pad_token_id": 0,
|
| 8 |
+
"transformers_version": "4.57.6"
|
| 9 |
+
}
|
continuous/smollm2/mc1_b16_224_400m/finetune/merges.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
continuous/smollm2/mc1_b16_224_400m/finetune/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1732c659d05eab9f6d05b881ff25b4ac6a4716b32d8f17bfb6bd6e22fc16ed4c
|
| 3 |
+
size 3819597280
|
continuous/smollm2/mc1_b16_224_400m/finetune/special_tokens_map.json
ADDED
|
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"additional_special_tokens": [
|
| 3 |
+
"<|im_start|>",
|
| 4 |
+
"<|im_end|>"
|
| 5 |
+
],
|
| 6 |
+
"bos_token": {
|
| 7 |
+
"content": "<|im_start|>",
|
| 8 |
+
"lstrip": false,
|
| 9 |
+
"normalized": false,
|
| 10 |
+
"rstrip": false,
|
| 11 |
+
"single_word": false
|
| 12 |
+
},
|
| 13 |
+
"eos_token": {
|
| 14 |
+
"content": "<|im_end|>",
|
| 15 |
+
"lstrip": false,
|
| 16 |
+
"normalized": false,
|
| 17 |
+
"rstrip": false,
|
| 18 |
+
"single_word": false
|
| 19 |
+
},
|
| 20 |
+
"pad_token": "<|endoftext|>",
|
| 21 |
+
"unk_token": {
|
| 22 |
+
"content": "<|endoftext|>",
|
| 23 |
+
"lstrip": false,
|
| 24 |
+
"normalized": false,
|
| 25 |
+
"rstrip": false,
|
| 26 |
+
"single_word": false
|
| 27 |
+
}
|
| 28 |
+
}
|
continuous/smollm2/mc1_b16_224_400m/finetune/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
continuous/smollm2/mc1_b16_224_400m/finetune/tokenizer_config.json
ADDED
|
@@ -0,0 +1,155 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_prefix_space": false,
|
| 3 |
+
"added_tokens_decoder": {
|
| 4 |
+
"0": {
|
| 5 |
+
"content": "<|endoftext|>",
|
| 6 |
+
"lstrip": false,
|
| 7 |
+
"normalized": false,
|
| 8 |
+
"rstrip": false,
|
| 9 |
+
"single_word": false,
|
| 10 |
+
"special": true
|
| 11 |
+
},
|
| 12 |
+
"1": {
|
| 13 |
+
"content": "<|im_start|>",
|
| 14 |
+
"lstrip": false,
|
| 15 |
+
"normalized": false,
|
| 16 |
+
"rstrip": false,
|
| 17 |
+
"single_word": false,
|
| 18 |
+
"special": true
|
| 19 |
+
},
|
| 20 |
+
"2": {
|
| 21 |
+
"content": "<|im_end|>",
|
| 22 |
+
"lstrip": false,
|
| 23 |
+
"normalized": false,
|
| 24 |
+
"rstrip": false,
|
| 25 |
+
"single_word": false,
|
| 26 |
+
"special": true
|
| 27 |
+
},
|
| 28 |
+
"3": {
|
| 29 |
+
"content": "<repo_name>",
|
| 30 |
+
"lstrip": false,
|
| 31 |
+
"normalized": false,
|
| 32 |
+
"rstrip": false,
|
| 33 |
+
"single_word": false,
|
| 34 |
+
"special": true
|
| 35 |
+
},
|
| 36 |
+
"4": {
|
| 37 |
+
"content": "<reponame>",
|
| 38 |
+
"lstrip": false,
|
| 39 |
+
"normalized": false,
|
| 40 |
+
"rstrip": false,
|
| 41 |
+
"single_word": false,
|
| 42 |
+
"special": true
|
| 43 |
+
},
|
| 44 |
+
"5": {
|
| 45 |
+
"content": "<file_sep>",
|
| 46 |
+
"lstrip": false,
|
| 47 |
+
"normalized": false,
|
| 48 |
+
"rstrip": false,
|
| 49 |
+
"single_word": false,
|
| 50 |
+
"special": true
|
| 51 |
+
},
|
| 52 |
+
"6": {
|
| 53 |
+
"content": "<filename>",
|
| 54 |
+
"lstrip": false,
|
| 55 |
+
"normalized": false,
|
| 56 |
+
"rstrip": false,
|
| 57 |
+
"single_word": false,
|
| 58 |
+
"special": true
|
| 59 |
+
},
|
| 60 |
+
"7": {
|
| 61 |
+
"content": "<gh_stars>",
|
| 62 |
+
"lstrip": false,
|
| 63 |
+
"normalized": false,
|
| 64 |
+
"rstrip": false,
|
| 65 |
+
"single_word": false,
|
| 66 |
+
"special": true
|
| 67 |
+
},
|
| 68 |
+
"8": {
|
| 69 |
+
"content": "<issue_start>",
|
| 70 |
+
"lstrip": false,
|
| 71 |
+
"normalized": false,
|
| 72 |
+
"rstrip": false,
|
| 73 |
+
"single_word": false,
|
| 74 |
+
"special": true
|
| 75 |
+
},
|
| 76 |
+
"9": {
|
| 77 |
+
"content": "<issue_comment>",
|
| 78 |
+
"lstrip": false,
|
| 79 |
+
"normalized": false,
|
| 80 |
+
"rstrip": false,
|
| 81 |
+
"single_word": false,
|
| 82 |
+
"special": true
|
| 83 |
+
},
|
| 84 |
+
"10": {
|
| 85 |
+
"content": "<issue_closed>",
|
| 86 |
+
"lstrip": false,
|
| 87 |
+
"normalized": false,
|
| 88 |
+
"rstrip": false,
|
| 89 |
+
"single_word": false,
|
| 90 |
+
"special": true
|
| 91 |
+
},
|
| 92 |
+
"11": {
|
| 93 |
+
"content": "<jupyter_start>",
|
| 94 |
+
"lstrip": false,
|
| 95 |
+
"normalized": false,
|
| 96 |
+
"rstrip": false,
|
| 97 |
+
"single_word": false,
|
| 98 |
+
"special": true
|
| 99 |
+
},
|
| 100 |
+
"12": {
|
| 101 |
+
"content": "<jupyter_text>",
|
| 102 |
+
"lstrip": false,
|
| 103 |
+
"normalized": false,
|
| 104 |
+
"rstrip": false,
|
| 105 |
+
"single_word": false,
|
| 106 |
+
"special": true
|
| 107 |
+
},
|
| 108 |
+
"13": {
|
| 109 |
+
"content": "<jupyter_code>",
|
| 110 |
+
"lstrip": false,
|
| 111 |
+
"normalized": false,
|
| 112 |
+
"rstrip": false,
|
| 113 |
+
"single_word": false,
|
| 114 |
+
"special": true
|
| 115 |
+
},
|
| 116 |
+
"14": {
|
| 117 |
+
"content": "<jupyter_output>",
|
| 118 |
+
"lstrip": false,
|
| 119 |
+
"normalized": false,
|
| 120 |
+
"rstrip": false,
|
| 121 |
+
"single_word": false,
|
| 122 |
+
"special": true
|
| 123 |
+
},
|
| 124 |
+
"15": {
|
| 125 |
+
"content": "<jupyter_script>",
|
| 126 |
+
"lstrip": false,
|
| 127 |
+
"normalized": false,
|
| 128 |
+
"rstrip": false,
|
| 129 |
+
"single_word": false,
|
| 130 |
+
"special": true
|
| 131 |
+
},
|
| 132 |
+
"16": {
|
| 133 |
+
"content": "<empty_output>",
|
| 134 |
+
"lstrip": false,
|
| 135 |
+
"normalized": false,
|
| 136 |
+
"rstrip": false,
|
| 137 |
+
"single_word": false,
|
| 138 |
+
"special": true
|
| 139 |
+
}
|
| 140 |
+
},
|
| 141 |
+
"additional_special_tokens": [
|
| 142 |
+
"<|im_start|>",
|
| 143 |
+
"<|im_end|>"
|
| 144 |
+
],
|
| 145 |
+
"bos_token": "<|im_start|>",
|
| 146 |
+
"clean_up_tokenization_spaces": false,
|
| 147 |
+
"eos_token": "<|im_end|>",
|
| 148 |
+
"extra_special_tokens": {},
|
| 149 |
+
"model_max_length": 8192,
|
| 150 |
+
"pad_token": "<|endoftext|>",
|
| 151 |
+
"padding_side": "right",
|
| 152 |
+
"tokenizer_class": "GPT2Tokenizer",
|
| 153 |
+
"unk_token": "<|endoftext|>",
|
| 154 |
+
"vocab_size": 49152
|
| 155 |
+
}
|
continuous/smollm2/mc1_b16_224_400m/finetune/trainer_state.json
ADDED
|
@@ -0,0 +1,1856 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 1.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 2597,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.003850596842510589,
|
| 14 |
+
"grad_norm": 3.2572810649871826,
|
| 15 |
+
"learning_rate": 1.153846153846154e-06,
|
| 16 |
+
"loss": 1.2671,
|
| 17 |
+
"step": 10
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.007701193685021178,
|
| 21 |
+
"grad_norm": 2.1146886348724365,
|
| 22 |
+
"learning_rate": 2.435897435897436e-06,
|
| 23 |
+
"loss": 1.1236,
|
| 24 |
+
"step": 20
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.011551790527531768,
|
| 28 |
+
"grad_norm": 2.0527374744415283,
|
| 29 |
+
"learning_rate": 3.7179487179487184e-06,
|
| 30 |
+
"loss": 1.0042,
|
| 31 |
+
"step": 30
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.015402387370042356,
|
| 35 |
+
"grad_norm": 1.3700536489486694,
|
| 36 |
+
"learning_rate": 5e-06,
|
| 37 |
+
"loss": 0.9555,
|
| 38 |
+
"step": 40
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.019252984212552945,
|
| 42 |
+
"grad_norm": 1.2790406942367554,
|
| 43 |
+
"learning_rate": 6.282051282051282e-06,
|
| 44 |
+
"loss": 0.9413,
|
| 45 |
+
"step": 50
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.023103581055063535,
|
| 49 |
+
"grad_norm": 0.2883168160915375,
|
| 50 |
+
"learning_rate": 7.564102564102564e-06,
|
| 51 |
+
"loss": 0.8723,
|
| 52 |
+
"step": 60
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.026954177897574125,
|
| 56 |
+
"grad_norm": 0.24443593621253967,
|
| 57 |
+
"learning_rate": 8.846153846153847e-06,
|
| 58 |
+
"loss": 0.8291,
|
| 59 |
+
"step": 70
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.030804774740084712,
|
| 63 |
+
"grad_norm": 0.24563875794410706,
|
| 64 |
+
"learning_rate": 9.999996111488723e-06,
|
| 65 |
+
"loss": 0.8437,
|
| 66 |
+
"step": 80
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.0346553715825953,
|
| 70 |
+
"grad_norm": 0.21684777736663818,
|
| 71 |
+
"learning_rate": 9.999529497453782e-06,
|
| 72 |
+
"loss": 0.831,
|
| 73 |
+
"step": 90
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.03850596842510589,
|
| 77 |
+
"grad_norm": 0.21515122056007385,
|
| 78 |
+
"learning_rate": 9.998285264324193e-06,
|
| 79 |
+
"loss": 0.8014,
|
| 80 |
+
"step": 100
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.04235656526761648,
|
| 84 |
+
"grad_norm": 0.21681255102157593,
|
| 85 |
+
"learning_rate": 9.996263605626052e-06,
|
| 86 |
+
"loss": 0.7936,
|
| 87 |
+
"step": 110
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.04620716211012707,
|
| 91 |
+
"grad_norm": 0.21227523684501648,
|
| 92 |
+
"learning_rate": 9.99346483580503e-06,
|
| 93 |
+
"loss": 0.8127,
|
| 94 |
+
"step": 120
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.05005775895263766,
|
| 98 |
+
"grad_norm": 0.22039848566055298,
|
| 99 |
+
"learning_rate": 9.989889390177467e-06,
|
| 100 |
+
"loss": 0.7888,
|
| 101 |
+
"step": 130
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.05390835579514825,
|
| 105 |
+
"grad_norm": 0.2273125946521759,
|
| 106 |
+
"learning_rate": 9.985537824862647e-06,
|
| 107 |
+
"loss": 0.8043,
|
| 108 |
+
"step": 140
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.05775895263765884,
|
| 112 |
+
"grad_norm": 0.22452707588672638,
|
| 113 |
+
"learning_rate": 9.980410816696317e-06,
|
| 114 |
+
"loss": 0.7805,
|
| 115 |
+
"step": 150
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.061609549480169425,
|
| 119 |
+
"grad_norm": 0.20935779809951782,
|
| 120 |
+
"learning_rate": 9.97450916312541e-06,
|
| 121 |
+
"loss": 0.7875,
|
| 122 |
+
"step": 160
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.06546014632268002,
|
| 126 |
+
"grad_norm": 0.180038183927536,
|
| 127 |
+
"learning_rate": 9.967833782084004e-06,
|
| 128 |
+
"loss": 0.787,
|
| 129 |
+
"step": 170
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.0693107431651906,
|
| 133 |
+
"grad_norm": 0.1984352469444275,
|
| 134 |
+
"learning_rate": 9.960385711850554e-06,
|
| 135 |
+
"loss": 0.8184,
|
| 136 |
+
"step": 180
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.07316134000770119,
|
| 140 |
+
"grad_norm": 0.20937885344028473,
|
| 141 |
+
"learning_rate": 9.952166110886394e-06,
|
| 142 |
+
"loss": 0.8061,
|
| 143 |
+
"step": 190
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.07701193685021178,
|
| 147 |
+
"grad_norm": 0.20725709199905396,
|
| 148 |
+
"learning_rate": 9.943176257655567e-06,
|
| 149 |
+
"loss": 0.8037,
|
| 150 |
+
"step": 200
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.08086253369272237,
|
| 154 |
+
"grad_norm": 0.21083995699882507,
|
| 155 |
+
"learning_rate": 9.933417550425952e-06,
|
| 156 |
+
"loss": 0.792,
|
| 157 |
+
"step": 210
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.08471313053523297,
|
| 161 |
+
"grad_norm": 0.2056027352809906,
|
| 162 |
+
"learning_rate": 9.9228915070518e-06,
|
| 163 |
+
"loss": 0.7806,
|
| 164 |
+
"step": 220
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.08856372737774355,
|
| 168 |
+
"grad_norm": 0.20162631571292877,
|
| 169 |
+
"learning_rate": 9.911599764737632e-06,
|
| 170 |
+
"loss": 0.7772,
|
| 171 |
+
"step": 230
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.09241432422025414,
|
| 175 |
+
"grad_norm": 0.1930011361837387,
|
| 176 |
+
"learning_rate": 9.899544079783606e-06,
|
| 177 |
+
"loss": 0.771,
|
| 178 |
+
"step": 240
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.09626492106276473,
|
| 182 |
+
"grad_norm": 0.1973913609981537,
|
| 183 |
+
"learning_rate": 9.886726327312338e-06,
|
| 184 |
+
"loss": 0.7887,
|
| 185 |
+
"step": 250
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.10011551790527531,
|
| 189 |
+
"grad_norm": 0.18832847476005554,
|
| 190 |
+
"learning_rate": 9.87314850097724e-06,
|
| 191 |
+
"loss": 0.7847,
|
| 192 |
+
"step": 260
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.1039661147477859,
|
| 196 |
+
"grad_norm": 0.19065648317337036,
|
| 197 |
+
"learning_rate": 9.858812712652452e-06,
|
| 198 |
+
"loss": 0.7692,
|
| 199 |
+
"step": 270
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.1078167115902965,
|
| 203 |
+
"grad_norm": 0.18433961272239685,
|
| 204 |
+
"learning_rate": 9.843721192104341e-06,
|
| 205 |
+
"loss": 0.766,
|
| 206 |
+
"step": 280
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.11166730843280709,
|
| 210 |
+
"grad_norm": 0.20792338252067566,
|
| 211 |
+
"learning_rate": 9.827876286644698e-06,
|
| 212 |
+
"loss": 0.7986,
|
| 213 |
+
"step": 290
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.11551790527531768,
|
| 217 |
+
"grad_norm": 0.18988417088985443,
|
| 218 |
+
"learning_rate": 9.811280460765643e-06,
|
| 219 |
+
"loss": 0.7563,
|
| 220 |
+
"step": 300
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.11936850211782826,
|
| 224 |
+
"grad_norm": 0.191196471452713,
|
| 225 |
+
"learning_rate": 9.793936295756292e-06,
|
| 226 |
+
"loss": 0.7611,
|
| 227 |
+
"step": 310
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.12321909896033885,
|
| 231 |
+
"grad_norm": 0.19292491674423218,
|
| 232 |
+
"learning_rate": 9.775846489301279e-06,
|
| 233 |
+
"loss": 0.7704,
|
| 234 |
+
"step": 320
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.12706969580284944,
|
| 238 |
+
"grad_norm": 0.18285511434078217,
|
| 239 |
+
"learning_rate": 9.757013855061157e-06,
|
| 240 |
+
"loss": 0.7714,
|
| 241 |
+
"step": 330
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.13092029264536004,
|
| 245 |
+
"grad_norm": 0.1902669221162796,
|
| 246 |
+
"learning_rate": 9.737441322234758e-06,
|
| 247 |
+
"loss": 0.775,
|
| 248 |
+
"step": 340
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.1347708894878706,
|
| 252 |
+
"grad_norm": 0.2036568969488144,
|
| 253 |
+
"learning_rate": 9.717131935103605e-06,
|
| 254 |
+
"loss": 0.7822,
|
| 255 |
+
"step": 350
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.1386214863303812,
|
| 259 |
+
"grad_norm": 0.1882489025592804,
|
| 260 |
+
"learning_rate": 9.696088852558398e-06,
|
| 261 |
+
"loss": 0.7628,
|
| 262 |
+
"step": 360
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.1424720831728918,
|
| 266 |
+
"grad_norm": 0.18622733652591705,
|
| 267 |
+
"learning_rate": 9.674315347607686e-06,
|
| 268 |
+
"loss": 0.7906,
|
| 269 |
+
"step": 370
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.14632268001540238,
|
| 273 |
+
"grad_norm": 0.17874756455421448,
|
| 274 |
+
"learning_rate": 9.651814806868796e-06,
|
| 275 |
+
"loss": 0.7555,
|
| 276 |
+
"step": 380
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.15017327685791299,
|
| 280 |
+
"grad_norm": 0.1910037100315094,
|
| 281 |
+
"learning_rate": 9.62859073004107e-06,
|
| 282 |
+
"loss": 0.7576,
|
| 283 |
+
"step": 390
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.15402387370042356,
|
| 287 |
+
"grad_norm": 0.17350363731384277,
|
| 288 |
+
"learning_rate": 9.604646729361547e-06,
|
| 289 |
+
"loss": 0.7659,
|
| 290 |
+
"step": 400
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.15787447054293416,
|
| 294 |
+
"grad_norm": 0.20693738758563995,
|
| 295 |
+
"learning_rate": 9.579986529043099e-06,
|
| 296 |
+
"loss": 0.7607,
|
| 297 |
+
"step": 410
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.16172506738544473,
|
| 301 |
+
"grad_norm": 0.20047707855701447,
|
| 302 |
+
"learning_rate": 9.554613964695189e-06,
|
| 303 |
+
"loss": 0.7773,
|
| 304 |
+
"step": 420
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 0.16557566422795533,
|
| 308 |
+
"grad_norm": 0.1861504763364792,
|
| 309 |
+
"learning_rate": 9.528532982727281e-06,
|
| 310 |
+
"loss": 0.759,
|
| 311 |
+
"step": 430
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.16942626107046593,
|
| 315 |
+
"grad_norm": 0.178996741771698,
|
| 316 |
+
"learning_rate": 9.501747639735018e-06,
|
| 317 |
+
"loss": 0.7655,
|
| 318 |
+
"step": 440
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.1732768579129765,
|
| 322 |
+
"grad_norm": 0.20141951739788055,
|
| 323 |
+
"learning_rate": 9.474262101869272e-06,
|
| 324 |
+
"loss": 0.75,
|
| 325 |
+
"step": 450
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.1771274547554871,
|
| 329 |
+
"grad_norm": 0.2120140790939331,
|
| 330 |
+
"learning_rate": 9.446080644188141e-06,
|
| 331 |
+
"loss": 0.7598,
|
| 332 |
+
"step": 460
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.18097805159799768,
|
| 336 |
+
"grad_norm": 0.19112084805965424,
|
| 337 |
+
"learning_rate": 9.417207649992017e-06,
|
| 338 |
+
"loss": 0.7607,
|
| 339 |
+
"step": 470
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.18482864844050828,
|
| 343 |
+
"grad_norm": 0.19272735714912415,
|
| 344 |
+
"learning_rate": 9.387647610141816e-06,
|
| 345 |
+
"loss": 0.7598,
|
| 346 |
+
"step": 480
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.18867924528301888,
|
| 350 |
+
"grad_norm": 0.18592801690101624,
|
| 351 |
+
"learning_rate": 9.35740512236047e-06,
|
| 352 |
+
"loss": 0.754,
|
| 353 |
+
"step": 490
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.19252984212552945,
|
| 357 |
+
"grad_norm": 0.17990081012248993,
|
| 358 |
+
"learning_rate": 9.326484890517809e-06,
|
| 359 |
+
"loss": 0.759,
|
| 360 |
+
"step": 500
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.19638043896804006,
|
| 364 |
+
"grad_norm": 0.2126895934343338,
|
| 365 |
+
"learning_rate": 9.294891723898926e-06,
|
| 366 |
+
"loss": 0.7478,
|
| 367 |
+
"step": 510
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.20023103581055063,
|
| 371 |
+
"grad_norm": 0.19312377274036407,
|
| 372 |
+
"learning_rate": 9.262630536456153e-06,
|
| 373 |
+
"loss": 0.7722,
|
| 374 |
+
"step": 520
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.20408163265306123,
|
| 378 |
+
"grad_norm": 0.19161541759967804,
|
| 379 |
+
"learning_rate": 9.229706346044749e-06,
|
| 380 |
+
"loss": 0.7546,
|
| 381 |
+
"step": 530
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.2079322294955718,
|
| 385 |
+
"grad_norm": 0.19222302734851837,
|
| 386 |
+
"learning_rate": 9.196124273642422e-06,
|
| 387 |
+
"loss": 0.7652,
|
| 388 |
+
"step": 540
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.2117828263380824,
|
| 392 |
+
"grad_norm": 0.18440645933151245,
|
| 393 |
+
"learning_rate": 9.161889542552842e-06,
|
| 394 |
+
"loss": 0.7712,
|
| 395 |
+
"step": 550
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.215633423180593,
|
| 399 |
+
"grad_norm": 0.18490299582481384,
|
| 400 |
+
"learning_rate": 9.127007477593193e-06,
|
| 401 |
+
"loss": 0.7727,
|
| 402 |
+
"step": 560
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.21948402002310358,
|
| 406 |
+
"grad_norm": 0.20648887753486633,
|
| 407 |
+
"learning_rate": 9.091483504265974e-06,
|
| 408 |
+
"loss": 0.7868,
|
| 409 |
+
"step": 570
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.22333461686561418,
|
| 413 |
+
"grad_norm": 0.22577454149723053,
|
| 414 |
+
"learning_rate": 9.055323147915116e-06,
|
| 415 |
+
"loss": 0.7547,
|
| 416 |
+
"step": 580
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.22718521370812475,
|
| 420 |
+
"grad_norm": 0.1867440938949585,
|
| 421 |
+
"learning_rate": 9.018532032866585e-06,
|
| 422 |
+
"loss": 0.7709,
|
| 423 |
+
"step": 590
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.23103581055063535,
|
| 427 |
+
"grad_norm": 0.17543499171733856,
|
| 428 |
+
"learning_rate": 8.981115881553579e-06,
|
| 429 |
+
"loss": 0.7668,
|
| 430 |
+
"step": 600
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 0.23488640739314592,
|
| 434 |
+
"grad_norm": 0.21719646453857422,
|
| 435 |
+
"learning_rate": 8.943080513626476e-06,
|
| 436 |
+
"loss": 0.7862,
|
| 437 |
+
"step": 610
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 0.23873700423565652,
|
| 441 |
+
"grad_norm": 0.19247278571128845,
|
| 442 |
+
"learning_rate": 8.904431845047641e-06,
|
| 443 |
+
"loss": 0.7651,
|
| 444 |
+
"step": 620
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.24258760107816713,
|
| 448 |
+
"grad_norm": 0.18842528760433197,
|
| 449 |
+
"learning_rate": 8.86517588717128e-06,
|
| 450 |
+
"loss": 0.7761,
|
| 451 |
+
"step": 630
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 0.2464381979206777,
|
| 455 |
+
"grad_norm": 0.19773036241531372,
|
| 456 |
+
"learning_rate": 8.82531874580844e-06,
|
| 457 |
+
"loss": 0.7744,
|
| 458 |
+
"step": 640
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.25028879476318827,
|
| 462 |
+
"grad_norm": 0.17201146483421326,
|
| 463 |
+
"learning_rate": 8.784866620277314e-06,
|
| 464 |
+
"loss": 0.7355,
|
| 465 |
+
"step": 650
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.25413939160569887,
|
| 469 |
+
"grad_norm": 0.1932615488767624,
|
| 470 |
+
"learning_rate": 8.743825802439018e-06,
|
| 471 |
+
"loss": 0.7601,
|
| 472 |
+
"step": 660
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 0.2579899884482095,
|
| 476 |
+
"grad_norm": 0.1857462376356125,
|
| 477 |
+
"learning_rate": 8.702202675718956e-06,
|
| 478 |
+
"loss": 0.7702,
|
| 479 |
+
"step": 670
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.2618405852907201,
|
| 483 |
+
"grad_norm": 0.17275774478912354,
|
| 484 |
+
"learning_rate": 8.660003714113959e-06,
|
| 485 |
+
"loss": 0.7513,
|
| 486 |
+
"step": 680
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 0.2656911821332307,
|
| 490 |
+
"grad_norm": 0.18267491459846497,
|
| 491 |
+
"learning_rate": 8.617235481185323e-06,
|
| 492 |
+
"loss": 0.7603,
|
| 493 |
+
"step": 690
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 0.2695417789757412,
|
| 497 |
+
"grad_norm": 0.19103997945785522,
|
| 498 |
+
"learning_rate": 8.573904629037931e-06,
|
| 499 |
+
"loss": 0.7739,
|
| 500 |
+
"step": 700
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 0.2733923758182518,
|
| 504 |
+
"grad_norm": 0.18015512824058533,
|
| 505 |
+
"learning_rate": 8.530017897285585e-06,
|
| 506 |
+
"loss": 0.7552,
|
| 507 |
+
"step": 710
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.2772429726607624,
|
| 511 |
+
"grad_norm": 0.18993747234344482,
|
| 512 |
+
"learning_rate": 8.485582112002746e-06,
|
| 513 |
+
"loss": 0.7559,
|
| 514 |
+
"step": 720
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 0.281093569503273,
|
| 518 |
+
"grad_norm": 0.18540197610855103,
|
| 519 |
+
"learning_rate": 8.440604184662811e-06,
|
| 520 |
+
"loss": 0.7657,
|
| 521 |
+
"step": 730
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 0.2849441663457836,
|
| 525 |
+
"grad_norm": 0.18727004528045654,
|
| 526 |
+
"learning_rate": 8.395091111063118e-06,
|
| 527 |
+
"loss": 0.7506,
|
| 528 |
+
"step": 740
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 0.28879476318829417,
|
| 532 |
+
"grad_norm": 0.1914101243019104,
|
| 533 |
+
"learning_rate": 8.349049970236822e-06,
|
| 534 |
+
"loss": 0.7483,
|
| 535 |
+
"step": 750
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 0.29264536003080477,
|
| 539 |
+
"grad_norm": 0.18833674490451813,
|
| 540 |
+
"learning_rate": 8.302487923351842e-06,
|
| 541 |
+
"loss": 0.7454,
|
| 542 |
+
"step": 760
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 0.29649595687331537,
|
| 546 |
+
"grad_norm": 0.170330211520195,
|
| 547 |
+
"learning_rate": 8.255412212597021e-06,
|
| 548 |
+
"loss": 0.7634,
|
| 549 |
+
"step": 770
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.30034655371582597,
|
| 553 |
+
"grad_norm": 0.18962769210338593,
|
| 554 |
+
"learning_rate": 8.207830160055688e-06,
|
| 555 |
+
"loss": 0.7601,
|
| 556 |
+
"step": 780
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 0.3041971505583365,
|
| 560 |
+
"grad_norm": 0.17611345648765564,
|
| 561 |
+
"learning_rate": 8.159749166566786e-06,
|
| 562 |
+
"loss": 0.7423,
|
| 563 |
+
"step": 790
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 0.3080477474008471,
|
| 567 |
+
"grad_norm": 0.1823180913925171,
|
| 568 |
+
"learning_rate": 8.11117671057377e-06,
|
| 569 |
+
"loss": 0.7641,
|
| 570 |
+
"step": 800
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 0.3118983442433577,
|
| 574 |
+
"grad_norm": 0.17702196538448334,
|
| 575 |
+
"learning_rate": 8.062120346961404e-06,
|
| 576 |
+
"loss": 0.7632,
|
| 577 |
+
"step": 810
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 0.3157489410858683,
|
| 581 |
+
"grad_norm": 0.18140192329883575,
|
| 582 |
+
"learning_rate": 8.012587705880705e-06,
|
| 583 |
+
"loss": 0.7602,
|
| 584 |
+
"step": 820
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 0.3195995379283789,
|
| 588 |
+
"grad_norm": 0.18093664944171906,
|
| 589 |
+
"learning_rate": 7.962586491562146e-06,
|
| 590 |
+
"loss": 0.7457,
|
| 591 |
+
"step": 830
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 0.32345013477088946,
|
| 595 |
+
"grad_norm": 0.18367372453212738,
|
| 596 |
+
"learning_rate": 7.912124481117357e-06,
|
| 597 |
+
"loss": 0.7461,
|
| 598 |
+
"step": 840
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 0.32730073161340006,
|
| 602 |
+
"grad_norm": 0.1731540709733963,
|
| 603 |
+
"learning_rate": 7.861209523329494e-06,
|
| 604 |
+
"loss": 0.756,
|
| 605 |
+
"step": 850
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 0.33115132845591067,
|
| 609 |
+
"grad_norm": 0.1841742843389511,
|
| 610 |
+
"learning_rate": 7.809849537432432e-06,
|
| 611 |
+
"loss": 0.7475,
|
| 612 |
+
"step": 860
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 0.33500192529842127,
|
| 616 |
+
"grad_norm": 0.17925141751766205,
|
| 617 |
+
"learning_rate": 7.758052511879035e-06,
|
| 618 |
+
"loss": 0.7515,
|
| 619 |
+
"step": 870
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 0.33885252214093187,
|
| 623 |
+
"grad_norm": 0.1823624223470688,
|
| 624 |
+
"learning_rate": 7.705826503098642e-06,
|
| 625 |
+
"loss": 0.755,
|
| 626 |
+
"step": 880
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 0.3427031189834424,
|
| 630 |
+
"grad_norm": 0.18201403319835663,
|
| 631 |
+
"learning_rate": 7.653179634243979e-06,
|
| 632 |
+
"loss": 0.748,
|
| 633 |
+
"step": 890
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 0.346553715825953,
|
| 637 |
+
"grad_norm": 0.16994622349739075,
|
| 638 |
+
"learning_rate": 7.600120093927696e-06,
|
| 639 |
+
"loss": 0.7503,
|
| 640 |
+
"step": 900
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 0.3504043126684636,
|
| 644 |
+
"grad_norm": 0.19723546504974365,
|
| 645 |
+
"learning_rate": 7.546656134948729e-06,
|
| 646 |
+
"loss": 0.7401,
|
| 647 |
+
"step": 910
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 0.3542549095109742,
|
| 651 |
+
"grad_norm": 0.17284683883190155,
|
| 652 |
+
"learning_rate": 7.492796073008661e-06,
|
| 653 |
+
"loss": 0.7478,
|
| 654 |
+
"step": 920
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.3581055063534848,
|
| 658 |
+
"grad_norm": 0.19395524263381958,
|
| 659 |
+
"learning_rate": 7.438548285418326e-06,
|
| 660 |
+
"loss": 0.7293,
|
| 661 |
+
"step": 930
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.36195610319599536,
|
| 665 |
+
"grad_norm": 0.1948285549879074,
|
| 666 |
+
"learning_rate": 7.3839212097948e-06,
|
| 667 |
+
"loss": 0.7601,
|
| 668 |
+
"step": 940
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 0.36580670003850596,
|
| 672 |
+
"grad_norm": 0.20074774324893951,
|
| 673 |
+
"learning_rate": 7.328923342749039e-06,
|
| 674 |
+
"loss": 0.776,
|
| 675 |
+
"step": 950
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.36965729688101656,
|
| 679 |
+
"grad_norm": 0.1806403249502182,
|
| 680 |
+
"learning_rate": 7.273563238564319e-06,
|
| 681 |
+
"loss": 0.7722,
|
| 682 |
+
"step": 960
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 0.37350789372352716,
|
| 686 |
+
"grad_norm": 0.1911482959985733,
|
| 687 |
+
"learning_rate": 7.217849507865724e-06,
|
| 688 |
+
"loss": 0.741,
|
| 689 |
+
"step": 970
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 0.37735849056603776,
|
| 693 |
+
"grad_norm": 0.1777525097131729,
|
| 694 |
+
"learning_rate": 7.161790816280858e-06,
|
| 695 |
+
"loss": 0.7504,
|
| 696 |
+
"step": 980
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 0.3812090874085483,
|
| 700 |
+
"grad_norm": 0.19722962379455566,
|
| 701 |
+
"learning_rate": 7.1053958830920125e-06,
|
| 702 |
+
"loss": 0.7372,
|
| 703 |
+
"step": 990
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.3850596842510589,
|
| 707 |
+
"grad_norm": 0.1947478950023651,
|
| 708 |
+
"learning_rate": 7.0486734798799715e-06,
|
| 709 |
+
"loss": 0.7504,
|
| 710 |
+
"step": 1000
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 0.3889102810935695,
|
| 714 |
+
"grad_norm": 0.17489983141422272,
|
| 715 |
+
"learning_rate": 6.991632429159708e-06,
|
| 716 |
+
"loss": 0.7687,
|
| 717 |
+
"step": 1010
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 0.3927608779360801,
|
| 721 |
+
"grad_norm": 0.19125287234783173,
|
| 722 |
+
"learning_rate": 6.934281603008132e-06,
|
| 723 |
+
"loss": 0.7673,
|
| 724 |
+
"step": 1020
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"epoch": 0.39661147477859066,
|
| 728 |
+
"grad_norm": 0.167512446641922,
|
| 729 |
+
"learning_rate": 6.876629921684143e-06,
|
| 730 |
+
"loss": 0.7631,
|
| 731 |
+
"step": 1030
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 0.40046207162110126,
|
| 735 |
+
"grad_norm": 0.19253911077976227,
|
| 736 |
+
"learning_rate": 6.818686352241199e-06,
|
| 737 |
+
"loss": 0.7516,
|
| 738 |
+
"step": 1040
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 0.40431266846361186,
|
| 742 |
+
"grad_norm": 0.16116246581077576,
|
| 743 |
+
"learning_rate": 6.760459907132573e-06,
|
| 744 |
+
"loss": 0.7464,
|
| 745 |
+
"step": 1050
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 0.40816326530612246,
|
| 749 |
+
"grad_norm": 0.1655367612838745,
|
| 750 |
+
"learning_rate": 6.701959642809589e-06,
|
| 751 |
+
"loss": 0.7527,
|
| 752 |
+
"step": 1060
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 0.41201386214863306,
|
| 756 |
+
"grad_norm": 0.17807863652706146,
|
| 757 |
+
"learning_rate": 6.643194658312987e-06,
|
| 758 |
+
"loss": 0.7443,
|
| 759 |
+
"step": 1070
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.4158644589911436,
|
| 763 |
+
"grad_norm": 0.16834115982055664,
|
| 764 |
+
"learning_rate": 6.584174093857676e-06,
|
| 765 |
+
"loss": 0.7698,
|
| 766 |
+
"step": 1080
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 0.4197150558336542,
|
| 770 |
+
"grad_norm": 0.186019629240036,
|
| 771 |
+
"learning_rate": 6.5249071294110735e-06,
|
| 772 |
+
"loss": 0.7534,
|
| 773 |
+
"step": 1090
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 0.4235656526761648,
|
| 777 |
+
"grad_norm": 0.1853189766407013,
|
| 778 |
+
"learning_rate": 6.465402983265274e-06,
|
| 779 |
+
"loss": 0.7645,
|
| 780 |
+
"step": 1100
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 0.4274162495186754,
|
| 784 |
+
"grad_norm": 0.1970815360546112,
|
| 785 |
+
"learning_rate": 6.405670910603243e-06,
|
| 786 |
+
"loss": 0.7573,
|
| 787 |
+
"step": 1110
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 0.431266846361186,
|
| 791 |
+
"grad_norm": 0.1987958401441574,
|
| 792 |
+
"learning_rate": 6.345720202059291e-06,
|
| 793 |
+
"loss": 0.7543,
|
| 794 |
+
"step": 1120
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 0.43511744320369655,
|
| 798 |
+
"grad_norm": 0.18732966482639313,
|
| 799 |
+
"learning_rate": 6.285560182274008e-06,
|
| 800 |
+
"loss": 0.7351,
|
| 801 |
+
"step": 1130
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 0.43896804004620715,
|
| 805 |
+
"grad_norm": 0.19548921287059784,
|
| 806 |
+
"learning_rate": 6.225200208443936e-06,
|
| 807 |
+
"loss": 0.7593,
|
| 808 |
+
"step": 1140
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 0.44281863688871775,
|
| 812 |
+
"grad_norm": 0.17334434390068054,
|
| 813 |
+
"learning_rate": 6.16464966886616e-06,
|
| 814 |
+
"loss": 0.7437,
|
| 815 |
+
"step": 1150
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 0.44666923373122835,
|
| 819 |
+
"grad_norm": 0.22224017977714539,
|
| 820 |
+
"learning_rate": 6.103917981478065e-06,
|
| 821 |
+
"loss": 0.7738,
|
| 822 |
+
"step": 1160
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 0.45051983057373896,
|
| 826 |
+
"grad_norm": 0.16305769979953766,
|
| 827 |
+
"learning_rate": 6.043014592392489e-06,
|
| 828 |
+
"loss": 0.7802,
|
| 829 |
+
"step": 1170
|
| 830 |
+
},
|
| 831 |
+
{
|
| 832 |
+
"epoch": 0.4543704274162495,
|
| 833 |
+
"grad_norm": 0.18167075514793396,
|
| 834 |
+
"learning_rate": 5.981948974428485e-06,
|
| 835 |
+
"loss": 0.7601,
|
| 836 |
+
"step": 1180
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 0.4582210242587601,
|
| 840 |
+
"grad_norm": 0.1657552868127823,
|
| 841 |
+
"learning_rate": 5.920730625637934e-06,
|
| 842 |
+
"loss": 0.7363,
|
| 843 |
+
"step": 1190
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 0.4620716211012707,
|
| 847 |
+
"grad_norm": 0.1790909320116043,
|
| 848 |
+
"learning_rate": 5.8593690678282355e-06,
|
| 849 |
+
"loss": 0.7453,
|
| 850 |
+
"step": 1200
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 0.4659222179437813,
|
| 854 |
+
"grad_norm": 0.17890852689743042,
|
| 855 |
+
"learning_rate": 5.797873845081301e-06,
|
| 856 |
+
"loss": 0.7442,
|
| 857 |
+
"step": 1210
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 0.46977281478629185,
|
| 861 |
+
"grad_norm": 0.19185440242290497,
|
| 862 |
+
"learning_rate": 5.736254522269075e-06,
|
| 863 |
+
"loss": 0.7691,
|
| 864 |
+
"step": 1220
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 0.47362341162880245,
|
| 868 |
+
"grad_norm": 0.17831683158874512,
|
| 869 |
+
"learning_rate": 5.6745206835658386e-06,
|
| 870 |
+
"loss": 0.7651,
|
| 871 |
+
"step": 1230
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 0.47747400847131305,
|
| 875 |
+
"grad_norm": 0.1693374216556549,
|
| 876 |
+
"learning_rate": 5.612681930957495e-06,
|
| 877 |
+
"loss": 0.7437,
|
| 878 |
+
"step": 1240
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 0.48132460531382365,
|
| 882 |
+
"grad_norm": 0.17822201550006866,
|
| 883 |
+
"learning_rate": 5.550747882748095e-06,
|
| 884 |
+
"loss": 0.7368,
|
| 885 |
+
"step": 1250
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 0.48517520215633425,
|
| 889 |
+
"grad_norm": 0.19760315120220184,
|
| 890 |
+
"learning_rate": 5.4887281720638195e-06,
|
| 891 |
+
"loss": 0.7505,
|
| 892 |
+
"step": 1260
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 0.4890257989988448,
|
| 896 |
+
"grad_norm": 0.19030965864658356,
|
| 897 |
+
"learning_rate": 5.42663244535466e-06,
|
| 898 |
+
"loss": 0.7269,
|
| 899 |
+
"step": 1270
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 0.4928763958413554,
|
| 903 |
+
"grad_norm": 0.1867971569299698,
|
| 904 |
+
"learning_rate": 5.3644703608940205e-06,
|
| 905 |
+
"loss": 0.7441,
|
| 906 |
+
"step": 1280
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 0.496726992683866,
|
| 910 |
+
"grad_norm": 0.17360694706439972,
|
| 911 |
+
"learning_rate": 5.302251587276491e-06,
|
| 912 |
+
"loss": 0.7561,
|
| 913 |
+
"step": 1290
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 0.5005775895263765,
|
| 917 |
+
"grad_norm": 0.1783231645822525,
|
| 918 |
+
"learning_rate": 5.2399858019140005e-06,
|
| 919 |
+
"loss": 0.7462,
|
| 920 |
+
"step": 1300
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 0.5044281863688872,
|
| 924 |
+
"grad_norm": 0.1832990050315857,
|
| 925 |
+
"learning_rate": 5.177682689530615e-06,
|
| 926 |
+
"loss": 0.7649,
|
| 927 |
+
"step": 1310
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 0.5082787832113977,
|
| 931 |
+
"grad_norm": 0.1927586793899536,
|
| 932 |
+
"learning_rate": 5.11535194065619e-06,
|
| 933 |
+
"loss": 0.7359,
|
| 934 |
+
"step": 1320
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 0.5121293800539084,
|
| 938 |
+
"grad_norm": 0.18354566395282745,
|
| 939 |
+
"learning_rate": 5.053003250119113e-06,
|
| 940 |
+
"loss": 0.7441,
|
| 941 |
+
"step": 1330
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 0.515979976896419,
|
| 945 |
+
"grad_norm": 0.1749369204044342,
|
| 946 |
+
"learning_rate": 4.9906463155383925e-06,
|
| 947 |
+
"loss": 0.7496,
|
| 948 |
+
"step": 1340
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 0.5198305737389295,
|
| 952 |
+
"grad_norm": 0.1803244650363922,
|
| 953 |
+
"learning_rate": 4.9282908358153046e-06,
|
| 954 |
+
"loss": 0.7479,
|
| 955 |
+
"step": 1350
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 0.5236811705814401,
|
| 959 |
+
"grad_norm": 0.21021044254302979,
|
| 960 |
+
"learning_rate": 4.8659465096248345e-06,
|
| 961 |
+
"loss": 0.7633,
|
| 962 |
+
"step": 1360
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 0.5275317674239507,
|
| 966 |
+
"grad_norm": 0.19374245405197144,
|
| 967 |
+
"learning_rate": 4.803623033907169e-06,
|
| 968 |
+
"loss": 0.7556,
|
| 969 |
+
"step": 1370
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 0.5313823642664613,
|
| 973 |
+
"grad_norm": 0.16912402212619781,
|
| 974 |
+
"learning_rate": 4.741330102359441e-06,
|
| 975 |
+
"loss": 0.769,
|
| 976 |
+
"step": 1380
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 0.5352329611089719,
|
| 980 |
+
"grad_norm": 0.1814538538455963,
|
| 981 |
+
"learning_rate": 4.679077403927996e-06,
|
| 982 |
+
"loss": 0.7439,
|
| 983 |
+
"step": 1390
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 0.5390835579514824,
|
| 987 |
+
"grad_norm": 0.17590168118476868,
|
| 988 |
+
"learning_rate": 4.616874621301373e-06,
|
| 989 |
+
"loss": 0.7458,
|
| 990 |
+
"step": 1400
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 0.5429341547939931,
|
| 994 |
+
"grad_norm": 0.18883827328681946,
|
| 995 |
+
"learning_rate": 4.554731429404293e-06,
|
| 996 |
+
"loss": 0.7644,
|
| 997 |
+
"step": 1410
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 0.5467847516365036,
|
| 1001 |
+
"grad_norm": 0.18659785389900208,
|
| 1002 |
+
"learning_rate": 4.492657493892822e-06,
|
| 1003 |
+
"loss": 0.746,
|
| 1004 |
+
"step": 1420
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 0.5506353484790143,
|
| 1008 |
+
"grad_norm": 0.17946568131446838,
|
| 1009 |
+
"learning_rate": 4.430662469650999e-06,
|
| 1010 |
+
"loss": 0.7593,
|
| 1011 |
+
"step": 1430
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 0.5544859453215248,
|
| 1015 |
+
"grad_norm": 0.1915060132741928,
|
| 1016 |
+
"learning_rate": 4.368755999289118e-06,
|
| 1017 |
+
"loss": 0.7606,
|
| 1018 |
+
"step": 1440
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 0.5583365421640354,
|
| 1022 |
+
"grad_norm": 0.18806183338165283,
|
| 1023 |
+
"learning_rate": 4.306947711643949e-06,
|
| 1024 |
+
"loss": 0.7444,
|
| 1025 |
+
"step": 1450
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 0.562187139006546,
|
| 1029 |
+
"grad_norm": 0.1815820336341858,
|
| 1030 |
+
"learning_rate": 4.245247220281068e-06,
|
| 1031 |
+
"loss": 0.7503,
|
| 1032 |
+
"step": 1460
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 0.5660377358490566,
|
| 1036 |
+
"grad_norm": 0.18319587409496307,
|
| 1037 |
+
"learning_rate": 4.183664121999587e-06,
|
| 1038 |
+
"loss": 0.7457,
|
| 1039 |
+
"step": 1470
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"epoch": 0.5698883326915672,
|
| 1043 |
+
"grad_norm": 0.19464389979839325,
|
| 1044 |
+
"learning_rate": 4.122207995339474e-06,
|
| 1045 |
+
"loss": 0.7391,
|
| 1046 |
+
"step": 1480
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 0.5737389295340778,
|
| 1050 |
+
"grad_norm": 0.18123389780521393,
|
| 1051 |
+
"learning_rate": 4.060888399091734e-06,
|
| 1052 |
+
"loss": 0.748,
|
| 1053 |
+
"step": 1490
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 0.5775895263765883,
|
| 1057 |
+
"grad_norm": 0.17612437903881073,
|
| 1058 |
+
"learning_rate": 3.999714870811639e-06,
|
| 1059 |
+
"loss": 0.7601,
|
| 1060 |
+
"step": 1500
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 0.581440123219099,
|
| 1064 |
+
"grad_norm": 0.17643168568611145,
|
| 1065 |
+
"learning_rate": 3.938696925335275e-06,
|
| 1066 |
+
"loss": 0.7531,
|
| 1067 |
+
"step": 1510
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 0.5852907200616095,
|
| 1071 |
+
"grad_norm": 0.1881759762763977,
|
| 1072 |
+
"learning_rate": 3.87784405329962e-06,
|
| 1073 |
+
"loss": 0.7616,
|
| 1074 |
+
"step": 1520
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 0.5891413169041202,
|
| 1078 |
+
"grad_norm": 0.1721840351819992,
|
| 1079 |
+
"learning_rate": 3.817165719666382e-06,
|
| 1080 |
+
"loss": 0.7607,
|
| 1081 |
+
"step": 1530
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 0.5929919137466307,
|
| 1085 |
+
"grad_norm": 0.18025976419448853,
|
| 1086 |
+
"learning_rate": 3.756671362249841e-06,
|
| 1087 |
+
"loss": 0.7631,
|
| 1088 |
+
"step": 1540
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 0.5968425105891413,
|
| 1092 |
+
"grad_norm": 0.16595016419887543,
|
| 1093 |
+
"learning_rate": 3.6963703902488907e-06,
|
| 1094 |
+
"loss": 0.7506,
|
| 1095 |
+
"step": 1550
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 0.6006931074316519,
|
| 1099 |
+
"grad_norm": 0.16796435415744781,
|
| 1100 |
+
"learning_rate": 3.6362721827835655e-06,
|
| 1101 |
+
"loss": 0.7621,
|
| 1102 |
+
"step": 1560
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 0.6045437042741625,
|
| 1106 |
+
"grad_norm": 0.1921037882566452,
|
| 1107 |
+
"learning_rate": 3.576386087436212e-06,
|
| 1108 |
+
"loss": 0.7665,
|
| 1109 |
+
"step": 1570
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 0.608394301116673,
|
| 1113 |
+
"grad_norm": 0.19326242804527283,
|
| 1114 |
+
"learning_rate": 3.5167214187975913e-06,
|
| 1115 |
+
"loss": 0.7592,
|
| 1116 |
+
"step": 1580
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 0.6122448979591837,
|
| 1120 |
+
"grad_norm": 0.16668424010276794,
|
| 1121 |
+
"learning_rate": 3.457287457018086e-06,
|
| 1122 |
+
"loss": 0.7549,
|
| 1123 |
+
"step": 1590
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 0.6160954948016942,
|
| 1127 |
+
"grad_norm": 0.1726846843957901,
|
| 1128 |
+
"learning_rate": 3.3980934463643003e-06,
|
| 1129 |
+
"loss": 0.7508,
|
| 1130 |
+
"step": 1600
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 0.6199460916442049,
|
| 1134 |
+
"grad_norm": 0.1611778736114502,
|
| 1135 |
+
"learning_rate": 3.3391485937812053e-06,
|
| 1136 |
+
"loss": 0.7287,
|
| 1137 |
+
"step": 1610
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 0.6237966884867154,
|
| 1141 |
+
"grad_norm": 0.17922617495059967,
|
| 1142 |
+
"learning_rate": 3.280462067460115e-06,
|
| 1143 |
+
"loss": 0.729,
|
| 1144 |
+
"step": 1620
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 0.627647285329226,
|
| 1148 |
+
"grad_norm": 0.17332057654857635,
|
| 1149 |
+
"learning_rate": 3.222042995412669e-06,
|
| 1150 |
+
"loss": 0.7555,
|
| 1151 |
+
"step": 1630
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 0.6314978821717366,
|
| 1155 |
+
"grad_norm": 0.16504517197608948,
|
| 1156 |
+
"learning_rate": 3.163900464051089e-06,
|
| 1157 |
+
"loss": 0.7464,
|
| 1158 |
+
"step": 1640
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"epoch": 0.6353484790142472,
|
| 1162 |
+
"grad_norm": 0.19378060102462769,
|
| 1163 |
+
"learning_rate": 3.1060435167748826e-06,
|
| 1164 |
+
"loss": 0.7619,
|
| 1165 |
+
"step": 1650
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 0.6391990758567578,
|
| 1169 |
+
"grad_norm": 0.18351104855537415,
|
| 1170 |
+
"learning_rate": 3.048481152564253e-06,
|
| 1171 |
+
"loss": 0.738,
|
| 1172 |
+
"step": 1660
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 0.6430496726992684,
|
| 1176 |
+
"grad_norm": 0.18634767830371857,
|
| 1177 |
+
"learning_rate": 2.9912223245804017e-06,
|
| 1178 |
+
"loss": 0.7562,
|
| 1179 |
+
"step": 1670
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 0.6469002695417789,
|
| 1183 |
+
"grad_norm": 0.1780145913362503,
|
| 1184 |
+
"learning_rate": 2.934275938772981e-06,
|
| 1185 |
+
"loss": 0.7483,
|
| 1186 |
+
"step": 1680
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 0.6507508663842896,
|
| 1190 |
+
"grad_norm": 0.18683426082134247,
|
| 1191 |
+
"learning_rate": 2.8776508524948676e-06,
|
| 1192 |
+
"loss": 0.7433,
|
| 1193 |
+
"step": 1690
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 0.6546014632268001,
|
| 1197 |
+
"grad_norm": 0.17372776567935944,
|
| 1198 |
+
"learning_rate": 2.821355873124503e-06,
|
| 1199 |
+
"loss": 0.7457,
|
| 1200 |
+
"step": 1700
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 0.6584520600693108,
|
| 1204 |
+
"grad_norm": 0.19463691115379333,
|
| 1205 |
+
"learning_rate": 2.7653997566960144e-06,
|
| 1206 |
+
"loss": 0.7515,
|
| 1207 |
+
"step": 1710
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 0.6623026569118213,
|
| 1211 |
+
"grad_norm": 0.20363306999206543,
|
| 1212 |
+
"learning_rate": 2.709791206537309e-06,
|
| 1213 |
+
"loss": 0.7436,
|
| 1214 |
+
"step": 1720
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 0.6661532537543319,
|
| 1218 |
+
"grad_norm": 0.1963336169719696,
|
| 1219 |
+
"learning_rate": 2.6545388719163747e-06,
|
| 1220 |
+
"loss": 0.7501,
|
| 1221 |
+
"step": 1730
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 0.6700038505968425,
|
| 1225 |
+
"grad_norm": 0.17929668724536896,
|
| 1226 |
+
"learning_rate": 2.599651346695979e-06,
|
| 1227 |
+
"loss": 0.7586,
|
| 1228 |
+
"step": 1740
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 0.6738544474393531,
|
| 1232 |
+
"grad_norm": 0.1720041036605835,
|
| 1233 |
+
"learning_rate": 2.545137167997007e-06,
|
| 1234 |
+
"loss": 0.7511,
|
| 1235 |
+
"step": 1750
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 0.6777050442818637,
|
| 1239 |
+
"grad_norm": 0.18184269964694977,
|
| 1240 |
+
"learning_rate": 2.4910048148705972e-06,
|
| 1241 |
+
"loss": 0.7451,
|
| 1242 |
+
"step": 1760
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 0.6815556411243743,
|
| 1246 |
+
"grad_norm": 0.18685950338840485,
|
| 1247 |
+
"learning_rate": 2.4372627069793327e-06,
|
| 1248 |
+
"loss": 0.7801,
|
| 1249 |
+
"step": 1770
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 0.6854062379668848,
|
| 1253 |
+
"grad_norm": 0.17641423642635345,
|
| 1254 |
+
"learning_rate": 2.3839192032876486e-06,
|
| 1255 |
+
"loss": 0.7559,
|
| 1256 |
+
"step": 1780
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 0.6892568348093955,
|
| 1260 |
+
"grad_norm": 0.18199770152568817,
|
| 1261 |
+
"learning_rate": 2.3309826007617047e-06,
|
| 1262 |
+
"loss": 0.7728,
|
| 1263 |
+
"step": 1790
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 0.693107431651906,
|
| 1267 |
+
"grad_norm": 0.18037888407707214,
|
| 1268 |
+
"learning_rate": 2.278461133078881e-06,
|
| 1269 |
+
"loss": 0.7412,
|
| 1270 |
+
"step": 1800
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 0.6969580284944167,
|
| 1274 |
+
"grad_norm": 0.180389404296875,
|
| 1275 |
+
"learning_rate": 2.2263629693471258e-06,
|
| 1276 |
+
"loss": 0.7347,
|
| 1277 |
+
"step": 1810
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 0.7008086253369272,
|
| 1281 |
+
"grad_norm": 0.1683788299560547,
|
| 1282 |
+
"learning_rate": 2.1746962128343414e-06,
|
| 1283 |
+
"loss": 0.7577,
|
| 1284 |
+
"step": 1820
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 0.7046592221794378,
|
| 1288 |
+
"grad_norm": 0.17885299026966095,
|
| 1289 |
+
"learning_rate": 2.1234688997080183e-06,
|
| 1290 |
+
"loss": 0.7415,
|
| 1291 |
+
"step": 1830
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 0.7085098190219484,
|
| 1295 |
+
"grad_norm": 0.178343266248703,
|
| 1296 |
+
"learning_rate": 2.0726889977853053e-06,
|
| 1297 |
+
"loss": 0.7461,
|
| 1298 |
+
"step": 1840
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 0.712360415864459,
|
| 1302 |
+
"grad_norm": 0.1892879158258438,
|
| 1303 |
+
"learning_rate": 2.022364405293703e-06,
|
| 1304 |
+
"loss": 0.7614,
|
| 1305 |
+
"step": 1850
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 0.7162110127069696,
|
| 1309 |
+
"grad_norm": 0.18038122355937958,
|
| 1310 |
+
"learning_rate": 1.9725029496425814e-06,
|
| 1311 |
+
"loss": 0.7589,
|
| 1312 |
+
"step": 1860
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 0.7200616095494802,
|
| 1316 |
+
"grad_norm": 0.17650437355041504,
|
| 1317 |
+
"learning_rate": 1.923112386205727e-06,
|
| 1318 |
+
"loss": 0.7488,
|
| 1319 |
+
"step": 1870
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 0.7239122063919907,
|
| 1323 |
+
"grad_norm": 0.1750689446926117,
|
| 1324 |
+
"learning_rate": 1.874200397115079e-06,
|
| 1325 |
+
"loss": 0.76,
|
| 1326 |
+
"step": 1880
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 0.7277628032345014,
|
| 1330 |
+
"grad_norm": 0.1856498420238495,
|
| 1331 |
+
"learning_rate": 1.825774590065857e-06,
|
| 1332 |
+
"loss": 0.7456,
|
| 1333 |
+
"step": 1890
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 0.7316134000770119,
|
| 1337 |
+
"grad_norm": 0.19084976613521576,
|
| 1338 |
+
"learning_rate": 1.7778424971332813e-06,
|
| 1339 |
+
"loss": 0.7568,
|
| 1340 |
+
"step": 1900
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 0.7354639969195226,
|
| 1344 |
+
"grad_norm": 0.17145371437072754,
|
| 1345 |
+
"learning_rate": 1.7304115736010407e-06,
|
| 1346 |
+
"loss": 0.7531,
|
| 1347 |
+
"step": 1910
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 0.7393145937620331,
|
| 1351 |
+
"grad_norm": 0.1788429617881775,
|
| 1352 |
+
"learning_rate": 1.6834891968017108e-06,
|
| 1353 |
+
"loss": 0.7442,
|
| 1354 |
+
"step": 1920
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 0.7431651906045437,
|
| 1358 |
+
"grad_norm": 0.1934567540884018,
|
| 1359 |
+
"learning_rate": 1.6370826649692884e-06,
|
| 1360 |
+
"loss": 0.7624,
|
| 1361 |
+
"step": 1930
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 0.7470157874470543,
|
| 1365 |
+
"grad_norm": 0.17106842994689941,
|
| 1366 |
+
"learning_rate": 1.5911991961040463e-06,
|
| 1367 |
+
"loss": 0.7601,
|
| 1368 |
+
"step": 1940
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 0.7508663842895649,
|
| 1372 |
+
"grad_norm": 0.17626892030239105,
|
| 1373 |
+
"learning_rate": 1.545845926849851e-06,
|
| 1374 |
+
"loss": 0.7429,
|
| 1375 |
+
"step": 1950
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 0.7547169811320755,
|
| 1379 |
+
"grad_norm": 0.18997561931610107,
|
| 1380 |
+
"learning_rate": 1.5010299113841397e-06,
|
| 1381 |
+
"loss": 0.7684,
|
| 1382 |
+
"step": 1960
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 0.7585675779745861,
|
| 1386 |
+
"grad_norm": 0.18567365407943726,
|
| 1387 |
+
"learning_rate": 1.4567581203207238e-06,
|
| 1388 |
+
"loss": 0.7588,
|
| 1389 |
+
"step": 1970
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 0.7624181748170966,
|
| 1393 |
+
"grad_norm": 0.19825080037117004,
|
| 1394 |
+
"learning_rate": 1.413037439625593e-06,
|
| 1395 |
+
"loss": 0.7572,
|
| 1396 |
+
"step": 1980
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 0.7662687716596073,
|
| 1400 |
+
"grad_norm": 0.20226572453975677,
|
| 1401 |
+
"learning_rate": 1.3698746695458847e-06,
|
| 1402 |
+
"loss": 0.7467,
|
| 1403 |
+
"step": 1990
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 0.7701193685021178,
|
| 1407 |
+
"grad_norm": 0.17460748553276062,
|
| 1408 |
+
"learning_rate": 1.3272765235521745e-06,
|
| 1409 |
+
"loss": 0.7485,
|
| 1410 |
+
"step": 2000
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 0.7739699653446284,
|
| 1414 |
+
"grad_norm": 0.17261159420013428,
|
| 1415 |
+
"learning_rate": 1.2852496272942877e-06,
|
| 1416 |
+
"loss": 0.7547,
|
| 1417 |
+
"step": 2010
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 0.777820562187139,
|
| 1421 |
+
"grad_norm": 0.1672951579093933,
|
| 1422 |
+
"learning_rate": 1.2438005175707396e-06,
|
| 1423 |
+
"loss": 0.7371,
|
| 1424 |
+
"step": 2020
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 0.7816711590296496,
|
| 1428 |
+
"grad_norm": 0.174899622797966,
|
| 1429 |
+
"learning_rate": 1.2029356413120297e-06,
|
| 1430 |
+
"loss": 0.7308,
|
| 1431 |
+
"step": 2030
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 0.7855217558721602,
|
| 1435 |
+
"grad_norm": 0.1784731149673462,
|
| 1436 |
+
"learning_rate": 1.1626613545778786e-06,
|
| 1437 |
+
"loss": 0.7459,
|
| 1438 |
+
"step": 2040
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 0.7893723527146708,
|
| 1442 |
+
"grad_norm": 0.17027370631694794,
|
| 1443 |
+
"learning_rate": 1.1229839215686294e-06,
|
| 1444 |
+
"loss": 0.7496,
|
| 1445 |
+
"step": 2050
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 0.7932229495571813,
|
| 1449 |
+
"grad_norm": 0.16291554272174835,
|
| 1450 |
+
"learning_rate": 1.0839095136509188e-06,
|
| 1451 |
+
"loss": 0.7334,
|
| 1452 |
+
"step": 2060
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 0.797073546399692,
|
| 1456 |
+
"grad_norm": 0.19742004573345184,
|
| 1457 |
+
"learning_rate": 1.045444208397791e-06,
|
| 1458 |
+
"loss": 0.7503,
|
| 1459 |
+
"step": 2070
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 0.8009241432422025,
|
| 1463 |
+
"grad_norm": 0.1687980443239212,
|
| 1464 |
+
"learning_rate": 1.007593988643402e-06,
|
| 1465 |
+
"loss": 0.7277,
|
| 1466 |
+
"step": 2080
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 0.8047747400847132,
|
| 1470 |
+
"grad_norm": 0.16688750684261322,
|
| 1471 |
+
"learning_rate": 9.703647415524613e-07,
|
| 1472 |
+
"loss": 0.7548,
|
| 1473 |
+
"step": 2090
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 0.8086253369272237,
|
| 1477 |
+
"grad_norm": 0.20113372802734375,
|
| 1478 |
+
"learning_rate": 9.337622577045485e-07,
|
| 1479 |
+
"loss": 0.7467,
|
| 1480 |
+
"step": 2100
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"epoch": 0.8124759337697343,
|
| 1484 |
+
"grad_norm": 0.17319852113723755,
|
| 1485 |
+
"learning_rate": 8.977922301934588e-07,
|
| 1486 |
+
"loss": 0.7492,
|
| 1487 |
+
"step": 2110
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 0.8163265306122449,
|
| 1491 |
+
"grad_norm": 0.18718205392360687,
|
| 1492 |
+
"learning_rate": 8.624602537417009e-07,
|
| 1493 |
+
"loss": 0.7635,
|
| 1494 |
+
"step": 2120
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 0.8201771274547555,
|
| 1498 |
+
"grad_norm": 0.17614679038524628,
|
| 1499 |
+
"learning_rate": 8.277718238303123e-07,
|
| 1500 |
+
"loss": 0.7455,
|
| 1501 |
+
"step": 2130
|
| 1502 |
+
},
|
| 1503 |
+
{
|
| 1504 |
+
"epoch": 0.8240277242972661,
|
| 1505 |
+
"grad_norm": 0.18698488175868988,
|
| 1506 |
+
"learning_rate": 7.937323358440935e-07,
|
| 1507 |
+
"loss": 0.7479,
|
| 1508 |
+
"step": 2140
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 0.8278783211397767,
|
| 1512 |
+
"grad_norm": 0.17944324016571045,
|
| 1513 |
+
"learning_rate": 7.603470842324223e-07,
|
| 1514 |
+
"loss": 0.7359,
|
| 1515 |
+
"step": 2150
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"epoch": 0.8317289179822872,
|
| 1519 |
+
"grad_norm": 0.17073459923267365,
|
| 1520 |
+
"learning_rate": 7.276212616857575e-07,
|
| 1521 |
+
"loss": 0.7576,
|
| 1522 |
+
"step": 2160
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"epoch": 0.8355795148247979,
|
| 1526 |
+
"grad_norm": 0.1806793063879013,
|
| 1527 |
+
"learning_rate": 6.955599583279837e-07,
|
| 1528 |
+
"loss": 0.7503,
|
| 1529 |
+
"step": 2170
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 0.8394301116673084,
|
| 1533 |
+
"grad_norm": 0.17134448885917664,
|
| 1534 |
+
"learning_rate": 6.641681609246981e-07,
|
| 1535 |
+
"loss": 0.7477,
|
| 1536 |
+
"step": 2180
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 0.8432807085098191,
|
| 1540 |
+
"grad_norm": 0.18635715544223785,
|
| 1541 |
+
"learning_rate": 6.334507521075711e-07,
|
| 1542 |
+
"loss": 0.7472,
|
| 1543 |
+
"step": 2190
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 0.8471313053523296,
|
| 1547 |
+
"grad_norm": 0.18426838517189026,
|
| 1548 |
+
"learning_rate": 6.034125096149179e-07,
|
| 1549 |
+
"loss": 0.75,
|
| 1550 |
+
"step": 2200
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 0.8509819021948402,
|
| 1554 |
+
"grad_norm": 0.18644101917743683,
|
| 1555 |
+
"learning_rate": 5.740581055485722e-07,
|
| 1556 |
+
"loss": 0.7218,
|
| 1557 |
+
"step": 2210
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 0.8548324990373508,
|
| 1561 |
+
"grad_norm": 0.17110180854797363,
|
| 1562 |
+
"learning_rate": 5.453921056471961e-07,
|
| 1563 |
+
"loss": 0.7544,
|
| 1564 |
+
"step": 2220
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 0.8586830958798614,
|
| 1568 |
+
"grad_norm": 0.18948741257190704,
|
| 1569 |
+
"learning_rate": 5.174189685761288e-07,
|
| 1570 |
+
"loss": 0.7475,
|
| 1571 |
+
"step": 2230
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 0.862533692722372,
|
| 1575 |
+
"grad_norm": 0.16897393763065338,
|
| 1576 |
+
"learning_rate": 4.901430452338979e-07,
|
| 1577 |
+
"loss": 0.7592,
|
| 1578 |
+
"step": 2240
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 0.8663842895648826,
|
| 1582 |
+
"grad_norm": 0.17033852636814117,
|
| 1583 |
+
"learning_rate": 4.6356857807548286e-07,
|
| 1584 |
+
"loss": 0.7457,
|
| 1585 |
+
"step": 2250
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 0.8702348864073931,
|
| 1589 |
+
"grad_norm": 0.20045344531536102,
|
| 1590 |
+
"learning_rate": 4.376997004524541e-07,
|
| 1591 |
+
"loss": 0.7289,
|
| 1592 |
+
"step": 2260
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 0.8740854832499038,
|
| 1596 |
+
"grad_norm": 0.1762336641550064,
|
| 1597 |
+
"learning_rate": 4.12540435970073e-07,
|
| 1598 |
+
"loss": 0.7477,
|
| 1599 |
+
"step": 2270
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 0.8779360800924143,
|
| 1603 |
+
"grad_norm": 0.17106330394744873,
|
| 1604 |
+
"learning_rate": 3.880946978614703e-07,
|
| 1605 |
+
"loss": 0.7538,
|
| 1606 |
+
"step": 2280
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"epoch": 0.881786676934925,
|
| 1610 |
+
"grad_norm": 0.1719740331172943,
|
| 1611 |
+
"learning_rate": 3.643662883789878e-07,
|
| 1612 |
+
"loss": 0.7562,
|
| 1613 |
+
"step": 2290
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 0.8856372737774355,
|
| 1617 |
+
"grad_norm": 0.16801229119300842,
|
| 1618 |
+
"learning_rate": 3.41358898202781e-07,
|
| 1619 |
+
"loss": 0.738,
|
| 1620 |
+
"step": 2300
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 0.889487870619946,
|
| 1624 |
+
"grad_norm": 0.16703911125659943,
|
| 1625 |
+
"learning_rate": 3.190761058667741e-07,
|
| 1626 |
+
"loss": 0.7489,
|
| 1627 |
+
"step": 2310
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 0.8933384674624567,
|
| 1631 |
+
"grad_norm": 0.16585545241832733,
|
| 1632 |
+
"learning_rate": 2.9752137720206463e-07,
|
| 1633 |
+
"loss": 0.75,
|
| 1634 |
+
"step": 2320
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 0.8971890643049673,
|
| 1638 |
+
"grad_norm": 0.17833101749420166,
|
| 1639 |
+
"learning_rate": 2.7669806479785067e-07,
|
| 1640 |
+
"loss": 0.761,
|
| 1641 |
+
"step": 2330
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 0.9010396611474779,
|
| 1645 |
+
"grad_norm": 0.16858963668346405,
|
| 1646 |
+
"learning_rate": 2.566094074799763e-07,
|
| 1647 |
+
"loss": 0.7513,
|
| 1648 |
+
"step": 2340
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 0.9048902579899885,
|
| 1652 |
+
"grad_norm": 0.17592516541481018,
|
| 1653 |
+
"learning_rate": 2.3725852980716557e-07,
|
| 1654 |
+
"loss": 0.7325,
|
| 1655 |
+
"step": 2350
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 0.908740854832499,
|
| 1659 |
+
"grad_norm": 0.16890260577201843,
|
| 1660 |
+
"learning_rate": 2.1864844158503983e-07,
|
| 1661 |
+
"loss": 0.7486,
|
| 1662 |
+
"step": 2360
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 0.9125914516750097,
|
| 1666 |
+
"grad_norm": 0.19499064981937408,
|
| 1667 |
+
"learning_rate": 2.007820373979702e-07,
|
| 1668 |
+
"loss": 0.7425,
|
| 1669 |
+
"step": 2370
|
| 1670 |
+
},
|
| 1671 |
+
{
|
| 1672 |
+
"epoch": 0.9164420485175202,
|
| 1673 |
+
"grad_norm": 0.16899104416370392,
|
| 1674 |
+
"learning_rate": 1.8366209615886532e-07,
|
| 1675 |
+
"loss": 0.7515,
|
| 1676 |
+
"step": 2380
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 0.9202926453600309,
|
| 1680 |
+
"grad_norm": 0.173374742269516,
|
| 1681 |
+
"learning_rate": 1.6729128067693613e-07,
|
| 1682 |
+
"loss": 0.7645,
|
| 1683 |
+
"step": 2390
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 0.9241432422025414,
|
| 1687 |
+
"grad_norm": 0.19350120425224304,
|
| 1688 |
+
"learning_rate": 1.5167213724353426e-07,
|
| 1689 |
+
"loss": 0.7582,
|
| 1690 |
+
"step": 2400
|
| 1691 |
+
},
|
| 1692 |
+
{
|
| 1693 |
+
"epoch": 0.927993839045052,
|
| 1694 |
+
"grad_norm": 0.16743911802768707,
|
| 1695 |
+
"learning_rate": 1.3680709523610058e-07,
|
| 1696 |
+
"loss": 0.7365,
|
| 1697 |
+
"step": 2410
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"epoch": 0.9318444358875626,
|
| 1701 |
+
"grad_norm": 0.17094437777996063,
|
| 1702 |
+
"learning_rate": 1.2269846674030418e-07,
|
| 1703 |
+
"loss": 0.7664,
|
| 1704 |
+
"step": 2420
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"epoch": 0.9356950327300732,
|
| 1708 |
+
"grad_norm": 0.172334223985672,
|
| 1709 |
+
"learning_rate": 1.0934844619042595e-07,
|
| 1710 |
+
"loss": 0.759,
|
| 1711 |
+
"step": 2430
|
| 1712 |
+
},
|
| 1713 |
+
{
|
| 1714 |
+
"epoch": 0.9395456295725837,
|
| 1715 |
+
"grad_norm": 0.1744299679994583,
|
| 1716 |
+
"learning_rate": 9.675911002803961e-08,
|
| 1717 |
+
"loss": 0.7491,
|
| 1718 |
+
"step": 2440
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 0.9433962264150944,
|
| 1722 |
+
"grad_norm": 0.18744736909866333,
|
| 1723 |
+
"learning_rate": 8.493241637904381e-08,
|
| 1724 |
+
"loss": 0.7372,
|
| 1725 |
+
"step": 2450
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 0.9472468232576049,
|
| 1729 |
+
"grad_norm": 0.19414865970611572,
|
| 1730 |
+
"learning_rate": 7.387020474909689e-08,
|
| 1731 |
+
"loss": 0.7643,
|
| 1732 |
+
"step": 2460
|
| 1733 |
+
},
|
| 1734 |
+
{
|
| 1735 |
+
"epoch": 0.9510974201001156,
|
| 1736 |
+
"grad_norm": 0.20816007256507874,
|
| 1737 |
+
"learning_rate": 6.357419573750856e-08,
|
| 1738 |
+
"loss": 0.7589,
|
| 1739 |
+
"step": 2470
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 0.9549480169426261,
|
| 1743 |
+
"grad_norm": 0.1783704161643982,
|
| 1744 |
+
"learning_rate": 5.4045990769613365e-08,
|
| 1745 |
+
"loss": 0.7444,
|
| 1746 |
+
"step": 2480
|
| 1747 |
+
},
|
| 1748 |
+
{
|
| 1749 |
+
"epoch": 0.9587986137851366,
|
| 1750 |
+
"grad_norm": 0.17478413879871368,
|
| 1751 |
+
"learning_rate": 4.5287071847694434e-08,
|
| 1752 |
+
"loss": 0.7468,
|
| 1753 |
+
"step": 2490
|
| 1754 |
+
},
|
| 1755 |
+
{
|
| 1756 |
+
"epoch": 0.9626492106276473,
|
| 1757 |
+
"grad_norm": 0.1698388308286667,
|
| 1758 |
+
"learning_rate": 3.7298801320467817e-08,
|
| 1759 |
+
"loss": 0.7488,
|
| 1760 |
+
"step": 2500
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 0.9664998074701578,
|
| 1764 |
+
"grad_norm": 0.16845713555812836,
|
| 1765 |
+
"learning_rate": 3.0082421671192576e-08,
|
| 1766 |
+
"loss": 0.748,
|
| 1767 |
+
"step": 2510
|
| 1768 |
+
},
|
| 1769 |
+
{
|
| 1770 |
+
"epoch": 0.9703504043126685,
|
| 1771 |
+
"grad_norm": 0.1755690723657608,
|
| 1772 |
+
"learning_rate": 2.3639055324410355e-08,
|
| 1773 |
+
"loss": 0.7617,
|
| 1774 |
+
"step": 2520
|
| 1775 |
+
},
|
| 1776 |
+
{
|
| 1777 |
+
"epoch": 0.974201001155179,
|
| 1778 |
+
"grad_norm": 0.17934748530387878,
|
| 1779 |
+
"learning_rate": 1.7969704471369477e-08,
|
| 1780 |
+
"loss": 0.7425,
|
| 1781 |
+
"step": 2530
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 0.9780515979976896,
|
| 1785 |
+
"grad_norm": 0.16723528504371643,
|
| 1786 |
+
"learning_rate": 1.3075250914142413e-08,
|
| 1787 |
+
"loss": 0.7421,
|
| 1788 |
+
"step": 2540
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"epoch": 0.9819021948402002,
|
| 1792 |
+
"grad_norm": 0.16724160313606262,
|
| 1793 |
+
"learning_rate": 8.956455928476049e-09,
|
| 1794 |
+
"loss": 0.7573,
|
| 1795 |
+
"step": 2550
|
| 1796 |
+
},
|
| 1797 |
+
{
|
| 1798 |
+
"epoch": 0.9857527916827108,
|
| 1799 |
+
"grad_norm": 0.16738183796405792,
|
| 1800 |
+
"learning_rate": 5.6139601453786275e-09,
|
| 1801 |
+
"loss": 0.7431,
|
| 1802 |
+
"step": 2560
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 0.9896033885252215,
|
| 1806 |
+
"grad_norm": 0.17542411386966705,
|
| 1807 |
+
"learning_rate": 3.0482834514805602e-09,
|
| 1808 |
+
"loss": 0.7387,
|
| 1809 |
+
"step": 2570
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 0.993453985367732,
|
| 1813 |
+
"grad_norm": 0.17070399224758148,
|
| 1814 |
+
"learning_rate": 1.2598249081713366e-09,
|
| 1815 |
+
"loss": 0.73,
|
| 1816 |
+
"step": 2580
|
| 1817 |
+
},
|
| 1818 |
+
{
|
| 1819 |
+
"epoch": 0.9973045822102425,
|
| 1820 |
+
"grad_norm": 0.17743350565433502,
|
| 1821 |
+
"learning_rate": 2.488626895302826e-10,
|
| 1822 |
+
"loss": 0.7624,
|
| 1823 |
+
"step": 2590
|
| 1824 |
+
},
|
| 1825 |
+
{
|
| 1826 |
+
"epoch": 1.0,
|
| 1827 |
+
"step": 2597,
|
| 1828 |
+
"total_flos": 5.969420038266946e+18,
|
| 1829 |
+
"train_loss": 0.7636532414450662,
|
| 1830 |
+
"train_runtime": 14664.736,
|
| 1831 |
+
"train_samples_per_second": 45.34,
|
| 1832 |
+
"train_steps_per_second": 0.177
|
| 1833 |
+
}
|
| 1834 |
+
],
|
| 1835 |
+
"logging_steps": 10,
|
| 1836 |
+
"max_steps": 2597,
|
| 1837 |
+
"num_input_tokens_seen": 584317136,
|
| 1838 |
+
"num_train_epochs": 1,
|
| 1839 |
+
"save_steps": 500,
|
| 1840 |
+
"stateful_callbacks": {
|
| 1841 |
+
"TrainerControl": {
|
| 1842 |
+
"args": {
|
| 1843 |
+
"should_epoch_stop": false,
|
| 1844 |
+
"should_evaluate": false,
|
| 1845 |
+
"should_log": false,
|
| 1846 |
+
"should_save": true,
|
| 1847 |
+
"should_training_stop": true
|
| 1848 |
+
},
|
| 1849 |
+
"attributes": {}
|
| 1850 |
+
}
|
| 1851 |
+
},
|
| 1852 |
+
"total_flos": 5.969420038266946e+18,
|
| 1853 |
+
"train_batch_size": 16,
|
| 1854 |
+
"trial_name": null,
|
| 1855 |
+
"trial_params": null
|
| 1856 |
+
}
|
continuous/smollm2/mc1_b16_224_400m/finetune/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:90158f95fc9655ce3baa2a75ce7a0fd901588268b573d1f46e5002a8adb769d0
|
| 3 |
+
size 6264
|
continuous/smollm2/mc1_b16_224_400m/finetune/vocab.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
continuous/smollm2/mc1_b16_224_400m/pretrain/config.json
ADDED
|
@@ -0,0 +1,70 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_faster_video": false,
|
| 3 |
+
"add_time_instruction": false,
|
| 4 |
+
"architectures": [
|
| 5 |
+
"LlamaForCausalLM"
|
| 6 |
+
],
|
| 7 |
+
"attention_bias": false,
|
| 8 |
+
"attention_dropout": 0.0,
|
| 9 |
+
"bos_token_id": 1,
|
| 10 |
+
"dtype": "bfloat16",
|
| 11 |
+
"eos_token_id": 2,
|
| 12 |
+
"faster_token_stride": 10,
|
| 13 |
+
"force_sample": false,
|
| 14 |
+
"head_dim": 64,
|
| 15 |
+
"hidden_act": "silu",
|
| 16 |
+
"hidden_size": 2048,
|
| 17 |
+
"image_aspect_ratio": "square",
|
| 18 |
+
"image_crop_resolution": null,
|
| 19 |
+
"image_grid_pinpoints": null,
|
| 20 |
+
"image_split_resolution": null,
|
| 21 |
+
"initializer_range": 0.02,
|
| 22 |
+
"intermediate_size": 8192,
|
| 23 |
+
"max_position_embeddings": 8192,
|
| 24 |
+
"mlp_bias": false,
|
| 25 |
+
"mm_hidden_size": 768,
|
| 26 |
+
"mm_newline_position": "grid",
|
| 27 |
+
"mm_patch_merge_type": "flat",
|
| 28 |
+
"mm_projector_lr": null,
|
| 29 |
+
"mm_projector_type": "mlp2x_gelu",
|
| 30 |
+
"mm_resampler_type": null,
|
| 31 |
+
"mm_spatial_pool_mode": "bilinear",
|
| 32 |
+
"mm_spatial_pool_stride": null,
|
| 33 |
+
"mm_tunable_parts": "mm_mlp_adapter",
|
| 34 |
+
"mm_use_im_patch_token": false,
|
| 35 |
+
"mm_use_im_start_end": false,
|
| 36 |
+
"mm_vision_select_feature": "patch",
|
| 37 |
+
"mm_vision_select_layer": -2,
|
| 38 |
+
"mm_vision_tower": "open_clip_hub:ViT-B-16",
|
| 39 |
+
"mm_vision_tower_lr": null,
|
| 40 |
+
"model_type": "llama",
|
| 41 |
+
"num_attention_heads": 32,
|
| 42 |
+
"num_hidden_layers": 24,
|
| 43 |
+
"num_key_value_heads": 32,
|
| 44 |
+
"pad_token_id": 0,
|
| 45 |
+
"pos_skipping_range": 4096,
|
| 46 |
+
"pretraining_tp": 1,
|
| 47 |
+
"rms_norm_eps": 1e-05,
|
| 48 |
+
"rope_scaling": null,
|
| 49 |
+
"rope_theta": 130000,
|
| 50 |
+
"tie_word_embeddings": true,
|
| 51 |
+
"tokenizer_model_max_length": 8192,
|
| 52 |
+
"tokenizer_padding_side": "right",
|
| 53 |
+
"transformers.js_config": {
|
| 54 |
+
"dtype": "q4",
|
| 55 |
+
"kv_cache_dtype": {
|
| 56 |
+
"fp16": "float16",
|
| 57 |
+
"q4f16": "float16"
|
| 58 |
+
},
|
| 59 |
+
"use_external_data_format": {
|
| 60 |
+
"model.onnx": true,
|
| 61 |
+
"model_fp16.onnx": true
|
| 62 |
+
}
|
| 63 |
+
},
|
| 64 |
+
"transformers_version": "4.57.6",
|
| 65 |
+
"use_cache": true,
|
| 66 |
+
"use_mm_proj": true,
|
| 67 |
+
"use_pos_skipping": false,
|
| 68 |
+
"vision_tower_pretrained": "/cache/ckpt/download/tokenizer/continuous/mc1/mc1_b16_224_400m.pt",
|
| 69 |
+
"vocab_size": 49152
|
| 70 |
+
}
|
continuous/smollm2/mc1_b16_224_400m/pretrain/mm_projector.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9699ea23327885cc5ece8143a7cd76776e5f3ac792d837d96192673562149aba
|
| 3 |
+
size 23087160
|
continuous/smollm2/mc1_b16_224_400m/pretrain/trainer_state.json
ADDED
|
@@ -0,0 +1,421 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 1.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 545,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.01834862385321101,
|
| 14 |
+
"grad_norm": 1.6609845161437988,
|
| 15 |
+
"learning_rate": 0.00032142857142857147,
|
| 16 |
+
"loss": 3.6939,
|
| 17 |
+
"step": 10
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.03669724770642202,
|
| 21 |
+
"grad_norm": 0.27033036947250366,
|
| 22 |
+
"learning_rate": 0.0006785714285714287,
|
| 23 |
+
"loss": 2.9534,
|
| 24 |
+
"step": 20
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.05504587155963303,
|
| 28 |
+
"grad_norm": 0.11400279402732849,
|
| 29 |
+
"learning_rate": 0.0009999907688176173,
|
| 30 |
+
"loss": 2.8001,
|
| 31 |
+
"step": 30
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.07339449541284404,
|
| 35 |
+
"grad_norm": 0.12151011824607849,
|
| 36 |
+
"learning_rate": 0.0009988834393115766,
|
| 37 |
+
"loss": 2.7469,
|
| 38 |
+
"step": 40
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.09174311926605505,
|
| 42 |
+
"grad_norm": 0.09218265861272812,
|
| 43 |
+
"learning_rate": 0.0009959345572940772,
|
| 44 |
+
"loss": 2.7233,
|
| 45 |
+
"step": 50
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.11009174311926606,
|
| 49 |
+
"grad_norm": 0.11360713094472885,
|
| 50 |
+
"learning_rate": 0.0009911550081155983,
|
| 51 |
+
"loss": 2.6839,
|
| 52 |
+
"step": 60
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.12844036697247707,
|
| 56 |
+
"grad_norm": 0.12789812684059143,
|
| 57 |
+
"learning_rate": 0.0009845624347566433,
|
| 58 |
+
"loss": 2.6475,
|
| 59 |
+
"step": 70
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.14678899082568808,
|
| 63 |
+
"grad_norm": 0.16777727007865906,
|
| 64 |
+
"learning_rate": 0.0009761811727013547,
|
| 65 |
+
"loss": 2.6109,
|
| 66 |
+
"step": 80
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.1651376146788991,
|
| 70 |
+
"grad_norm": 0.11202109605073929,
|
| 71 |
+
"learning_rate": 0.0009660421601067666,
|
| 72 |
+
"loss": 2.5729,
|
| 73 |
+
"step": 90
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.1834862385321101,
|
| 77 |
+
"grad_norm": 0.1622268259525299,
|
| 78 |
+
"learning_rate": 0.0009541828235992926,
|
| 79 |
+
"loss": 2.5455,
|
| 80 |
+
"step": 100
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.2018348623853211,
|
| 84 |
+
"grad_norm": 0.12131012976169586,
|
| 85 |
+
"learning_rate": 0.0009406469401200151,
|
| 86 |
+
"loss": 2.5089,
|
| 87 |
+
"step": 110
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.22018348623853212,
|
| 91 |
+
"grad_norm": 0.14192640781402588,
|
| 92 |
+
"learning_rate": 0.0009254844753287492,
|
| 93 |
+
"loss": 2.4861,
|
| 94 |
+
"step": 120
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.23853211009174313,
|
| 98 |
+
"grad_norm": 0.11952310800552368,
|
| 99 |
+
"learning_rate": 0.0009087513991633924,
|
| 100 |
+
"loss": 2.4478,
|
| 101 |
+
"step": 130
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.25688073394495414,
|
| 105 |
+
"grad_norm": 0.13647273182868958,
|
| 106 |
+
"learning_rate": 0.0008905094792353917,
|
| 107 |
+
"loss": 2.4142,
|
| 108 |
+
"step": 140
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.27522935779816515,
|
| 112 |
+
"grad_norm": 0.1494603455066681,
|
| 113 |
+
"learning_rate": 0.0008708260528239789,
|
| 114 |
+
"loss": 2.3956,
|
| 115 |
+
"step": 150
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.29357798165137616,
|
| 119 |
+
"grad_norm": 0.14114491641521454,
|
| 120 |
+
"learning_rate": 0.0008497737783108237,
|
| 121 |
+
"loss": 2.361,
|
| 122 |
+
"step": 160
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.3119266055045872,
|
| 126 |
+
"grad_norm": 0.09423563629388809,
|
| 127 |
+
"learning_rate": 0.0008274303669726426,
|
| 128 |
+
"loss": 2.3422,
|
| 129 |
+
"step": 170
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.3302752293577982,
|
| 133 |
+
"grad_norm": 0.1314387023448944,
|
| 134 |
+
"learning_rate": 0.0008038782961218137,
|
| 135 |
+
"loss": 2.3263,
|
| 136 |
+
"step": 180
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.3486238532110092,
|
| 140 |
+
"grad_norm": 0.10563050955533981,
|
| 141 |
+
"learning_rate": 0.000779204504653894,
|
| 142 |
+
"loss": 2.3115,
|
| 143 |
+
"step": 190
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.3669724770642202,
|
| 147 |
+
"grad_norm": 0.11541625112295151,
|
| 148 |
+
"learning_rate": 0.0007535000721258791,
|
| 149 |
+
"loss": 2.2837,
|
| 150 |
+
"step": 200
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.3853211009174312,
|
| 154 |
+
"grad_norm": 0.1178816556930542,
|
| 155 |
+
"learning_rate": 0.00072685988254984,
|
| 156 |
+
"loss": 2.2797,
|
| 157 |
+
"step": 210
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.4036697247706422,
|
| 161 |
+
"grad_norm": 0.14665955305099487,
|
| 162 |
+
"learning_rate": 0.0006993822741429906,
|
| 163 |
+
"loss": 2.2852,
|
| 164 |
+
"step": 220
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.42201834862385323,
|
| 168 |
+
"grad_norm": 0.10501314699649811,
|
| 169 |
+
"learning_rate": 0.0006711686763270818,
|
| 170 |
+
"loss": 2.2737,
|
| 171 |
+
"step": 230
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.44036697247706424,
|
| 175 |
+
"grad_norm": 0.08842355012893677,
|
| 176 |
+
"learning_rate": 0.0006423232353170831,
|
| 177 |
+
"loss": 2.264,
|
| 178 |
+
"step": 240
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.45871559633027525,
|
| 182 |
+
"grad_norm": 0.10759062319993973,
|
| 183 |
+
"learning_rate": 0.0006129524296812335,
|
| 184 |
+
"loss": 2.264,
|
| 185 |
+
"step": 250
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.47706422018348627,
|
| 189 |
+
"grad_norm": 0.1079496219754219,
|
| 190 |
+
"learning_rate": 0.0005831646772915651,
|
| 191 |
+
"loss": 2.2684,
|
| 192 |
+
"step": 260
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.4954128440366973,
|
| 196 |
+
"grad_norm": 0.11650674045085907,
|
| 197 |
+
"learning_rate": 0.0005530699351157809,
|
| 198 |
+
"loss": 2.2451,
|
| 199 |
+
"step": 270
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.5137614678899083,
|
| 203 |
+
"grad_norm": 0.10438396781682968,
|
| 204 |
+
"learning_rate": 0.0005227792933277943,
|
| 205 |
+
"loss": 2.2512,
|
| 206 |
+
"step": 280
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.5321100917431193,
|
| 210 |
+
"grad_norm": 0.08998076617717743,
|
| 211 |
+
"learning_rate": 0.0004924045652352092,
|
| 212 |
+
"loss": 2.2368,
|
| 213 |
+
"step": 290
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.5504587155963303,
|
| 217 |
+
"grad_norm": 0.09093569964170456,
|
| 218 |
+
"learning_rate": 0.0004620578745374633,
|
| 219 |
+
"loss": 2.2368,
|
| 220 |
+
"step": 300
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.5688073394495413,
|
| 224 |
+
"grad_norm": 0.124798484146595,
|
| 225 |
+
"learning_rate": 0.0004318512414382059,
|
| 226 |
+
"loss": 2.2411,
|
| 227 |
+
"step": 310
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.5871559633027523,
|
| 231 |
+
"grad_norm": 0.09100601822137833,
|
| 232 |
+
"learning_rate": 0.0004018961691397148,
|
| 233 |
+
"loss": 2.2415,
|
| 234 |
+
"step": 320
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.6055045871559633,
|
| 238 |
+
"grad_norm": 0.10746273398399353,
|
| 239 |
+
"learning_rate": 0.0003723032322457458,
|
| 240 |
+
"loss": 2.2285,
|
| 241 |
+
"step": 330
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.6238532110091743,
|
| 245 |
+
"grad_norm": 0.13037382066249847,
|
| 246 |
+
"learning_rate": 0.0003431816685921625,
|
| 247 |
+
"loss": 2.2293,
|
| 248 |
+
"step": 340
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.6422018348623854,
|
| 252 |
+
"grad_norm": 0.10620160400867462,
|
| 253 |
+
"learning_rate": 0.0003146389760120416,
|
| 254 |
+
"loss": 2.2283,
|
| 255 |
+
"step": 350
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.6605504587155964,
|
| 259 |
+
"grad_norm": 0.10957541316747665,
|
| 260 |
+
"learning_rate": 0.00028678051552373485,
|
| 261 |
+
"loss": 2.2297,
|
| 262 |
+
"step": 360
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.6788990825688074,
|
| 266 |
+
"grad_norm": 0.08884303271770477,
|
| 267 |
+
"learning_rate": 0.0002597091224066581,
|
| 268 |
+
"loss": 2.2252,
|
| 269 |
+
"step": 370
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.6972477064220184,
|
| 273 |
+
"grad_norm": 0.08508676290512085,
|
| 274 |
+
"learning_rate": 0.0002335247266004629,
|
| 275 |
+
"loss": 2.231,
|
| 276 |
+
"step": 380
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.7155963302752294,
|
| 280 |
+
"grad_norm": 0.08625798672437668,
|
| 281 |
+
"learning_rate": 0.000208323983828828,
|
| 282 |
+
"loss": 2.2225,
|
| 283 |
+
"step": 390
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.7339449541284404,
|
| 287 |
+
"grad_norm": 0.08349300920963287,
|
| 288 |
+
"learning_rate": 0.0001841999188095224,
|
| 289 |
+
"loss": 2.2141,
|
| 290 |
+
"step": 400
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.7522935779816514,
|
| 294 |
+
"grad_norm": 0.08350767940282822,
|
| 295 |
+
"learning_rate": 0.00016124158186777676,
|
| 296 |
+
"loss": 2.2201,
|
| 297 |
+
"step": 410
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.7706422018348624,
|
| 301 |
+
"grad_norm": 0.09674710035324097,
|
| 302 |
+
"learning_rate": 0.0001395337202205211,
|
| 303 |
+
"loss": 2.2161,
|
| 304 |
+
"step": 420
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 0.7889908256880734,
|
| 308 |
+
"grad_norm": 0.09312637150287628,
|
| 309 |
+
"learning_rate": 0.00011915646514489292,
|
| 310 |
+
"loss": 2.2126,
|
| 311 |
+
"step": 430
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.8073394495412844,
|
| 315 |
+
"grad_norm": 0.09449267387390137,
|
| 316 |
+
"learning_rate": 0.00010018503618578818,
|
| 317 |
+
"loss": 2.2225,
|
| 318 |
+
"step": 440
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.8256880733944955,
|
| 322 |
+
"grad_norm": 0.08690215647220612,
|
| 323 |
+
"learning_rate": 8.268946349432583e-05,
|
| 324 |
+
"loss": 2.2135,
|
| 325 |
+
"step": 450
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.8440366972477065,
|
| 329 |
+
"grad_norm": 0.08425020426511765,
|
| 330 |
+
"learning_rate": 6.673432932217117e-05,
|
| 331 |
+
"loss": 2.2091,
|
| 332 |
+
"step": 460
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.8623853211009175,
|
| 336 |
+
"grad_norm": 0.09241657704114914,
|
| 337 |
+
"learning_rate": 5.2378529625954684e-05,
|
| 338 |
+
"loss": 2.2019,
|
| 339 |
+
"step": 470
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.8807339449541285,
|
| 343 |
+
"grad_norm": 0.08600675314664841,
|
| 344 |
+
"learning_rate": 3.9675056661785556e-05,
|
| 345 |
+
"loss": 2.2033,
|
| 346 |
+
"step": 480
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.8990825688073395,
|
| 350 |
+
"grad_norm": 0.0849154144525528,
|
| 351 |
+
"learning_rate": 2.8670803372379538e-05,
|
| 352 |
+
"loss": 2.2112,
|
| 353 |
+
"step": 490
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.9174311926605505,
|
| 357 |
+
"grad_norm": 0.08045026659965515,
|
| 358 |
+
"learning_rate": 1.9406390288876586e-05,
|
| 359 |
+
"loss": 2.2156,
|
| 360 |
+
"step": 500
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.9357798165137615,
|
| 364 |
+
"grad_norm": 0.08131886273622513,
|
| 365 |
+
"learning_rate": 1.191601558631461e-05,
|
| 366 |
+
"loss": 2.206,
|
| 367 |
+
"step": 510
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.9541284403669725,
|
| 371 |
+
"grad_norm": 0.0832924172282219,
|
| 372 |
+
"learning_rate": 6.2273288462559305e-06,
|
| 373 |
+
"loss": 2.2051,
|
| 374 |
+
"step": 520
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.9724770642201835,
|
| 378 |
+
"grad_norm": 0.08890211582183838,
|
| 379 |
+
"learning_rate": 2.361328992552314e-06,
|
| 380 |
+
"loss": 2.207,
|
| 381 |
+
"step": 530
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.9908256880733946,
|
| 385 |
+
"grad_norm": 0.08312161266803741,
|
| 386 |
+
"learning_rate": 3.322867770044202e-07,
|
| 387 |
+
"loss": 2.2036,
|
| 388 |
+
"step": 540
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 1.0,
|
| 392 |
+
"step": 545,
|
| 393 |
+
"total_flos": 3.540700722154701e+17,
|
| 394 |
+
"train_loss": 2.3636895485974234,
|
| 395 |
+
"train_runtime": 1382.7808,
|
| 396 |
+
"train_samples_per_second": 403.627,
|
| 397 |
+
"train_steps_per_second": 0.394
|
| 398 |
+
}
|
| 399 |
+
],
|
| 400 |
+
"logging_steps": 10,
|
| 401 |
+
"max_steps": 545,
|
| 402 |
+
"num_input_tokens_seen": 34658176,
|
| 403 |
+
"num_train_epochs": 1,
|
| 404 |
+
"save_steps": 500,
|
| 405 |
+
"stateful_callbacks": {
|
| 406 |
+
"TrainerControl": {
|
| 407 |
+
"args": {
|
| 408 |
+
"should_epoch_stop": false,
|
| 409 |
+
"should_evaluate": false,
|
| 410 |
+
"should_log": false,
|
| 411 |
+
"should_save": false,
|
| 412 |
+
"should_training_stop": false
|
| 413 |
+
},
|
| 414 |
+
"attributes": {}
|
| 415 |
+
}
|
| 416 |
+
},
|
| 417 |
+
"total_flos": 3.540700722154701e+17,
|
| 418 |
+
"train_batch_size": 64,
|
| 419 |
+
"trial_name": null,
|
| 420 |
+
"trial_params": null
|
| 421 |
+
}
|
continuous/smollm2/mc1_b32_224_2.5b/finetune/chat_template.jinja
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{% for message in messages %}{% if loop.first and messages[0]['role'] != 'system' %}{{ '<|im_start|>system
|
| 2 |
+
You are a helpful AI assistant named SmolLM, trained by Hugging Face<|im_end|>
|
| 3 |
+
' }}{% endif %}{{'<|im_start|>' + message['role'] + '
|
| 4 |
+
' + message['content'] + '<|im_end|>' + '
|
| 5 |
+
'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant
|
| 6 |
+
' }}{% endif %}
|
continuous/smollm2/mc1_b32_224_2.5b/finetune/config.json
ADDED
|
@@ -0,0 +1,70 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_faster_video": false,
|
| 3 |
+
"add_time_instruction": false,
|
| 4 |
+
"architectures": [
|
| 5 |
+
"LlavaLlamaForCausalLM"
|
| 6 |
+
],
|
| 7 |
+
"attention_bias": false,
|
| 8 |
+
"attention_dropout": 0.0,
|
| 9 |
+
"bos_token_id": 1,
|
| 10 |
+
"dtype": "bfloat16",
|
| 11 |
+
"eos_token_id": 2,
|
| 12 |
+
"faster_token_stride": 10,
|
| 13 |
+
"force_sample": false,
|
| 14 |
+
"head_dim": 64,
|
| 15 |
+
"hidden_act": "silu",
|
| 16 |
+
"hidden_size": 2048,
|
| 17 |
+
"image_aspect_ratio": "square",
|
| 18 |
+
"image_crop_resolution": null,
|
| 19 |
+
"image_grid_pinpoints": null,
|
| 20 |
+
"image_split_resolution": null,
|
| 21 |
+
"initializer_range": 0.02,
|
| 22 |
+
"intermediate_size": 8192,
|
| 23 |
+
"max_position_embeddings": 8192,
|
| 24 |
+
"mlp_bias": false,
|
| 25 |
+
"mm_hidden_size": 768,
|
| 26 |
+
"mm_newline_position": "grid",
|
| 27 |
+
"mm_patch_merge_type": "flat",
|
| 28 |
+
"mm_projector_lr": 0.0001,
|
| 29 |
+
"mm_projector_type": "mlp2x_gelu",
|
| 30 |
+
"mm_resampler_type": null,
|
| 31 |
+
"mm_spatial_pool_mode": "bilinear",
|
| 32 |
+
"mm_spatial_pool_stride": null,
|
| 33 |
+
"mm_tunable_parts": "mm_mlp_adapter,mm_language_model",
|
| 34 |
+
"mm_use_im_patch_token": false,
|
| 35 |
+
"mm_use_im_start_end": false,
|
| 36 |
+
"mm_vision_select_feature": "patch",
|
| 37 |
+
"mm_vision_select_layer": -2,
|
| 38 |
+
"mm_vision_tower": "open_clip_hub:ViT-B-32",
|
| 39 |
+
"mm_vision_tower_lr": null,
|
| 40 |
+
"model_type": "llama",
|
| 41 |
+
"num_attention_heads": 32,
|
| 42 |
+
"num_hidden_layers": 24,
|
| 43 |
+
"num_key_value_heads": 32,
|
| 44 |
+
"pad_token_id": 0,
|
| 45 |
+
"pos_skipping_range": 4096,
|
| 46 |
+
"pretraining_tp": 1,
|
| 47 |
+
"rms_norm_eps": 1e-05,
|
| 48 |
+
"rope_scaling": null,
|
| 49 |
+
"rope_theta": 130000,
|
| 50 |
+
"tie_word_embeddings": true,
|
| 51 |
+
"tokenizer_model_max_length": 8192,
|
| 52 |
+
"tokenizer_padding_side": "right",
|
| 53 |
+
"transformers.js_config": {
|
| 54 |
+
"dtype": "q4",
|
| 55 |
+
"kv_cache_dtype": {
|
| 56 |
+
"fp16": "float16",
|
| 57 |
+
"q4f16": "float16"
|
| 58 |
+
},
|
| 59 |
+
"use_external_data_format": {
|
| 60 |
+
"model.onnx": true,
|
| 61 |
+
"model_fp16.onnx": true
|
| 62 |
+
}
|
| 63 |
+
},
|
| 64 |
+
"transformers_version": "4.57.6",
|
| 65 |
+
"use_cache": true,
|
| 66 |
+
"use_mm_proj": true,
|
| 67 |
+
"use_pos_skipping": false,
|
| 68 |
+
"vision_tower_pretrained": "/cache/ckpt/download/tokenizer/continuous/mc1/mc1_b32_224_2.5b.pt",
|
| 69 |
+
"vocab_size": 49152
|
| 70 |
+
}
|
continuous/smollm2/mc1_b32_224_2.5b/finetune/generation_config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 1,
|
| 4 |
+
"eos_token_id": [
|
| 5 |
+
2
|
| 6 |
+
],
|
| 7 |
+
"pad_token_id": 0,
|
| 8 |
+
"transformers_version": "4.57.6"
|
| 9 |
+
}
|
continuous/smollm2/mc1_b32_224_2.5b/finetune/merges.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
continuous/smollm2/mc1_b32_224_2.5b/finetune/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5d92d8976c8e398a43edda0a5b35d0e35ec2b76db50c0baa8183b04d016e56ae
|
| 3 |
+
size 3822910424
|
continuous/smollm2/mc1_b32_224_2.5b/finetune/special_tokens_map.json
ADDED
|
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"additional_special_tokens": [
|
| 3 |
+
"<|im_start|>",
|
| 4 |
+
"<|im_end|>"
|
| 5 |
+
],
|
| 6 |
+
"bos_token": {
|
| 7 |
+
"content": "<|im_start|>",
|
| 8 |
+
"lstrip": false,
|
| 9 |
+
"normalized": false,
|
| 10 |
+
"rstrip": false,
|
| 11 |
+
"single_word": false
|
| 12 |
+
},
|
| 13 |
+
"eos_token": {
|
| 14 |
+
"content": "<|im_end|>",
|
| 15 |
+
"lstrip": false,
|
| 16 |
+
"normalized": false,
|
| 17 |
+
"rstrip": false,
|
| 18 |
+
"single_word": false
|
| 19 |
+
},
|
| 20 |
+
"pad_token": "<|endoftext|>",
|
| 21 |
+
"unk_token": {
|
| 22 |
+
"content": "<|endoftext|>",
|
| 23 |
+
"lstrip": false,
|
| 24 |
+
"normalized": false,
|
| 25 |
+
"rstrip": false,
|
| 26 |
+
"single_word": false
|
| 27 |
+
}
|
| 28 |
+
}
|
continuous/smollm2/mc1_b32_224_2.5b/finetune/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
continuous/smollm2/mc1_b32_224_2.5b/finetune/tokenizer_config.json
ADDED
|
@@ -0,0 +1,155 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_prefix_space": false,
|
| 3 |
+
"added_tokens_decoder": {
|
| 4 |
+
"0": {
|
| 5 |
+
"content": "<|endoftext|>",
|
| 6 |
+
"lstrip": false,
|
| 7 |
+
"normalized": false,
|
| 8 |
+
"rstrip": false,
|
| 9 |
+
"single_word": false,
|
| 10 |
+
"special": true
|
| 11 |
+
},
|
| 12 |
+
"1": {
|
| 13 |
+
"content": "<|im_start|>",
|
| 14 |
+
"lstrip": false,
|
| 15 |
+
"normalized": false,
|
| 16 |
+
"rstrip": false,
|
| 17 |
+
"single_word": false,
|
| 18 |
+
"special": true
|
| 19 |
+
},
|
| 20 |
+
"2": {
|
| 21 |
+
"content": "<|im_end|>",
|
| 22 |
+
"lstrip": false,
|
| 23 |
+
"normalized": false,
|
| 24 |
+
"rstrip": false,
|
| 25 |
+
"single_word": false,
|
| 26 |
+
"special": true
|
| 27 |
+
},
|
| 28 |
+
"3": {
|
| 29 |
+
"content": "<repo_name>",
|
| 30 |
+
"lstrip": false,
|
| 31 |
+
"normalized": false,
|
| 32 |
+
"rstrip": false,
|
| 33 |
+
"single_word": false,
|
| 34 |
+
"special": true
|
| 35 |
+
},
|
| 36 |
+
"4": {
|
| 37 |
+
"content": "<reponame>",
|
| 38 |
+
"lstrip": false,
|
| 39 |
+
"normalized": false,
|
| 40 |
+
"rstrip": false,
|
| 41 |
+
"single_word": false,
|
| 42 |
+
"special": true
|
| 43 |
+
},
|
| 44 |
+
"5": {
|
| 45 |
+
"content": "<file_sep>",
|
| 46 |
+
"lstrip": false,
|
| 47 |
+
"normalized": false,
|
| 48 |
+
"rstrip": false,
|
| 49 |
+
"single_word": false,
|
| 50 |
+
"special": true
|
| 51 |
+
},
|
| 52 |
+
"6": {
|
| 53 |
+
"content": "<filename>",
|
| 54 |
+
"lstrip": false,
|
| 55 |
+
"normalized": false,
|
| 56 |
+
"rstrip": false,
|
| 57 |
+
"single_word": false,
|
| 58 |
+
"special": true
|
| 59 |
+
},
|
| 60 |
+
"7": {
|
| 61 |
+
"content": "<gh_stars>",
|
| 62 |
+
"lstrip": false,
|
| 63 |
+
"normalized": false,
|
| 64 |
+
"rstrip": false,
|
| 65 |
+
"single_word": false,
|
| 66 |
+
"special": true
|
| 67 |
+
},
|
| 68 |
+
"8": {
|
| 69 |
+
"content": "<issue_start>",
|
| 70 |
+
"lstrip": false,
|
| 71 |
+
"normalized": false,
|
| 72 |
+
"rstrip": false,
|
| 73 |
+
"single_word": false,
|
| 74 |
+
"special": true
|
| 75 |
+
},
|
| 76 |
+
"9": {
|
| 77 |
+
"content": "<issue_comment>",
|
| 78 |
+
"lstrip": false,
|
| 79 |
+
"normalized": false,
|
| 80 |
+
"rstrip": false,
|
| 81 |
+
"single_word": false,
|
| 82 |
+
"special": true
|
| 83 |
+
},
|
| 84 |
+
"10": {
|
| 85 |
+
"content": "<issue_closed>",
|
| 86 |
+
"lstrip": false,
|
| 87 |
+
"normalized": false,
|
| 88 |
+
"rstrip": false,
|
| 89 |
+
"single_word": false,
|
| 90 |
+
"special": true
|
| 91 |
+
},
|
| 92 |
+
"11": {
|
| 93 |
+
"content": "<jupyter_start>",
|
| 94 |
+
"lstrip": false,
|
| 95 |
+
"normalized": false,
|
| 96 |
+
"rstrip": false,
|
| 97 |
+
"single_word": false,
|
| 98 |
+
"special": true
|
| 99 |
+
},
|
| 100 |
+
"12": {
|
| 101 |
+
"content": "<jupyter_text>",
|
| 102 |
+
"lstrip": false,
|
| 103 |
+
"normalized": false,
|
| 104 |
+
"rstrip": false,
|
| 105 |
+
"single_word": false,
|
| 106 |
+
"special": true
|
| 107 |
+
},
|
| 108 |
+
"13": {
|
| 109 |
+
"content": "<jupyter_code>",
|
| 110 |
+
"lstrip": false,
|
| 111 |
+
"normalized": false,
|
| 112 |
+
"rstrip": false,
|
| 113 |
+
"single_word": false,
|
| 114 |
+
"special": true
|
| 115 |
+
},
|
| 116 |
+
"14": {
|
| 117 |
+
"content": "<jupyter_output>",
|
| 118 |
+
"lstrip": false,
|
| 119 |
+
"normalized": false,
|
| 120 |
+
"rstrip": false,
|
| 121 |
+
"single_word": false,
|
| 122 |
+
"special": true
|
| 123 |
+
},
|
| 124 |
+
"15": {
|
| 125 |
+
"content": "<jupyter_script>",
|
| 126 |
+
"lstrip": false,
|
| 127 |
+
"normalized": false,
|
| 128 |
+
"rstrip": false,
|
| 129 |
+
"single_word": false,
|
| 130 |
+
"special": true
|
| 131 |
+
},
|
| 132 |
+
"16": {
|
| 133 |
+
"content": "<empty_output>",
|
| 134 |
+
"lstrip": false,
|
| 135 |
+
"normalized": false,
|
| 136 |
+
"rstrip": false,
|
| 137 |
+
"single_word": false,
|
| 138 |
+
"special": true
|
| 139 |
+
}
|
| 140 |
+
},
|
| 141 |
+
"additional_special_tokens": [
|
| 142 |
+
"<|im_start|>",
|
| 143 |
+
"<|im_end|>"
|
| 144 |
+
],
|
| 145 |
+
"bos_token": "<|im_start|>",
|
| 146 |
+
"clean_up_tokenization_spaces": false,
|
| 147 |
+
"eos_token": "<|im_end|>",
|
| 148 |
+
"extra_special_tokens": {},
|
| 149 |
+
"model_max_length": 8192,
|
| 150 |
+
"pad_token": "<|endoftext|>",
|
| 151 |
+
"padding_side": "right",
|
| 152 |
+
"tokenizer_class": "GPT2Tokenizer",
|
| 153 |
+
"unk_token": "<|endoftext|>",
|
| 154 |
+
"vocab_size": 49152
|
| 155 |
+
}
|
continuous/smollm2/mc1_b32_224_2.5b/finetune/trainer_state.json
ADDED
|
@@ -0,0 +1,1856 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 1.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 2597,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.003850596842510589,
|
| 14 |
+
"grad_norm": 3.0389933586120605,
|
| 15 |
+
"learning_rate": 1.153846153846154e-06,
|
| 16 |
+
"loss": 1.1961,
|
| 17 |
+
"step": 10
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.007701193685021178,
|
| 21 |
+
"grad_norm": 2.103576183319092,
|
| 22 |
+
"learning_rate": 2.435897435897436e-06,
|
| 23 |
+
"loss": 1.11,
|
| 24 |
+
"step": 20
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.011551790527531768,
|
| 28 |
+
"grad_norm": 2.082343339920044,
|
| 29 |
+
"learning_rate": 3.7179487179487184e-06,
|
| 30 |
+
"loss": 1.0261,
|
| 31 |
+
"step": 30
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.015402387370042356,
|
| 35 |
+
"grad_norm": 1.421505331993103,
|
| 36 |
+
"learning_rate": 5e-06,
|
| 37 |
+
"loss": 0.9856,
|
| 38 |
+
"step": 40
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.019252984212552945,
|
| 42 |
+
"grad_norm": 1.35994553565979,
|
| 43 |
+
"learning_rate": 6.282051282051282e-06,
|
| 44 |
+
"loss": 0.9691,
|
| 45 |
+
"step": 50
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.023103581055063535,
|
| 49 |
+
"grad_norm": 0.33109214901924133,
|
| 50 |
+
"learning_rate": 7.564102564102564e-06,
|
| 51 |
+
"loss": 0.8968,
|
| 52 |
+
"step": 60
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.026954177897574125,
|
| 56 |
+
"grad_norm": 0.26070114970207214,
|
| 57 |
+
"learning_rate": 8.846153846153847e-06,
|
| 58 |
+
"loss": 0.8484,
|
| 59 |
+
"step": 70
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.030804774740084712,
|
| 63 |
+
"grad_norm": 0.27182310819625854,
|
| 64 |
+
"learning_rate": 9.999996111488723e-06,
|
| 65 |
+
"loss": 0.8597,
|
| 66 |
+
"step": 80
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.0346553715825953,
|
| 70 |
+
"grad_norm": 0.257328063249588,
|
| 71 |
+
"learning_rate": 9.999529497453782e-06,
|
| 72 |
+
"loss": 0.8445,
|
| 73 |
+
"step": 90
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.03850596842510589,
|
| 77 |
+
"grad_norm": 0.2523109018802643,
|
| 78 |
+
"learning_rate": 9.998285264324193e-06,
|
| 79 |
+
"loss": 0.8128,
|
| 80 |
+
"step": 100
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.04235656526761648,
|
| 84 |
+
"grad_norm": 0.23408886790275574,
|
| 85 |
+
"learning_rate": 9.996263605626052e-06,
|
| 86 |
+
"loss": 0.8039,
|
| 87 |
+
"step": 110
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.04620716211012707,
|
| 91 |
+
"grad_norm": 0.23644255101680756,
|
| 92 |
+
"learning_rate": 9.99346483580503e-06,
|
| 93 |
+
"loss": 0.8231,
|
| 94 |
+
"step": 120
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.05005775895263766,
|
| 98 |
+
"grad_norm": 0.241837278008461,
|
| 99 |
+
"learning_rate": 9.989889390177467e-06,
|
| 100 |
+
"loss": 0.7996,
|
| 101 |
+
"step": 130
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.05390835579514825,
|
| 105 |
+
"grad_norm": 0.27184751629829407,
|
| 106 |
+
"learning_rate": 9.985537824862647e-06,
|
| 107 |
+
"loss": 0.8133,
|
| 108 |
+
"step": 140
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.05775895263765884,
|
| 112 |
+
"grad_norm": 0.272132009267807,
|
| 113 |
+
"learning_rate": 9.980410816696317e-06,
|
| 114 |
+
"loss": 0.7908,
|
| 115 |
+
"step": 150
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.061609549480169425,
|
| 119 |
+
"grad_norm": 0.21557871997356415,
|
| 120 |
+
"learning_rate": 9.97450916312541e-06,
|
| 121 |
+
"loss": 0.7973,
|
| 122 |
+
"step": 160
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.06546014632268002,
|
| 126 |
+
"grad_norm": 0.19836604595184326,
|
| 127 |
+
"learning_rate": 9.967833782084004e-06,
|
| 128 |
+
"loss": 0.7969,
|
| 129 |
+
"step": 170
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.0693107431651906,
|
| 133 |
+
"grad_norm": 0.23535935580730438,
|
| 134 |
+
"learning_rate": 9.960385711850554e-06,
|
| 135 |
+
"loss": 0.8298,
|
| 136 |
+
"step": 180
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.07316134000770119,
|
| 140 |
+
"grad_norm": 0.23828832805156708,
|
| 141 |
+
"learning_rate": 9.952166110886394e-06,
|
| 142 |
+
"loss": 0.8172,
|
| 143 |
+
"step": 190
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.07701193685021178,
|
| 147 |
+
"grad_norm": 0.21886341273784637,
|
| 148 |
+
"learning_rate": 9.943176257655567e-06,
|
| 149 |
+
"loss": 0.8155,
|
| 150 |
+
"step": 200
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.08086253369272237,
|
| 154 |
+
"grad_norm": 0.20388799905776978,
|
| 155 |
+
"learning_rate": 9.933417550425952e-06,
|
| 156 |
+
"loss": 0.8027,
|
| 157 |
+
"step": 210
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.08471313053523297,
|
| 161 |
+
"grad_norm": 0.2167951762676239,
|
| 162 |
+
"learning_rate": 9.9228915070518e-06,
|
| 163 |
+
"loss": 0.7931,
|
| 164 |
+
"step": 220
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.08856372737774355,
|
| 168 |
+
"grad_norm": 0.2007095366716385,
|
| 169 |
+
"learning_rate": 9.911599764737632e-06,
|
| 170 |
+
"loss": 0.7893,
|
| 171 |
+
"step": 230
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.09241432422025414,
|
| 175 |
+
"grad_norm": 0.19667713344097137,
|
| 176 |
+
"learning_rate": 9.899544079783606e-06,
|
| 177 |
+
"loss": 0.7823,
|
| 178 |
+
"step": 240
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.09626492106276473,
|
| 182 |
+
"grad_norm": 0.20376968383789062,
|
| 183 |
+
"learning_rate": 9.886726327312338e-06,
|
| 184 |
+
"loss": 0.8007,
|
| 185 |
+
"step": 250
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.10011551790527531,
|
| 189 |
+
"grad_norm": 0.23066265881061554,
|
| 190 |
+
"learning_rate": 9.87314850097724e-06,
|
| 191 |
+
"loss": 0.7986,
|
| 192 |
+
"step": 260
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.1039661147477859,
|
| 196 |
+
"grad_norm": 0.22822503745555878,
|
| 197 |
+
"learning_rate": 9.858812712652452e-06,
|
| 198 |
+
"loss": 0.7816,
|
| 199 |
+
"step": 270
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.1078167115902965,
|
| 203 |
+
"grad_norm": 0.20772984623908997,
|
| 204 |
+
"learning_rate": 9.843721192104341e-06,
|
| 205 |
+
"loss": 0.7787,
|
| 206 |
+
"step": 280
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.11166730843280709,
|
| 210 |
+
"grad_norm": 0.2161320298910141,
|
| 211 |
+
"learning_rate": 9.827876286644698e-06,
|
| 212 |
+
"loss": 0.8104,
|
| 213 |
+
"step": 290
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.11551790527531768,
|
| 217 |
+
"grad_norm": 0.1832609474658966,
|
| 218 |
+
"learning_rate": 9.811280460765643e-06,
|
| 219 |
+
"loss": 0.769,
|
| 220 |
+
"step": 300
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.11936850211782826,
|
| 224 |
+
"grad_norm": 0.19735439121723175,
|
| 225 |
+
"learning_rate": 9.793936295756292e-06,
|
| 226 |
+
"loss": 0.7745,
|
| 227 |
+
"step": 310
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.12321909896033885,
|
| 231 |
+
"grad_norm": 0.19830678403377533,
|
| 232 |
+
"learning_rate": 9.775846489301279e-06,
|
| 233 |
+
"loss": 0.783,
|
| 234 |
+
"step": 320
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.12706969580284944,
|
| 238 |
+
"grad_norm": 0.202273428440094,
|
| 239 |
+
"learning_rate": 9.757013855061157e-06,
|
| 240 |
+
"loss": 0.785,
|
| 241 |
+
"step": 330
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.13092029264536004,
|
| 245 |
+
"grad_norm": 0.1965688169002533,
|
| 246 |
+
"learning_rate": 9.737441322234758e-06,
|
| 247 |
+
"loss": 0.7871,
|
| 248 |
+
"step": 340
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.1347708894878706,
|
| 252 |
+
"grad_norm": 0.20624244213104248,
|
| 253 |
+
"learning_rate": 9.717131935103605e-06,
|
| 254 |
+
"loss": 0.7945,
|
| 255 |
+
"step": 350
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.1386214863303812,
|
| 259 |
+
"grad_norm": 0.19883181154727936,
|
| 260 |
+
"learning_rate": 9.696088852558398e-06,
|
| 261 |
+
"loss": 0.775,
|
| 262 |
+
"step": 360
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.1424720831728918,
|
| 266 |
+
"grad_norm": 0.22585013508796692,
|
| 267 |
+
"learning_rate": 9.674315347607686e-06,
|
| 268 |
+
"loss": 0.8026,
|
| 269 |
+
"step": 370
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.14632268001540238,
|
| 273 |
+
"grad_norm": 0.213645800948143,
|
| 274 |
+
"learning_rate": 9.651814806868796e-06,
|
| 275 |
+
"loss": 0.7673,
|
| 276 |
+
"step": 380
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.15017327685791299,
|
| 280 |
+
"grad_norm": 0.20081381499767303,
|
| 281 |
+
"learning_rate": 9.62859073004107e-06,
|
| 282 |
+
"loss": 0.7702,
|
| 283 |
+
"step": 390
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.15402387370042356,
|
| 287 |
+
"grad_norm": 0.1744217872619629,
|
| 288 |
+
"learning_rate": 9.604646729361547e-06,
|
| 289 |
+
"loss": 0.7781,
|
| 290 |
+
"step": 400
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.15787447054293416,
|
| 294 |
+
"grad_norm": 0.2149629145860672,
|
| 295 |
+
"learning_rate": 9.579986529043099e-06,
|
| 296 |
+
"loss": 0.7747,
|
| 297 |
+
"step": 410
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.16172506738544473,
|
| 301 |
+
"grad_norm": 0.22572393715381622,
|
| 302 |
+
"learning_rate": 9.554613964695189e-06,
|
| 303 |
+
"loss": 0.7893,
|
| 304 |
+
"step": 420
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 0.16557566422795533,
|
| 308 |
+
"grad_norm": 0.1955273598432541,
|
| 309 |
+
"learning_rate": 9.528532982727281e-06,
|
| 310 |
+
"loss": 0.7708,
|
| 311 |
+
"step": 430
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.16942626107046593,
|
| 315 |
+
"grad_norm": 0.18861645460128784,
|
| 316 |
+
"learning_rate": 9.501747639735018e-06,
|
| 317 |
+
"loss": 0.7784,
|
| 318 |
+
"step": 440
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.1732768579129765,
|
| 322 |
+
"grad_norm": 0.2140524536371231,
|
| 323 |
+
"learning_rate": 9.474262101869272e-06,
|
| 324 |
+
"loss": 0.763,
|
| 325 |
+
"step": 450
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.1771274547554871,
|
| 329 |
+
"grad_norm": 0.20963913202285767,
|
| 330 |
+
"learning_rate": 9.446080644188141e-06,
|
| 331 |
+
"loss": 0.7726,
|
| 332 |
+
"step": 460
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.18097805159799768,
|
| 336 |
+
"grad_norm": 0.19561170041561127,
|
| 337 |
+
"learning_rate": 9.417207649992017e-06,
|
| 338 |
+
"loss": 0.7744,
|
| 339 |
+
"step": 470
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.18482864844050828,
|
| 343 |
+
"grad_norm": 0.21152062714099884,
|
| 344 |
+
"learning_rate": 9.387647610141816e-06,
|
| 345 |
+
"loss": 0.7744,
|
| 346 |
+
"step": 480
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.18867924528301888,
|
| 350 |
+
"grad_norm": 0.18556399643421173,
|
| 351 |
+
"learning_rate": 9.35740512236047e-06,
|
| 352 |
+
"loss": 0.7675,
|
| 353 |
+
"step": 490
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.19252984212552945,
|
| 357 |
+
"grad_norm": 0.1855229139328003,
|
| 358 |
+
"learning_rate": 9.326484890517809e-06,
|
| 359 |
+
"loss": 0.7707,
|
| 360 |
+
"step": 500
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.19638043896804006,
|
| 364 |
+
"grad_norm": 0.1991398185491562,
|
| 365 |
+
"learning_rate": 9.294891723898926e-06,
|
| 366 |
+
"loss": 0.7608,
|
| 367 |
+
"step": 510
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.20023103581055063,
|
| 371 |
+
"grad_norm": 0.19114550948143005,
|
| 372 |
+
"learning_rate": 9.262630536456153e-06,
|
| 373 |
+
"loss": 0.7857,
|
| 374 |
+
"step": 520
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.20408163265306123,
|
| 378 |
+
"grad_norm": 0.19820517301559448,
|
| 379 |
+
"learning_rate": 9.229706346044749e-06,
|
| 380 |
+
"loss": 0.7677,
|
| 381 |
+
"step": 530
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.2079322294955718,
|
| 385 |
+
"grad_norm": 0.19739079475402832,
|
| 386 |
+
"learning_rate": 9.196124273642422e-06,
|
| 387 |
+
"loss": 0.7779,
|
| 388 |
+
"step": 540
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.2117828263380824,
|
| 392 |
+
"grad_norm": 0.18581658601760864,
|
| 393 |
+
"learning_rate": 9.161889542552842e-06,
|
| 394 |
+
"loss": 0.783,
|
| 395 |
+
"step": 550
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.215633423180593,
|
| 399 |
+
"grad_norm": 0.2048119604587555,
|
| 400 |
+
"learning_rate": 9.127007477593193e-06,
|
| 401 |
+
"loss": 0.7852,
|
| 402 |
+
"step": 560
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.21948402002310358,
|
| 406 |
+
"grad_norm": 0.2061079740524292,
|
| 407 |
+
"learning_rate": 9.091483504265974e-06,
|
| 408 |
+
"loss": 0.8022,
|
| 409 |
+
"step": 570
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.22333461686561418,
|
| 413 |
+
"grad_norm": 0.2301781177520752,
|
| 414 |
+
"learning_rate": 9.055323147915116e-06,
|
| 415 |
+
"loss": 0.7679,
|
| 416 |
+
"step": 580
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.22718521370812475,
|
| 420 |
+
"grad_norm": 0.18711897730827332,
|
| 421 |
+
"learning_rate": 9.018532032866585e-06,
|
| 422 |
+
"loss": 0.7843,
|
| 423 |
+
"step": 590
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.23103581055063535,
|
| 427 |
+
"grad_norm": 0.17545408010482788,
|
| 428 |
+
"learning_rate": 8.981115881553579e-06,
|
| 429 |
+
"loss": 0.7813,
|
| 430 |
+
"step": 600
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 0.23488640739314592,
|
| 434 |
+
"grad_norm": 0.21313416957855225,
|
| 435 |
+
"learning_rate": 8.943080513626476e-06,
|
| 436 |
+
"loss": 0.7994,
|
| 437 |
+
"step": 610
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 0.23873700423565652,
|
| 441 |
+
"grad_norm": 0.18089522421360016,
|
| 442 |
+
"learning_rate": 8.904431845047641e-06,
|
| 443 |
+
"loss": 0.7771,
|
| 444 |
+
"step": 620
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.24258760107816713,
|
| 448 |
+
"grad_norm": 0.21252045035362244,
|
| 449 |
+
"learning_rate": 8.86517588717128e-06,
|
| 450 |
+
"loss": 0.7895,
|
| 451 |
+
"step": 630
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 0.2464381979206777,
|
| 455 |
+
"grad_norm": 0.19222480058670044,
|
| 456 |
+
"learning_rate": 8.82531874580844e-06,
|
| 457 |
+
"loss": 0.7858,
|
| 458 |
+
"step": 640
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.25028879476318827,
|
| 462 |
+
"grad_norm": 0.17959687113761902,
|
| 463 |
+
"learning_rate": 8.784866620277314e-06,
|
| 464 |
+
"loss": 0.7479,
|
| 465 |
+
"step": 650
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.25413939160569887,
|
| 469 |
+
"grad_norm": 0.19166706502437592,
|
| 470 |
+
"learning_rate": 8.743825802439018e-06,
|
| 471 |
+
"loss": 0.7744,
|
| 472 |
+
"step": 660
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 0.2579899884482095,
|
| 476 |
+
"grad_norm": 0.1829976588487625,
|
| 477 |
+
"learning_rate": 8.702202675718956e-06,
|
| 478 |
+
"loss": 0.7816,
|
| 479 |
+
"step": 670
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.2618405852907201,
|
| 483 |
+
"grad_norm": 0.205853670835495,
|
| 484 |
+
"learning_rate": 8.660003714113959e-06,
|
| 485 |
+
"loss": 0.7645,
|
| 486 |
+
"step": 680
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 0.2656911821332307,
|
| 490 |
+
"grad_norm": 0.19417399168014526,
|
| 491 |
+
"learning_rate": 8.617235481185323e-06,
|
| 492 |
+
"loss": 0.7742,
|
| 493 |
+
"step": 690
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 0.2695417789757412,
|
| 497 |
+
"grad_norm": 0.22963868081569672,
|
| 498 |
+
"learning_rate": 8.573904629037931e-06,
|
| 499 |
+
"loss": 0.7874,
|
| 500 |
+
"step": 700
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 0.2733923758182518,
|
| 504 |
+
"grad_norm": 0.1756202131509781,
|
| 505 |
+
"learning_rate": 8.530017897285585e-06,
|
| 506 |
+
"loss": 0.7677,
|
| 507 |
+
"step": 710
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.2772429726607624,
|
| 511 |
+
"grad_norm": 0.18162409961223602,
|
| 512 |
+
"learning_rate": 8.485582112002746e-06,
|
| 513 |
+
"loss": 0.7685,
|
| 514 |
+
"step": 720
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 0.281093569503273,
|
| 518 |
+
"grad_norm": 0.20109842717647552,
|
| 519 |
+
"learning_rate": 8.440604184662811e-06,
|
| 520 |
+
"loss": 0.7795,
|
| 521 |
+
"step": 730
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 0.2849441663457836,
|
| 525 |
+
"grad_norm": 0.18087700009346008,
|
| 526 |
+
"learning_rate": 8.395091111063118e-06,
|
| 527 |
+
"loss": 0.7642,
|
| 528 |
+
"step": 740
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 0.28879476318829417,
|
| 532 |
+
"grad_norm": 0.16643284261226654,
|
| 533 |
+
"learning_rate": 8.349049970236822e-06,
|
| 534 |
+
"loss": 0.7606,
|
| 535 |
+
"step": 750
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 0.29264536003080477,
|
| 539 |
+
"grad_norm": 0.18412572145462036,
|
| 540 |
+
"learning_rate": 8.302487923351842e-06,
|
| 541 |
+
"loss": 0.7583,
|
| 542 |
+
"step": 760
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 0.29649595687331537,
|
| 546 |
+
"grad_norm": 0.17171673476696014,
|
| 547 |
+
"learning_rate": 8.255412212597021e-06,
|
| 548 |
+
"loss": 0.7764,
|
| 549 |
+
"step": 770
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.30034655371582597,
|
| 553 |
+
"grad_norm": 0.20469729602336884,
|
| 554 |
+
"learning_rate": 8.207830160055688e-06,
|
| 555 |
+
"loss": 0.7716,
|
| 556 |
+
"step": 780
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 0.3041971505583365,
|
| 560 |
+
"grad_norm": 0.18164847791194916,
|
| 561 |
+
"learning_rate": 8.159749166566786e-06,
|
| 562 |
+
"loss": 0.7558,
|
| 563 |
+
"step": 790
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 0.3080477474008471,
|
| 567 |
+
"grad_norm": 0.17886154353618622,
|
| 568 |
+
"learning_rate": 8.11117671057377e-06,
|
| 569 |
+
"loss": 0.7776,
|
| 570 |
+
"step": 800
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 0.3118983442433577,
|
| 574 |
+
"grad_norm": 0.17253777384757996,
|
| 575 |
+
"learning_rate": 8.062120346961404e-06,
|
| 576 |
+
"loss": 0.7757,
|
| 577 |
+
"step": 810
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 0.3157489410858683,
|
| 581 |
+
"grad_norm": 0.17683227360248566,
|
| 582 |
+
"learning_rate": 8.012587705880705e-06,
|
| 583 |
+
"loss": 0.7733,
|
| 584 |
+
"step": 820
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 0.3195995379283789,
|
| 588 |
+
"grad_norm": 0.1790492683649063,
|
| 589 |
+
"learning_rate": 7.962586491562146e-06,
|
| 590 |
+
"loss": 0.7594,
|
| 591 |
+
"step": 830
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 0.32345013477088946,
|
| 595 |
+
"grad_norm": 0.18137134611606598,
|
| 596 |
+
"learning_rate": 7.912124481117357e-06,
|
| 597 |
+
"loss": 0.7594,
|
| 598 |
+
"step": 840
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 0.32730073161340006,
|
| 602 |
+
"grad_norm": 0.17885252833366394,
|
| 603 |
+
"learning_rate": 7.861209523329494e-06,
|
| 604 |
+
"loss": 0.7691,
|
| 605 |
+
"step": 850
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 0.33115132845591067,
|
| 609 |
+
"grad_norm": 0.19237762689590454,
|
| 610 |
+
"learning_rate": 7.809849537432432e-06,
|
| 611 |
+
"loss": 0.7613,
|
| 612 |
+
"step": 860
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 0.33500192529842127,
|
| 616 |
+
"grad_norm": 0.17438630759716034,
|
| 617 |
+
"learning_rate": 7.758052511879035e-06,
|
| 618 |
+
"loss": 0.7642,
|
| 619 |
+
"step": 870
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 0.33885252214093187,
|
| 623 |
+
"grad_norm": 0.21920138597488403,
|
| 624 |
+
"learning_rate": 7.705826503098642e-06,
|
| 625 |
+
"loss": 0.7676,
|
| 626 |
+
"step": 880
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 0.3427031189834424,
|
| 630 |
+
"grad_norm": 0.1898784637451172,
|
| 631 |
+
"learning_rate": 7.653179634243979e-06,
|
| 632 |
+
"loss": 0.7604,
|
| 633 |
+
"step": 890
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 0.346553715825953,
|
| 637 |
+
"grad_norm": 0.16759398579597473,
|
| 638 |
+
"learning_rate": 7.600120093927696e-06,
|
| 639 |
+
"loss": 0.7642,
|
| 640 |
+
"step": 900
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 0.3504043126684636,
|
| 644 |
+
"grad_norm": 0.1960635781288147,
|
| 645 |
+
"learning_rate": 7.546656134948729e-06,
|
| 646 |
+
"loss": 0.7543,
|
| 647 |
+
"step": 910
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 0.3542549095109742,
|
| 651 |
+
"grad_norm": 0.18080741167068481,
|
| 652 |
+
"learning_rate": 7.492796073008661e-06,
|
| 653 |
+
"loss": 0.7602,
|
| 654 |
+
"step": 920
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.3581055063534848,
|
| 658 |
+
"grad_norm": 0.23220692574977875,
|
| 659 |
+
"learning_rate": 7.438548285418326e-06,
|
| 660 |
+
"loss": 0.7422,
|
| 661 |
+
"step": 930
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.36195610319599536,
|
| 665 |
+
"grad_norm": 0.19567300379276276,
|
| 666 |
+
"learning_rate": 7.3839212097948e-06,
|
| 667 |
+
"loss": 0.7736,
|
| 668 |
+
"step": 940
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 0.36580670003850596,
|
| 672 |
+
"grad_norm": 0.24979831278324127,
|
| 673 |
+
"learning_rate": 7.328923342749039e-06,
|
| 674 |
+
"loss": 0.7904,
|
| 675 |
+
"step": 950
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.36965729688101656,
|
| 679 |
+
"grad_norm": 0.17791008949279785,
|
| 680 |
+
"learning_rate": 7.273563238564319e-06,
|
| 681 |
+
"loss": 0.7853,
|
| 682 |
+
"step": 960
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 0.37350789372352716,
|
| 686 |
+
"grad_norm": 0.2068287879228592,
|
| 687 |
+
"learning_rate": 7.217849507865724e-06,
|
| 688 |
+
"loss": 0.7554,
|
| 689 |
+
"step": 970
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 0.37735849056603776,
|
| 693 |
+
"grad_norm": 0.18160636723041534,
|
| 694 |
+
"learning_rate": 7.161790816280858e-06,
|
| 695 |
+
"loss": 0.7631,
|
| 696 |
+
"step": 980
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 0.3812090874085483,
|
| 700 |
+
"grad_norm": 0.18440678715705872,
|
| 701 |
+
"learning_rate": 7.1053958830920125e-06,
|
| 702 |
+
"loss": 0.7519,
|
| 703 |
+
"step": 990
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.3850596842510589,
|
| 707 |
+
"grad_norm": 0.18616710603237152,
|
| 708 |
+
"learning_rate": 7.0486734798799715e-06,
|
| 709 |
+
"loss": 0.7624,
|
| 710 |
+
"step": 1000
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 0.3889102810935695,
|
| 714 |
+
"grad_norm": 0.17475666105747223,
|
| 715 |
+
"learning_rate": 6.991632429159708e-06,
|
| 716 |
+
"loss": 0.7822,
|
| 717 |
+
"step": 1010
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 0.3927608779360801,
|
| 721 |
+
"grad_norm": 0.19472917914390564,
|
| 722 |
+
"learning_rate": 6.934281603008132e-06,
|
| 723 |
+
"loss": 0.7808,
|
| 724 |
+
"step": 1020
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"epoch": 0.39661147477859066,
|
| 728 |
+
"grad_norm": 0.16190624237060547,
|
| 729 |
+
"learning_rate": 6.876629921684143e-06,
|
| 730 |
+
"loss": 0.7771,
|
| 731 |
+
"step": 1030
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 0.40046207162110126,
|
| 735 |
+
"grad_norm": 0.1998361051082611,
|
| 736 |
+
"learning_rate": 6.818686352241199e-06,
|
| 737 |
+
"loss": 0.7644,
|
| 738 |
+
"step": 1040
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 0.40431266846361186,
|
| 742 |
+
"grad_norm": 0.16509808599948883,
|
| 743 |
+
"learning_rate": 6.760459907132573e-06,
|
| 744 |
+
"loss": 0.761,
|
| 745 |
+
"step": 1050
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 0.40816326530612246,
|
| 749 |
+
"grad_norm": 0.17539195716381073,
|
| 750 |
+
"learning_rate": 6.701959642809589e-06,
|
| 751 |
+
"loss": 0.7666,
|
| 752 |
+
"step": 1060
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 0.41201386214863306,
|
| 756 |
+
"grad_norm": 0.1725974977016449,
|
| 757 |
+
"learning_rate": 6.643194658312987e-06,
|
| 758 |
+
"loss": 0.7586,
|
| 759 |
+
"step": 1070
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.4158644589911436,
|
| 763 |
+
"grad_norm": 0.16963861882686615,
|
| 764 |
+
"learning_rate": 6.584174093857676e-06,
|
| 765 |
+
"loss": 0.7835,
|
| 766 |
+
"step": 1080
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 0.4197150558336542,
|
| 770 |
+
"grad_norm": 0.1849723905324936,
|
| 771 |
+
"learning_rate": 6.5249071294110735e-06,
|
| 772 |
+
"loss": 0.766,
|
| 773 |
+
"step": 1090
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 0.4235656526761648,
|
| 777 |
+
"grad_norm": 0.179683119058609,
|
| 778 |
+
"learning_rate": 6.465402983265274e-06,
|
| 779 |
+
"loss": 0.778,
|
| 780 |
+
"step": 1100
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 0.4274162495186754,
|
| 784 |
+
"grad_norm": 0.20142687857151031,
|
| 785 |
+
"learning_rate": 6.405670910603243e-06,
|
| 786 |
+
"loss": 0.7715,
|
| 787 |
+
"step": 1110
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 0.431266846361186,
|
| 791 |
+
"grad_norm": 0.20258887112140656,
|
| 792 |
+
"learning_rate": 6.345720202059291e-06,
|
| 793 |
+
"loss": 0.7686,
|
| 794 |
+
"step": 1120
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 0.43511744320369655,
|
| 798 |
+
"grad_norm": 0.2059461921453476,
|
| 799 |
+
"learning_rate": 6.285560182274008e-06,
|
| 800 |
+
"loss": 0.7487,
|
| 801 |
+
"step": 1130
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 0.43896804004620715,
|
| 805 |
+
"grad_norm": 0.1899777203798294,
|
| 806 |
+
"learning_rate": 6.225200208443936e-06,
|
| 807 |
+
"loss": 0.7732,
|
| 808 |
+
"step": 1140
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 0.44281863688871775,
|
| 812 |
+
"grad_norm": 0.1828649789094925,
|
| 813 |
+
"learning_rate": 6.16464966886616e-06,
|
| 814 |
+
"loss": 0.7587,
|
| 815 |
+
"step": 1150
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 0.44666923373122835,
|
| 819 |
+
"grad_norm": 0.21838313341140747,
|
| 820 |
+
"learning_rate": 6.103917981478065e-06,
|
| 821 |
+
"loss": 0.7875,
|
| 822 |
+
"step": 1160
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 0.45051983057373896,
|
| 826 |
+
"grad_norm": 0.19157014787197113,
|
| 827 |
+
"learning_rate": 6.043014592392489e-06,
|
| 828 |
+
"loss": 0.7944,
|
| 829 |
+
"step": 1170
|
| 830 |
+
},
|
| 831 |
+
{
|
| 832 |
+
"epoch": 0.4543704274162495,
|
| 833 |
+
"grad_norm": 0.1762637048959732,
|
| 834 |
+
"learning_rate": 5.981948974428485e-06,
|
| 835 |
+
"loss": 0.775,
|
| 836 |
+
"step": 1180
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 0.4582210242587601,
|
| 840 |
+
"grad_norm": 0.17550347745418549,
|
| 841 |
+
"learning_rate": 5.920730625637934e-06,
|
| 842 |
+
"loss": 0.751,
|
| 843 |
+
"step": 1190
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 0.4620716211012707,
|
| 847 |
+
"grad_norm": 0.1862068474292755,
|
| 848 |
+
"learning_rate": 5.8593690678282355e-06,
|
| 849 |
+
"loss": 0.7596,
|
| 850 |
+
"step": 1200
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 0.4659222179437813,
|
| 854 |
+
"grad_norm": 0.1813003271818161,
|
| 855 |
+
"learning_rate": 5.797873845081301e-06,
|
| 856 |
+
"loss": 0.7574,
|
| 857 |
+
"step": 1210
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 0.46977281478629185,
|
| 861 |
+
"grad_norm": 0.1999894082546234,
|
| 862 |
+
"learning_rate": 5.736254522269075e-06,
|
| 863 |
+
"loss": 0.7833,
|
| 864 |
+
"step": 1220
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 0.47362341162880245,
|
| 868 |
+
"grad_norm": 0.17487768828868866,
|
| 869 |
+
"learning_rate": 5.6745206835658386e-06,
|
| 870 |
+
"loss": 0.7784,
|
| 871 |
+
"step": 1230
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 0.47747400847131305,
|
| 875 |
+
"grad_norm": 0.17683789134025574,
|
| 876 |
+
"learning_rate": 5.612681930957495e-06,
|
| 877 |
+
"loss": 0.7574,
|
| 878 |
+
"step": 1240
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 0.48132460531382365,
|
| 882 |
+
"grad_norm": 0.17646785080432892,
|
| 883 |
+
"learning_rate": 5.550747882748095e-06,
|
| 884 |
+
"loss": 0.7506,
|
| 885 |
+
"step": 1250
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 0.48517520215633425,
|
| 889 |
+
"grad_norm": 0.1946573406457901,
|
| 890 |
+
"learning_rate": 5.4887281720638195e-06,
|
| 891 |
+
"loss": 0.7628,
|
| 892 |
+
"step": 1260
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 0.4890257989988448,
|
| 896 |
+
"grad_norm": 0.19261430203914642,
|
| 897 |
+
"learning_rate": 5.42663244535466e-06,
|
| 898 |
+
"loss": 0.7411,
|
| 899 |
+
"step": 1270
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 0.4928763958413554,
|
| 903 |
+
"grad_norm": 0.1834699660539627,
|
| 904 |
+
"learning_rate": 5.3644703608940205e-06,
|
| 905 |
+
"loss": 0.7564,
|
| 906 |
+
"step": 1280
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 0.496726992683866,
|
| 910 |
+
"grad_norm": 0.1755651980638504,
|
| 911 |
+
"learning_rate": 5.302251587276491e-06,
|
| 912 |
+
"loss": 0.7691,
|
| 913 |
+
"step": 1290
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 0.5005775895263765,
|
| 917 |
+
"grad_norm": 0.19190266728401184,
|
| 918 |
+
"learning_rate": 5.2399858019140005e-06,
|
| 919 |
+
"loss": 0.7601,
|
| 920 |
+
"step": 1300
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 0.5044281863688872,
|
| 924 |
+
"grad_norm": 0.19020673632621765,
|
| 925 |
+
"learning_rate": 5.177682689530615e-06,
|
| 926 |
+
"loss": 0.7779,
|
| 927 |
+
"step": 1310
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 0.5082787832113977,
|
| 931 |
+
"grad_norm": 0.1990414708852768,
|
| 932 |
+
"learning_rate": 5.11535194065619e-06,
|
| 933 |
+
"loss": 0.7498,
|
| 934 |
+
"step": 1320
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 0.5121293800539084,
|
| 938 |
+
"grad_norm": 0.18387247622013092,
|
| 939 |
+
"learning_rate": 5.053003250119113e-06,
|
| 940 |
+
"loss": 0.7579,
|
| 941 |
+
"step": 1330
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 0.515979976896419,
|
| 945 |
+
"grad_norm": 0.18690666556358337,
|
| 946 |
+
"learning_rate": 4.9906463155383925e-06,
|
| 947 |
+
"loss": 0.7637,
|
| 948 |
+
"step": 1340
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 0.5198305737389295,
|
| 952 |
+
"grad_norm": 0.18904241919517517,
|
| 953 |
+
"learning_rate": 4.9282908358153046e-06,
|
| 954 |
+
"loss": 0.7608,
|
| 955 |
+
"step": 1350
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 0.5236811705814401,
|
| 959 |
+
"grad_norm": 0.20967189967632294,
|
| 960 |
+
"learning_rate": 4.8659465096248345e-06,
|
| 961 |
+
"loss": 0.7775,
|
| 962 |
+
"step": 1360
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 0.5275317674239507,
|
| 966 |
+
"grad_norm": 0.19664627313613892,
|
| 967 |
+
"learning_rate": 4.803623033907169e-06,
|
| 968 |
+
"loss": 0.7691,
|
| 969 |
+
"step": 1370
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 0.5313823642664613,
|
| 973 |
+
"grad_norm": 0.1874246597290039,
|
| 974 |
+
"learning_rate": 4.741330102359441e-06,
|
| 975 |
+
"loss": 0.7826,
|
| 976 |
+
"step": 1380
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 0.5352329611089719,
|
| 980 |
+
"grad_norm": 0.18424846231937408,
|
| 981 |
+
"learning_rate": 4.679077403927996e-06,
|
| 982 |
+
"loss": 0.7587,
|
| 983 |
+
"step": 1390
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 0.5390835579514824,
|
| 987 |
+
"grad_norm": 0.1835668981075287,
|
| 988 |
+
"learning_rate": 4.616874621301373e-06,
|
| 989 |
+
"loss": 0.7598,
|
| 990 |
+
"step": 1400
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 0.5429341547939931,
|
| 994 |
+
"grad_norm": 0.19265849888324738,
|
| 995 |
+
"learning_rate": 4.554731429404293e-06,
|
| 996 |
+
"loss": 0.777,
|
| 997 |
+
"step": 1410
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 0.5467847516365036,
|
| 1001 |
+
"grad_norm": 0.19345396757125854,
|
| 1002 |
+
"learning_rate": 4.492657493892822e-06,
|
| 1003 |
+
"loss": 0.7608,
|
| 1004 |
+
"step": 1420
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 0.5506353484790143,
|
| 1008 |
+
"grad_norm": 0.19195866584777832,
|
| 1009 |
+
"learning_rate": 4.430662469650999e-06,
|
| 1010 |
+
"loss": 0.7734,
|
| 1011 |
+
"step": 1430
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 0.5544859453215248,
|
| 1015 |
+
"grad_norm": 0.19251754879951477,
|
| 1016 |
+
"learning_rate": 4.368755999289118e-06,
|
| 1017 |
+
"loss": 0.7738,
|
| 1018 |
+
"step": 1440
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 0.5583365421640354,
|
| 1022 |
+
"grad_norm": 0.18953853845596313,
|
| 1023 |
+
"learning_rate": 4.306947711643949e-06,
|
| 1024 |
+
"loss": 0.7582,
|
| 1025 |
+
"step": 1450
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 0.562187139006546,
|
| 1029 |
+
"grad_norm": 0.18174311518669128,
|
| 1030 |
+
"learning_rate": 4.245247220281068e-06,
|
| 1031 |
+
"loss": 0.7633,
|
| 1032 |
+
"step": 1460
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 0.5660377358490566,
|
| 1036 |
+
"grad_norm": 0.17605026066303253,
|
| 1037 |
+
"learning_rate": 4.183664121999587e-06,
|
| 1038 |
+
"loss": 0.7603,
|
| 1039 |
+
"step": 1470
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"epoch": 0.5698883326915672,
|
| 1043 |
+
"grad_norm": 0.1963917762041092,
|
| 1044 |
+
"learning_rate": 4.122207995339474e-06,
|
| 1045 |
+
"loss": 0.7525,
|
| 1046 |
+
"step": 1480
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 0.5737389295340778,
|
| 1050 |
+
"grad_norm": 0.18174195289611816,
|
| 1051 |
+
"learning_rate": 4.060888399091734e-06,
|
| 1052 |
+
"loss": 0.7618,
|
| 1053 |
+
"step": 1490
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 0.5775895263765883,
|
| 1057 |
+
"grad_norm": 0.1773805022239685,
|
| 1058 |
+
"learning_rate": 3.999714870811639e-06,
|
| 1059 |
+
"loss": 0.7735,
|
| 1060 |
+
"step": 1500
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 0.581440123219099,
|
| 1064 |
+
"grad_norm": 0.18712858855724335,
|
| 1065 |
+
"learning_rate": 3.938696925335275e-06,
|
| 1066 |
+
"loss": 0.7661,
|
| 1067 |
+
"step": 1510
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 0.5852907200616095,
|
| 1071 |
+
"grad_norm": 0.1916084736585617,
|
| 1072 |
+
"learning_rate": 3.87784405329962e-06,
|
| 1073 |
+
"loss": 0.7741,
|
| 1074 |
+
"step": 1520
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 0.5891413169041202,
|
| 1078 |
+
"grad_norm": 0.16869810223579407,
|
| 1079 |
+
"learning_rate": 3.817165719666382e-06,
|
| 1080 |
+
"loss": 0.7755,
|
| 1081 |
+
"step": 1530
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 0.5929919137466307,
|
| 1085 |
+
"grad_norm": 0.18597762286663055,
|
| 1086 |
+
"learning_rate": 3.756671362249841e-06,
|
| 1087 |
+
"loss": 0.7763,
|
| 1088 |
+
"step": 1540
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 0.5968425105891413,
|
| 1092 |
+
"grad_norm": 0.17557670176029205,
|
| 1093 |
+
"learning_rate": 3.6963703902488907e-06,
|
| 1094 |
+
"loss": 0.7634,
|
| 1095 |
+
"step": 1550
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 0.6006931074316519,
|
| 1099 |
+
"grad_norm": 0.17192763090133667,
|
| 1100 |
+
"learning_rate": 3.6362721827835655e-06,
|
| 1101 |
+
"loss": 0.7775,
|
| 1102 |
+
"step": 1560
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 0.6045437042741625,
|
| 1106 |
+
"grad_norm": 0.19951577484607697,
|
| 1107 |
+
"learning_rate": 3.576386087436212e-06,
|
| 1108 |
+
"loss": 0.7799,
|
| 1109 |
+
"step": 1570
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 0.608394301116673,
|
| 1113 |
+
"grad_norm": 0.1947396993637085,
|
| 1114 |
+
"learning_rate": 3.5167214187975913e-06,
|
| 1115 |
+
"loss": 0.7734,
|
| 1116 |
+
"step": 1580
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 0.6122448979591837,
|
| 1120 |
+
"grad_norm": 0.15874922275543213,
|
| 1121 |
+
"learning_rate": 3.457287457018086e-06,
|
| 1122 |
+
"loss": 0.769,
|
| 1123 |
+
"step": 1590
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 0.6160954948016942,
|
| 1127 |
+
"grad_norm": 0.17710931599140167,
|
| 1128 |
+
"learning_rate": 3.3980934463643003e-06,
|
| 1129 |
+
"loss": 0.7635,
|
| 1130 |
+
"step": 1600
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 0.6199460916442049,
|
| 1134 |
+
"grad_norm": 0.16742654144763947,
|
| 1135 |
+
"learning_rate": 3.3391485937812053e-06,
|
| 1136 |
+
"loss": 0.7423,
|
| 1137 |
+
"step": 1610
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 0.6237966884867154,
|
| 1141 |
+
"grad_norm": 0.19137488305568695,
|
| 1142 |
+
"learning_rate": 3.280462067460115e-06,
|
| 1143 |
+
"loss": 0.7437,
|
| 1144 |
+
"step": 1620
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 0.627647285329226,
|
| 1148 |
+
"grad_norm": 0.1792331486940384,
|
| 1149 |
+
"learning_rate": 3.222042995412669e-06,
|
| 1150 |
+
"loss": 0.7689,
|
| 1151 |
+
"step": 1630
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 0.6314978821717366,
|
| 1155 |
+
"grad_norm": 0.16917359828948975,
|
| 1156 |
+
"learning_rate": 3.163900464051089e-06,
|
| 1157 |
+
"loss": 0.7602,
|
| 1158 |
+
"step": 1640
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"epoch": 0.6353484790142472,
|
| 1162 |
+
"grad_norm": 0.1872333437204361,
|
| 1163 |
+
"learning_rate": 3.1060435167748826e-06,
|
| 1164 |
+
"loss": 0.7749,
|
| 1165 |
+
"step": 1650
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 0.6391990758567578,
|
| 1169 |
+
"grad_norm": 0.1776774823665619,
|
| 1170 |
+
"learning_rate": 3.048481152564253e-06,
|
| 1171 |
+
"loss": 0.7511,
|
| 1172 |
+
"step": 1660
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 0.6430496726992684,
|
| 1176 |
+
"grad_norm": 0.19339701533317566,
|
| 1177 |
+
"learning_rate": 2.9912223245804017e-06,
|
| 1178 |
+
"loss": 0.7698,
|
| 1179 |
+
"step": 1670
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 0.6469002695417789,
|
| 1183 |
+
"grad_norm": 0.17855679988861084,
|
| 1184 |
+
"learning_rate": 2.934275938772981e-06,
|
| 1185 |
+
"loss": 0.7624,
|
| 1186 |
+
"step": 1680
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 0.6507508663842896,
|
| 1190 |
+
"grad_norm": 0.18356171250343323,
|
| 1191 |
+
"learning_rate": 2.8776508524948676e-06,
|
| 1192 |
+
"loss": 0.7575,
|
| 1193 |
+
"step": 1690
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 0.6546014632268001,
|
| 1197 |
+
"grad_norm": 0.1728813648223877,
|
| 1198 |
+
"learning_rate": 2.821355873124503e-06,
|
| 1199 |
+
"loss": 0.7609,
|
| 1200 |
+
"step": 1700
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 0.6584520600693108,
|
| 1204 |
+
"grad_norm": 0.1868329644203186,
|
| 1205 |
+
"learning_rate": 2.7653997566960144e-06,
|
| 1206 |
+
"loss": 0.7641,
|
| 1207 |
+
"step": 1710
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 0.6623026569118213,
|
| 1211 |
+
"grad_norm": 0.20271888375282288,
|
| 1212 |
+
"learning_rate": 2.709791206537309e-06,
|
| 1213 |
+
"loss": 0.7573,
|
| 1214 |
+
"step": 1720
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 0.6661532537543319,
|
| 1218 |
+
"grad_norm": 0.1905214488506317,
|
| 1219 |
+
"learning_rate": 2.6545388719163747e-06,
|
| 1220 |
+
"loss": 0.7657,
|
| 1221 |
+
"step": 1730
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 0.6700038505968425,
|
| 1225 |
+
"grad_norm": 0.17898045480251312,
|
| 1226 |
+
"learning_rate": 2.599651346695979e-06,
|
| 1227 |
+
"loss": 0.7729,
|
| 1228 |
+
"step": 1740
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 0.6738544474393531,
|
| 1232 |
+
"grad_norm": 0.17109276354312897,
|
| 1233 |
+
"learning_rate": 2.545137167997007e-06,
|
| 1234 |
+
"loss": 0.7646,
|
| 1235 |
+
"step": 1750
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 0.6777050442818637,
|
| 1239 |
+
"grad_norm": 0.17974501848220825,
|
| 1240 |
+
"learning_rate": 2.4910048148705972e-06,
|
| 1241 |
+
"loss": 0.7598,
|
| 1242 |
+
"step": 1760
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 0.6815556411243743,
|
| 1246 |
+
"grad_norm": 0.19514746963977814,
|
| 1247 |
+
"learning_rate": 2.4372627069793327e-06,
|
| 1248 |
+
"loss": 0.7932,
|
| 1249 |
+
"step": 1770
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 0.6854062379668848,
|
| 1253 |
+
"grad_norm": 0.17416617274284363,
|
| 1254 |
+
"learning_rate": 2.3839192032876486e-06,
|
| 1255 |
+
"loss": 0.7703,
|
| 1256 |
+
"step": 1780
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 0.6892568348093955,
|
| 1260 |
+
"grad_norm": 0.18420708179473877,
|
| 1261 |
+
"learning_rate": 2.3309826007617047e-06,
|
| 1262 |
+
"loss": 0.7867,
|
| 1263 |
+
"step": 1790
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 0.693107431651906,
|
| 1267 |
+
"grad_norm": 0.17710979282855988,
|
| 1268 |
+
"learning_rate": 2.278461133078881e-06,
|
| 1269 |
+
"loss": 0.7546,
|
| 1270 |
+
"step": 1800
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 0.6969580284944167,
|
| 1274 |
+
"grad_norm": 0.18198853731155396,
|
| 1275 |
+
"learning_rate": 2.2263629693471258e-06,
|
| 1276 |
+
"loss": 0.7484,
|
| 1277 |
+
"step": 1810
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 0.7008086253369272,
|
| 1281 |
+
"grad_norm": 0.17320799827575684,
|
| 1282 |
+
"learning_rate": 2.1746962128343414e-06,
|
| 1283 |
+
"loss": 0.7723,
|
| 1284 |
+
"step": 1820
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 0.7046592221794378,
|
| 1288 |
+
"grad_norm": 0.18094617128372192,
|
| 1289 |
+
"learning_rate": 2.1234688997080183e-06,
|
| 1290 |
+
"loss": 0.7548,
|
| 1291 |
+
"step": 1830
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 0.7085098190219484,
|
| 1295 |
+
"grad_norm": 0.17804455757141113,
|
| 1296 |
+
"learning_rate": 2.0726889977853053e-06,
|
| 1297 |
+
"loss": 0.7596,
|
| 1298 |
+
"step": 1840
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 0.712360415864459,
|
| 1302 |
+
"grad_norm": 0.1994146853685379,
|
| 1303 |
+
"learning_rate": 2.022364405293703e-06,
|
| 1304 |
+
"loss": 0.775,
|
| 1305 |
+
"step": 1850
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 0.7162110127069696,
|
| 1309 |
+
"grad_norm": 0.18103033304214478,
|
| 1310 |
+
"learning_rate": 1.9725029496425814e-06,
|
| 1311 |
+
"loss": 0.7719,
|
| 1312 |
+
"step": 1860
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 0.7200616095494802,
|
| 1316 |
+
"grad_norm": 0.18564371764659882,
|
| 1317 |
+
"learning_rate": 1.923112386205727e-06,
|
| 1318 |
+
"loss": 0.7628,
|
| 1319 |
+
"step": 1870
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 0.7239122063919907,
|
| 1323 |
+
"grad_norm": 0.17494694888591766,
|
| 1324 |
+
"learning_rate": 1.874200397115079e-06,
|
| 1325 |
+
"loss": 0.7744,
|
| 1326 |
+
"step": 1880
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 0.7277628032345014,
|
| 1330 |
+
"grad_norm": 0.19359874725341797,
|
| 1331 |
+
"learning_rate": 1.825774590065857e-06,
|
| 1332 |
+
"loss": 0.7598,
|
| 1333 |
+
"step": 1890
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 0.7316134000770119,
|
| 1337 |
+
"grad_norm": 0.19449737668037415,
|
| 1338 |
+
"learning_rate": 1.7778424971332813e-06,
|
| 1339 |
+
"loss": 0.7707,
|
| 1340 |
+
"step": 1900
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 0.7354639969195226,
|
| 1344 |
+
"grad_norm": 0.1769935041666031,
|
| 1345 |
+
"learning_rate": 1.7304115736010407e-06,
|
| 1346 |
+
"loss": 0.7691,
|
| 1347 |
+
"step": 1910
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 0.7393145937620331,
|
| 1351 |
+
"grad_norm": 0.18241924047470093,
|
| 1352 |
+
"learning_rate": 1.6834891968017108e-06,
|
| 1353 |
+
"loss": 0.7578,
|
| 1354 |
+
"step": 1920
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 0.7431651906045437,
|
| 1358 |
+
"grad_norm": 0.18720492720603943,
|
| 1359 |
+
"learning_rate": 1.6370826649692884e-06,
|
| 1360 |
+
"loss": 0.7774,
|
| 1361 |
+
"step": 1930
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 0.7470157874470543,
|
| 1365 |
+
"grad_norm": 0.1795533299446106,
|
| 1366 |
+
"learning_rate": 1.5911991961040463e-06,
|
| 1367 |
+
"loss": 0.7739,
|
| 1368 |
+
"step": 1940
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 0.7508663842895649,
|
| 1372 |
+
"grad_norm": 0.18319252133369446,
|
| 1373 |
+
"learning_rate": 1.545845926849851e-06,
|
| 1374 |
+
"loss": 0.7558,
|
| 1375 |
+
"step": 1950
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 0.7547169811320755,
|
| 1379 |
+
"grad_norm": 0.19484229385852814,
|
| 1380 |
+
"learning_rate": 1.5010299113841397e-06,
|
| 1381 |
+
"loss": 0.7829,
|
| 1382 |
+
"step": 1960
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 0.7585675779745861,
|
| 1386 |
+
"grad_norm": 0.18326008319854736,
|
| 1387 |
+
"learning_rate": 1.4567581203207238e-06,
|
| 1388 |
+
"loss": 0.7731,
|
| 1389 |
+
"step": 1970
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 0.7624181748170966,
|
| 1393 |
+
"grad_norm": 0.19267138838768005,
|
| 1394 |
+
"learning_rate": 1.413037439625593e-06,
|
| 1395 |
+
"loss": 0.7711,
|
| 1396 |
+
"step": 1980
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 0.7662687716596073,
|
| 1400 |
+
"grad_norm": 0.19358649849891663,
|
| 1401 |
+
"learning_rate": 1.3698746695458847e-06,
|
| 1402 |
+
"loss": 0.7599,
|
| 1403 |
+
"step": 1990
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 0.7701193685021178,
|
| 1407 |
+
"grad_norm": 0.1800256371498108,
|
| 1408 |
+
"learning_rate": 1.3272765235521745e-06,
|
| 1409 |
+
"loss": 0.7616,
|
| 1410 |
+
"step": 2000
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 0.7739699653446284,
|
| 1414 |
+
"grad_norm": 0.1800033003091812,
|
| 1415 |
+
"learning_rate": 1.2852496272942877e-06,
|
| 1416 |
+
"loss": 0.7675,
|
| 1417 |
+
"step": 2010
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 0.777820562187139,
|
| 1421 |
+
"grad_norm": 0.17263908684253693,
|
| 1422 |
+
"learning_rate": 1.2438005175707396e-06,
|
| 1423 |
+
"loss": 0.7505,
|
| 1424 |
+
"step": 2020
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 0.7816711590296496,
|
| 1428 |
+
"grad_norm": 0.1786670982837677,
|
| 1429 |
+
"learning_rate": 1.2029356413120297e-06,
|
| 1430 |
+
"loss": 0.7449,
|
| 1431 |
+
"step": 2030
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 0.7855217558721602,
|
| 1435 |
+
"grad_norm": 0.18627265095710754,
|
| 1436 |
+
"learning_rate": 1.1626613545778786e-06,
|
| 1437 |
+
"loss": 0.76,
|
| 1438 |
+
"step": 2040
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 0.7893723527146708,
|
| 1442 |
+
"grad_norm": 0.17020221054553986,
|
| 1443 |
+
"learning_rate": 1.1229839215686294e-06,
|
| 1444 |
+
"loss": 0.7635,
|
| 1445 |
+
"step": 2050
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 0.7932229495571813,
|
| 1449 |
+
"grad_norm": 0.17699410021305084,
|
| 1450 |
+
"learning_rate": 1.0839095136509188e-06,
|
| 1451 |
+
"loss": 0.7484,
|
| 1452 |
+
"step": 2060
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 0.797073546399692,
|
| 1456 |
+
"grad_norm": 0.19386282563209534,
|
| 1457 |
+
"learning_rate": 1.045444208397791e-06,
|
| 1458 |
+
"loss": 0.7648,
|
| 1459 |
+
"step": 2070
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 0.8009241432422025,
|
| 1463 |
+
"grad_norm": 0.17186394333839417,
|
| 1464 |
+
"learning_rate": 1.007593988643402e-06,
|
| 1465 |
+
"loss": 0.7414,
|
| 1466 |
+
"step": 2080
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 0.8047747400847132,
|
| 1470 |
+
"grad_norm": 0.174429252743721,
|
| 1471 |
+
"learning_rate": 9.703647415524613e-07,
|
| 1472 |
+
"loss": 0.7676,
|
| 1473 |
+
"step": 2090
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 0.8086253369272237,
|
| 1477 |
+
"grad_norm": 0.17704562842845917,
|
| 1478 |
+
"learning_rate": 9.337622577045485e-07,
|
| 1479 |
+
"loss": 0.7611,
|
| 1480 |
+
"step": 2100
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"epoch": 0.8124759337697343,
|
| 1484 |
+
"grad_norm": 0.1772325187921524,
|
| 1485 |
+
"learning_rate": 8.977922301934588e-07,
|
| 1486 |
+
"loss": 0.7627,
|
| 1487 |
+
"step": 2110
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 0.8163265306122449,
|
| 1491 |
+
"grad_norm": 0.19124917685985565,
|
| 1492 |
+
"learning_rate": 8.624602537417009e-07,
|
| 1493 |
+
"loss": 0.7774,
|
| 1494 |
+
"step": 2120
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 0.8201771274547555,
|
| 1498 |
+
"grad_norm": 0.1882966309785843,
|
| 1499 |
+
"learning_rate": 8.277718238303123e-07,
|
| 1500 |
+
"loss": 0.7589,
|
| 1501 |
+
"step": 2130
|
| 1502 |
+
},
|
| 1503 |
+
{
|
| 1504 |
+
"epoch": 0.8240277242972661,
|
| 1505 |
+
"grad_norm": 0.1745186448097229,
|
| 1506 |
+
"learning_rate": 7.937323358440935e-07,
|
| 1507 |
+
"loss": 0.7611,
|
| 1508 |
+
"step": 2140
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 0.8278783211397767,
|
| 1512 |
+
"grad_norm": 0.1773451715707779,
|
| 1513 |
+
"learning_rate": 7.603470842324223e-07,
|
| 1514 |
+
"loss": 0.7508,
|
| 1515 |
+
"step": 2150
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"epoch": 0.8317289179822872,
|
| 1519 |
+
"grad_norm": 0.1728517711162567,
|
| 1520 |
+
"learning_rate": 7.276212616857575e-07,
|
| 1521 |
+
"loss": 0.7716,
|
| 1522 |
+
"step": 2160
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"epoch": 0.8355795148247979,
|
| 1526 |
+
"grad_norm": 0.17615918815135956,
|
| 1527 |
+
"learning_rate": 6.955599583279837e-07,
|
| 1528 |
+
"loss": 0.7654,
|
| 1529 |
+
"step": 2170
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 0.8394301116673084,
|
| 1533 |
+
"grad_norm": 0.17022207379341125,
|
| 1534 |
+
"learning_rate": 6.641681609246981e-07,
|
| 1535 |
+
"loss": 0.7604,
|
| 1536 |
+
"step": 2180
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 0.8432807085098191,
|
| 1540 |
+
"grad_norm": 0.18910086154937744,
|
| 1541 |
+
"learning_rate": 6.334507521075711e-07,
|
| 1542 |
+
"loss": 0.762,
|
| 1543 |
+
"step": 2190
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 0.8471313053523296,
|
| 1547 |
+
"grad_norm": 0.1876983940601349,
|
| 1548 |
+
"learning_rate": 6.034125096149179e-07,
|
| 1549 |
+
"loss": 0.7636,
|
| 1550 |
+
"step": 2200
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 0.8509819021948402,
|
| 1554 |
+
"grad_norm": 0.1867261379957199,
|
| 1555 |
+
"learning_rate": 5.740581055485722e-07,
|
| 1556 |
+
"loss": 0.7353,
|
| 1557 |
+
"step": 2210
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 0.8548324990373508,
|
| 1561 |
+
"grad_norm": 0.18015824258327484,
|
| 1562 |
+
"learning_rate": 5.453921056471961e-07,
|
| 1563 |
+
"loss": 0.7686,
|
| 1564 |
+
"step": 2220
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 0.8586830958798614,
|
| 1568 |
+
"grad_norm": 0.19256606698036194,
|
| 1569 |
+
"learning_rate": 5.174189685761288e-07,
|
| 1570 |
+
"loss": 0.7615,
|
| 1571 |
+
"step": 2230
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 0.862533692722372,
|
| 1575 |
+
"grad_norm": 0.1762619912624359,
|
| 1576 |
+
"learning_rate": 4.901430452338979e-07,
|
| 1577 |
+
"loss": 0.7727,
|
| 1578 |
+
"step": 2240
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 0.8663842895648826,
|
| 1582 |
+
"grad_norm": 0.18190613389015198,
|
| 1583 |
+
"learning_rate": 4.6356857807548286e-07,
|
| 1584 |
+
"loss": 0.7603,
|
| 1585 |
+
"step": 2250
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 0.8702348864073931,
|
| 1589 |
+
"grad_norm": 0.20101071894168854,
|
| 1590 |
+
"learning_rate": 4.376997004524541e-07,
|
| 1591 |
+
"loss": 0.743,
|
| 1592 |
+
"step": 2260
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 0.8740854832499038,
|
| 1596 |
+
"grad_norm": 0.17443805932998657,
|
| 1597 |
+
"learning_rate": 4.12540435970073e-07,
|
| 1598 |
+
"loss": 0.7618,
|
| 1599 |
+
"step": 2270
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 0.8779360800924143,
|
| 1603 |
+
"grad_norm": 0.1716483235359192,
|
| 1604 |
+
"learning_rate": 3.880946978614703e-07,
|
| 1605 |
+
"loss": 0.7663,
|
| 1606 |
+
"step": 2280
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"epoch": 0.881786676934925,
|
| 1610 |
+
"grad_norm": 0.18252786993980408,
|
| 1611 |
+
"learning_rate": 3.643662883789878e-07,
|
| 1612 |
+
"loss": 0.7719,
|
| 1613 |
+
"step": 2290
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 0.8856372737774355,
|
| 1617 |
+
"grad_norm": 0.16400329768657684,
|
| 1618 |
+
"learning_rate": 3.41358898202781e-07,
|
| 1619 |
+
"loss": 0.7517,
|
| 1620 |
+
"step": 2300
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 0.889487870619946,
|
| 1624 |
+
"grad_norm": 0.16708019375801086,
|
| 1625 |
+
"learning_rate": 3.190761058667741e-07,
|
| 1626 |
+
"loss": 0.7632,
|
| 1627 |
+
"step": 2310
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 0.8933384674624567,
|
| 1631 |
+
"grad_norm": 0.16746412217617035,
|
| 1632 |
+
"learning_rate": 2.9752137720206463e-07,
|
| 1633 |
+
"loss": 0.7655,
|
| 1634 |
+
"step": 2320
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 0.8971890643049673,
|
| 1638 |
+
"grad_norm": 0.1811438351869583,
|
| 1639 |
+
"learning_rate": 2.7669806479785067e-07,
|
| 1640 |
+
"loss": 0.7752,
|
| 1641 |
+
"step": 2330
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 0.9010396611474779,
|
| 1645 |
+
"grad_norm": 0.17560437321662903,
|
| 1646 |
+
"learning_rate": 2.566094074799763e-07,
|
| 1647 |
+
"loss": 0.7667,
|
| 1648 |
+
"step": 2340
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 0.9048902579899885,
|
| 1652 |
+
"grad_norm": 0.18160516023635864,
|
| 1653 |
+
"learning_rate": 2.3725852980716557e-07,
|
| 1654 |
+
"loss": 0.7463,
|
| 1655 |
+
"step": 2350
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 0.908740854832499,
|
| 1659 |
+
"grad_norm": 0.16930338740348816,
|
| 1660 |
+
"learning_rate": 2.1864844158503983e-07,
|
| 1661 |
+
"loss": 0.7628,
|
| 1662 |
+
"step": 2360
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 0.9125914516750097,
|
| 1666 |
+
"grad_norm": 0.19553492963314056,
|
| 1667 |
+
"learning_rate": 2.007820373979702e-07,
|
| 1668 |
+
"loss": 0.756,
|
| 1669 |
+
"step": 2370
|
| 1670 |
+
},
|
| 1671 |
+
{
|
| 1672 |
+
"epoch": 0.9164420485175202,
|
| 1673 |
+
"grad_norm": 0.1726168692111969,
|
| 1674 |
+
"learning_rate": 1.8366209615886532e-07,
|
| 1675 |
+
"loss": 0.7649,
|
| 1676 |
+
"step": 2380
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 0.9202926453600309,
|
| 1680 |
+
"grad_norm": 0.18009184300899506,
|
| 1681 |
+
"learning_rate": 1.6729128067693613e-07,
|
| 1682 |
+
"loss": 0.7776,
|
| 1683 |
+
"step": 2390
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 0.9241432422025414,
|
| 1687 |
+
"grad_norm": 0.19652995467185974,
|
| 1688 |
+
"learning_rate": 1.5167213724353426e-07,
|
| 1689 |
+
"loss": 0.771,
|
| 1690 |
+
"step": 2400
|
| 1691 |
+
},
|
| 1692 |
+
{
|
| 1693 |
+
"epoch": 0.927993839045052,
|
| 1694 |
+
"grad_norm": 0.1730128675699234,
|
| 1695 |
+
"learning_rate": 1.3680709523610058e-07,
|
| 1696 |
+
"loss": 0.7496,
|
| 1697 |
+
"step": 2410
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"epoch": 0.9318444358875626,
|
| 1701 |
+
"grad_norm": 0.16951508820056915,
|
| 1702 |
+
"learning_rate": 1.2269846674030418e-07,
|
| 1703 |
+
"loss": 0.7799,
|
| 1704 |
+
"step": 2420
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"epoch": 0.9356950327300732,
|
| 1708 |
+
"grad_norm": 0.1782931089401245,
|
| 1709 |
+
"learning_rate": 1.0934844619042595e-07,
|
| 1710 |
+
"loss": 0.7726,
|
| 1711 |
+
"step": 2430
|
| 1712 |
+
},
|
| 1713 |
+
{
|
| 1714 |
+
"epoch": 0.9395456295725837,
|
| 1715 |
+
"grad_norm": 0.17375195026397705,
|
| 1716 |
+
"learning_rate": 9.675911002803961e-08,
|
| 1717 |
+
"loss": 0.7634,
|
| 1718 |
+
"step": 2440
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 0.9433962264150944,
|
| 1722 |
+
"grad_norm": 0.1879037618637085,
|
| 1723 |
+
"learning_rate": 8.493241637904381e-08,
|
| 1724 |
+
"loss": 0.7506,
|
| 1725 |
+
"step": 2450
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 0.9472468232576049,
|
| 1729 |
+
"grad_norm": 0.18971092998981476,
|
| 1730 |
+
"learning_rate": 7.387020474909689e-08,
|
| 1731 |
+
"loss": 0.7786,
|
| 1732 |
+
"step": 2460
|
| 1733 |
+
},
|
| 1734 |
+
{
|
| 1735 |
+
"epoch": 0.9510974201001156,
|
| 1736 |
+
"grad_norm": 0.18743909895420074,
|
| 1737 |
+
"learning_rate": 6.357419573750856e-08,
|
| 1738 |
+
"loss": 0.773,
|
| 1739 |
+
"step": 2470
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 0.9549480169426261,
|
| 1743 |
+
"grad_norm": 0.1745530366897583,
|
| 1744 |
+
"learning_rate": 5.4045990769613365e-08,
|
| 1745 |
+
"loss": 0.7593,
|
| 1746 |
+
"step": 2480
|
| 1747 |
+
},
|
| 1748 |
+
{
|
| 1749 |
+
"epoch": 0.9587986137851366,
|
| 1750 |
+
"grad_norm": 0.17570818960666656,
|
| 1751 |
+
"learning_rate": 4.5287071847694434e-08,
|
| 1752 |
+
"loss": 0.7603,
|
| 1753 |
+
"step": 2490
|
| 1754 |
+
},
|
| 1755 |
+
{
|
| 1756 |
+
"epoch": 0.9626492106276473,
|
| 1757 |
+
"grad_norm": 0.17579486966133118,
|
| 1758 |
+
"learning_rate": 3.7298801320467817e-08,
|
| 1759 |
+
"loss": 0.7621,
|
| 1760 |
+
"step": 2500
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 0.9664998074701578,
|
| 1764 |
+
"grad_norm": 0.1743277609348297,
|
| 1765 |
+
"learning_rate": 3.0082421671192576e-08,
|
| 1766 |
+
"loss": 0.7611,
|
| 1767 |
+
"step": 2510
|
| 1768 |
+
},
|
| 1769 |
+
{
|
| 1770 |
+
"epoch": 0.9703504043126685,
|
| 1771 |
+
"grad_norm": 0.1863650381565094,
|
| 1772 |
+
"learning_rate": 2.3639055324410355e-08,
|
| 1773 |
+
"loss": 0.7748,
|
| 1774 |
+
"step": 2520
|
| 1775 |
+
},
|
| 1776 |
+
{
|
| 1777 |
+
"epoch": 0.974201001155179,
|
| 1778 |
+
"grad_norm": 0.17793843150138855,
|
| 1779 |
+
"learning_rate": 1.7969704471369477e-08,
|
| 1780 |
+
"loss": 0.7572,
|
| 1781 |
+
"step": 2530
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 0.9780515979976896,
|
| 1785 |
+
"grad_norm": 0.17045287787914276,
|
| 1786 |
+
"learning_rate": 1.3075250914142413e-08,
|
| 1787 |
+
"loss": 0.756,
|
| 1788 |
+
"step": 2540
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"epoch": 0.9819021948402002,
|
| 1792 |
+
"grad_norm": 0.16748173534870148,
|
| 1793 |
+
"learning_rate": 8.956455928476049e-09,
|
| 1794 |
+
"loss": 0.7719,
|
| 1795 |
+
"step": 2550
|
| 1796 |
+
},
|
| 1797 |
+
{
|
| 1798 |
+
"epoch": 0.9857527916827108,
|
| 1799 |
+
"grad_norm": 0.16757027804851532,
|
| 1800 |
+
"learning_rate": 5.6139601453786275e-09,
|
| 1801 |
+
"loss": 0.7568,
|
| 1802 |
+
"step": 2560
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 0.9896033885252215,
|
| 1806 |
+
"grad_norm": 0.17211169004440308,
|
| 1807 |
+
"learning_rate": 3.0482834514805602e-09,
|
| 1808 |
+
"loss": 0.7529,
|
| 1809 |
+
"step": 2570
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 0.993453985367732,
|
| 1813 |
+
"grad_norm": 0.17133864760398865,
|
| 1814 |
+
"learning_rate": 1.2598249081713366e-09,
|
| 1815 |
+
"loss": 0.7442,
|
| 1816 |
+
"step": 2580
|
| 1817 |
+
},
|
| 1818 |
+
{
|
| 1819 |
+
"epoch": 0.9973045822102425,
|
| 1820 |
+
"grad_norm": 0.1838298887014389,
|
| 1821 |
+
"learning_rate": 2.488626895302826e-10,
|
| 1822 |
+
"loss": 0.7757,
|
| 1823 |
+
"step": 2590
|
| 1824 |
+
},
|
| 1825 |
+
{
|
| 1826 |
+
"epoch": 1.0,
|
| 1827 |
+
"step": 2597,
|
| 1828 |
+
"total_flos": 5.975227850884645e+18,
|
| 1829 |
+
"train_loss": 0.776847034773094,
|
| 1830 |
+
"train_runtime": 14597.7139,
|
| 1831 |
+
"train_samples_per_second": 45.548,
|
| 1832 |
+
"train_steps_per_second": 0.178
|
| 1833 |
+
}
|
| 1834 |
+
],
|
| 1835 |
+
"logging_steps": 10,
|
| 1836 |
+
"max_steps": 2597,
|
| 1837 |
+
"num_input_tokens_seen": 584317136,
|
| 1838 |
+
"num_train_epochs": 1,
|
| 1839 |
+
"save_steps": 500,
|
| 1840 |
+
"stateful_callbacks": {
|
| 1841 |
+
"TrainerControl": {
|
| 1842 |
+
"args": {
|
| 1843 |
+
"should_epoch_stop": false,
|
| 1844 |
+
"should_evaluate": false,
|
| 1845 |
+
"should_log": false,
|
| 1846 |
+
"should_save": true,
|
| 1847 |
+
"should_training_stop": true
|
| 1848 |
+
},
|
| 1849 |
+
"attributes": {}
|
| 1850 |
+
}
|
| 1851 |
+
},
|
| 1852 |
+
"total_flos": 5.975227850884645e+18,
|
| 1853 |
+
"train_batch_size": 16,
|
| 1854 |
+
"trial_name": null,
|
| 1855 |
+
"trial_params": null
|
| 1856 |
+
}
|
continuous/smollm2/mc1_b32_224_2.5b/finetune/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6cc467d69de09e920700e8f58a1b8cc276bb950c8240d1b7fd61348891caf77c
|
| 3 |
+
size 6264
|
continuous/smollm2/mc1_b32_224_2.5b/finetune/vocab.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
continuous/smollm2/mc1_b32_224_2.5b/pretrain/config.json
ADDED
|
@@ -0,0 +1,70 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_faster_video": false,
|
| 3 |
+
"add_time_instruction": false,
|
| 4 |
+
"architectures": [
|
| 5 |
+
"LlamaForCausalLM"
|
| 6 |
+
],
|
| 7 |
+
"attention_bias": false,
|
| 8 |
+
"attention_dropout": 0.0,
|
| 9 |
+
"bos_token_id": 1,
|
| 10 |
+
"dtype": "bfloat16",
|
| 11 |
+
"eos_token_id": 2,
|
| 12 |
+
"faster_token_stride": 10,
|
| 13 |
+
"force_sample": false,
|
| 14 |
+
"head_dim": 64,
|
| 15 |
+
"hidden_act": "silu",
|
| 16 |
+
"hidden_size": 2048,
|
| 17 |
+
"image_aspect_ratio": "square",
|
| 18 |
+
"image_crop_resolution": null,
|
| 19 |
+
"image_grid_pinpoints": null,
|
| 20 |
+
"image_split_resolution": null,
|
| 21 |
+
"initializer_range": 0.02,
|
| 22 |
+
"intermediate_size": 8192,
|
| 23 |
+
"max_position_embeddings": 8192,
|
| 24 |
+
"mlp_bias": false,
|
| 25 |
+
"mm_hidden_size": 768,
|
| 26 |
+
"mm_newline_position": "grid",
|
| 27 |
+
"mm_patch_merge_type": "flat",
|
| 28 |
+
"mm_projector_lr": null,
|
| 29 |
+
"mm_projector_type": "mlp2x_gelu",
|
| 30 |
+
"mm_resampler_type": null,
|
| 31 |
+
"mm_spatial_pool_mode": "bilinear",
|
| 32 |
+
"mm_spatial_pool_stride": null,
|
| 33 |
+
"mm_tunable_parts": "mm_mlp_adapter",
|
| 34 |
+
"mm_use_im_patch_token": false,
|
| 35 |
+
"mm_use_im_start_end": false,
|
| 36 |
+
"mm_vision_select_feature": "patch",
|
| 37 |
+
"mm_vision_select_layer": -2,
|
| 38 |
+
"mm_vision_tower": "open_clip_hub:ViT-B-32",
|
| 39 |
+
"mm_vision_tower_lr": null,
|
| 40 |
+
"model_type": "llama",
|
| 41 |
+
"num_attention_heads": 32,
|
| 42 |
+
"num_hidden_layers": 24,
|
| 43 |
+
"num_key_value_heads": 32,
|
| 44 |
+
"pad_token_id": 0,
|
| 45 |
+
"pos_skipping_range": 4096,
|
| 46 |
+
"pretraining_tp": 1,
|
| 47 |
+
"rms_norm_eps": 1e-05,
|
| 48 |
+
"rope_scaling": null,
|
| 49 |
+
"rope_theta": 130000,
|
| 50 |
+
"tie_word_embeddings": true,
|
| 51 |
+
"tokenizer_model_max_length": 8192,
|
| 52 |
+
"tokenizer_padding_side": "right",
|
| 53 |
+
"transformers.js_config": {
|
| 54 |
+
"dtype": "q4",
|
| 55 |
+
"kv_cache_dtype": {
|
| 56 |
+
"fp16": "float16",
|
| 57 |
+
"q4f16": "float16"
|
| 58 |
+
},
|
| 59 |
+
"use_external_data_format": {
|
| 60 |
+
"model.onnx": true,
|
| 61 |
+
"model_fp16.onnx": true
|
| 62 |
+
}
|
| 63 |
+
},
|
| 64 |
+
"transformers_version": "4.57.6",
|
| 65 |
+
"use_cache": true,
|
| 66 |
+
"use_mm_proj": true,
|
| 67 |
+
"use_pos_skipping": false,
|
| 68 |
+
"vision_tower_pretrained": "/cache/ckpt/download/tokenizer/continuous/mc1/mc1_b32_224_2.5b.pt",
|
| 69 |
+
"vocab_size": 49152
|
| 70 |
+
}
|
continuous/smollm2/mc1_b32_224_2.5b/pretrain/mm_projector.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:55ad7a58c270123e9e77b546b615316ed506b72aff053d10324d579f1777e905
|
| 3 |
+
size 23087160
|
continuous/smollm2/mc1_b32_224_2.5b/pretrain/trainer_state.json
ADDED
|
@@ -0,0 +1,421 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 1.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 545,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.01834862385321101,
|
| 14 |
+
"grad_norm": 2.4924943447113037,
|
| 15 |
+
"learning_rate": 0.00032142857142857147,
|
| 16 |
+
"loss": 3.5416,
|
| 17 |
+
"step": 10
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.03669724770642202,
|
| 21 |
+
"grad_norm": 1.2021764516830444,
|
| 22 |
+
"learning_rate": 0.0006785714285714287,
|
| 23 |
+
"loss": 2.9147,
|
| 24 |
+
"step": 20
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.05504587155963303,
|
| 28 |
+
"grad_norm": 0.19937342405319214,
|
| 29 |
+
"learning_rate": 0.0009999907688176173,
|
| 30 |
+
"loss": 2.8116,
|
| 31 |
+
"step": 30
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.07339449541284404,
|
| 35 |
+
"grad_norm": 0.2077532857656479,
|
| 36 |
+
"learning_rate": 0.0009988834393115766,
|
| 37 |
+
"loss": 2.7453,
|
| 38 |
+
"step": 40
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.09174311926605505,
|
| 42 |
+
"grad_norm": 0.24068237841129303,
|
| 43 |
+
"learning_rate": 0.0009959345572940772,
|
| 44 |
+
"loss": 2.7248,
|
| 45 |
+
"step": 50
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.11009174311926606,
|
| 49 |
+
"grad_norm": 0.19938510656356812,
|
| 50 |
+
"learning_rate": 0.0009911550081155983,
|
| 51 |
+
"loss": 2.6838,
|
| 52 |
+
"step": 60
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.12844036697247707,
|
| 56 |
+
"grad_norm": 0.1496661901473999,
|
| 57 |
+
"learning_rate": 0.0009845624347566433,
|
| 58 |
+
"loss": 2.6412,
|
| 59 |
+
"step": 70
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.14678899082568808,
|
| 63 |
+
"grad_norm": 0.23687408864498138,
|
| 64 |
+
"learning_rate": 0.0009761811727013547,
|
| 65 |
+
"loss": 2.6045,
|
| 66 |
+
"step": 80
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.1651376146788991,
|
| 70 |
+
"grad_norm": 0.23572243750095367,
|
| 71 |
+
"learning_rate": 0.0009660421601067666,
|
| 72 |
+
"loss": 2.5712,
|
| 73 |
+
"step": 90
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.1834862385321101,
|
| 77 |
+
"grad_norm": 0.2259160876274109,
|
| 78 |
+
"learning_rate": 0.0009541828235992926,
|
| 79 |
+
"loss": 2.5484,
|
| 80 |
+
"step": 100
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.2018348623853211,
|
| 84 |
+
"grad_norm": 0.21544738113880157,
|
| 85 |
+
"learning_rate": 0.0009406469401200151,
|
| 86 |
+
"loss": 2.5195,
|
| 87 |
+
"step": 110
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.22018348623853212,
|
| 91 |
+
"grad_norm": 0.24060146510601044,
|
| 92 |
+
"learning_rate": 0.0009254844753287492,
|
| 93 |
+
"loss": 2.5058,
|
| 94 |
+
"step": 120
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.23853211009174313,
|
| 98 |
+
"grad_norm": 0.23342803120613098,
|
| 99 |
+
"learning_rate": 0.0009087513991633924,
|
| 100 |
+
"loss": 2.4805,
|
| 101 |
+
"step": 130
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.25688073394495414,
|
| 105 |
+
"grad_norm": 0.2516099214553833,
|
| 106 |
+
"learning_rate": 0.0008905094792353917,
|
| 107 |
+
"loss": 2.4626,
|
| 108 |
+
"step": 140
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.27522935779816515,
|
| 112 |
+
"grad_norm": 0.21936482191085815,
|
| 113 |
+
"learning_rate": 0.0008708260528239789,
|
| 114 |
+
"loss": 2.4613,
|
| 115 |
+
"step": 150
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.29357798165137616,
|
| 119 |
+
"grad_norm": 0.20484444499015808,
|
| 120 |
+
"learning_rate": 0.0008497737783108237,
|
| 121 |
+
"loss": 2.4369,
|
| 122 |
+
"step": 160
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.3119266055045872,
|
| 126 |
+
"grad_norm": 0.1828058511018753,
|
| 127 |
+
"learning_rate": 0.0008274303669726426,
|
| 128 |
+
"loss": 2.4227,
|
| 129 |
+
"step": 170
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.3302752293577982,
|
| 133 |
+
"grad_norm": 0.1932852864265442,
|
| 134 |
+
"learning_rate": 0.0008038782961218137,
|
| 135 |
+
"loss": 2.4032,
|
| 136 |
+
"step": 180
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.3486238532110092,
|
| 140 |
+
"grad_norm": 0.21173274517059326,
|
| 141 |
+
"learning_rate": 0.000779204504653894,
|
| 142 |
+
"loss": 2.3926,
|
| 143 |
+
"step": 190
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.3669724770642202,
|
| 147 |
+
"grad_norm": 0.19603575766086578,
|
| 148 |
+
"learning_rate": 0.0007535000721258791,
|
| 149 |
+
"loss": 2.3666,
|
| 150 |
+
"step": 200
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.3853211009174312,
|
| 154 |
+
"grad_norm": 0.19822457432746887,
|
| 155 |
+
"learning_rate": 0.00072685988254984,
|
| 156 |
+
"loss": 2.3603,
|
| 157 |
+
"step": 210
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.4036697247706422,
|
| 161 |
+
"grad_norm": 0.2273283749818802,
|
| 162 |
+
"learning_rate": 0.0006993822741429906,
|
| 163 |
+
"loss": 2.3663,
|
| 164 |
+
"step": 220
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.42201834862385323,
|
| 168 |
+
"grad_norm": 0.1874184012413025,
|
| 169 |
+
"learning_rate": 0.0006711686763270818,
|
| 170 |
+
"loss": 2.3546,
|
| 171 |
+
"step": 230
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.44036697247706424,
|
| 175 |
+
"grad_norm": 0.18195600807666779,
|
| 176 |
+
"learning_rate": 0.0006423232353170831,
|
| 177 |
+
"loss": 2.3437,
|
| 178 |
+
"step": 240
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.45871559633027525,
|
| 182 |
+
"grad_norm": 0.1985316127538681,
|
| 183 |
+
"learning_rate": 0.0006129524296812335,
|
| 184 |
+
"loss": 2.3387,
|
| 185 |
+
"step": 250
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.47706422018348627,
|
| 189 |
+
"grad_norm": 0.20488664507865906,
|
| 190 |
+
"learning_rate": 0.0005831646772915651,
|
| 191 |
+
"loss": 2.3452,
|
| 192 |
+
"step": 260
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.4954128440366973,
|
| 196 |
+
"grad_norm": 0.2367027848958969,
|
| 197 |
+
"learning_rate": 0.0005530699351157809,
|
| 198 |
+
"loss": 2.3221,
|
| 199 |
+
"step": 270
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.5137614678899083,
|
| 203 |
+
"grad_norm": 0.1752113699913025,
|
| 204 |
+
"learning_rate": 0.0005227792933277943,
|
| 205 |
+
"loss": 2.3294,
|
| 206 |
+
"step": 280
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.5321100917431193,
|
| 210 |
+
"grad_norm": 0.1663140207529068,
|
| 211 |
+
"learning_rate": 0.0004924045652352092,
|
| 212 |
+
"loss": 2.3135,
|
| 213 |
+
"step": 290
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.5504587155963303,
|
| 217 |
+
"grad_norm": 0.16536323726177216,
|
| 218 |
+
"learning_rate": 0.0004620578745374633,
|
| 219 |
+
"loss": 2.3143,
|
| 220 |
+
"step": 300
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.5688073394495413,
|
| 224 |
+
"grad_norm": 0.2159757912158966,
|
| 225 |
+
"learning_rate": 0.0004318512414382059,
|
| 226 |
+
"loss": 2.3197,
|
| 227 |
+
"step": 310
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.5871559633027523,
|
| 231 |
+
"grad_norm": 0.1734122782945633,
|
| 232 |
+
"learning_rate": 0.0004018961691397148,
|
| 233 |
+
"loss": 2.321,
|
| 234 |
+
"step": 320
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.6055045871559633,
|
| 238 |
+
"grad_norm": 0.24391157925128937,
|
| 239 |
+
"learning_rate": 0.0003723032322457458,
|
| 240 |
+
"loss": 2.304,
|
| 241 |
+
"step": 330
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.6238532110091743,
|
| 245 |
+
"grad_norm": 0.26052388548851013,
|
| 246 |
+
"learning_rate": 0.0003431816685921625,
|
| 247 |
+
"loss": 2.3082,
|
| 248 |
+
"step": 340
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.6422018348623854,
|
| 252 |
+
"grad_norm": 0.2261037826538086,
|
| 253 |
+
"learning_rate": 0.0003146389760120416,
|
| 254 |
+
"loss": 2.3095,
|
| 255 |
+
"step": 350
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.6605504587155964,
|
| 259 |
+
"grad_norm": 0.1853925734758377,
|
| 260 |
+
"learning_rate": 0.00028678051552373485,
|
| 261 |
+
"loss": 2.3098,
|
| 262 |
+
"step": 360
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.6788990825688074,
|
| 266 |
+
"grad_norm": 0.14747129380702972,
|
| 267 |
+
"learning_rate": 0.0002597091224066581,
|
| 268 |
+
"loss": 2.3032,
|
| 269 |
+
"step": 370
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.6972477064220184,
|
| 273 |
+
"grad_norm": 0.1802663952112198,
|
| 274 |
+
"learning_rate": 0.0002335247266004629,
|
| 275 |
+
"loss": 2.3112,
|
| 276 |
+
"step": 380
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.7155963302752294,
|
| 280 |
+
"grad_norm": 0.16374453902244568,
|
| 281 |
+
"learning_rate": 0.000208323983828828,
|
| 282 |
+
"loss": 2.2993,
|
| 283 |
+
"step": 390
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.7339449541284404,
|
| 287 |
+
"grad_norm": 0.1608005166053772,
|
| 288 |
+
"learning_rate": 0.0001841999188095224,
|
| 289 |
+
"loss": 2.2943,
|
| 290 |
+
"step": 400
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.7522935779816514,
|
| 294 |
+
"grad_norm": 0.16182105243206024,
|
| 295 |
+
"learning_rate": 0.00016124158186777676,
|
| 296 |
+
"loss": 2.2952,
|
| 297 |
+
"step": 410
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.7706422018348624,
|
| 301 |
+
"grad_norm": 0.1631917953491211,
|
| 302 |
+
"learning_rate": 0.0001395337202205211,
|
| 303 |
+
"loss": 2.2943,
|
| 304 |
+
"step": 420
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 0.7889908256880734,
|
| 308 |
+
"grad_norm": 0.16515028476715088,
|
| 309 |
+
"learning_rate": 0.00011915646514489292,
|
| 310 |
+
"loss": 2.291,
|
| 311 |
+
"step": 430
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.8073394495412844,
|
| 315 |
+
"grad_norm": 0.173843115568161,
|
| 316 |
+
"learning_rate": 0.00010018503618578818,
|
| 317 |
+
"loss": 2.2982,
|
| 318 |
+
"step": 440
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.8256880733944955,
|
| 322 |
+
"grad_norm": 0.16356681287288666,
|
| 323 |
+
"learning_rate": 8.268946349432583e-05,
|
| 324 |
+
"loss": 2.2978,
|
| 325 |
+
"step": 450
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.8440366972477065,
|
| 329 |
+
"grad_norm": 0.16404202580451965,
|
| 330 |
+
"learning_rate": 6.673432932217117e-05,
|
| 331 |
+
"loss": 2.2878,
|
| 332 |
+
"step": 460
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.8623853211009175,
|
| 336 |
+
"grad_norm": 0.16689321398735046,
|
| 337 |
+
"learning_rate": 5.2378529625954684e-05,
|
| 338 |
+
"loss": 2.2807,
|
| 339 |
+
"step": 470
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.8807339449541285,
|
| 343 |
+
"grad_norm": 0.15238650143146515,
|
| 344 |
+
"learning_rate": 3.9675056661785556e-05,
|
| 345 |
+
"loss": 2.2831,
|
| 346 |
+
"step": 480
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.8990825688073395,
|
| 350 |
+
"grad_norm": 0.15290220081806183,
|
| 351 |
+
"learning_rate": 2.8670803372379538e-05,
|
| 352 |
+
"loss": 2.2925,
|
| 353 |
+
"step": 490
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.9174311926605505,
|
| 357 |
+
"grad_norm": 0.14908067882061005,
|
| 358 |
+
"learning_rate": 1.9406390288876586e-05,
|
| 359 |
+
"loss": 2.295,
|
| 360 |
+
"step": 500
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.9357798165137615,
|
| 364 |
+
"grad_norm": 0.15913592278957367,
|
| 365 |
+
"learning_rate": 1.191601558631461e-05,
|
| 366 |
+
"loss": 2.2835,
|
| 367 |
+
"step": 510
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.9541284403669725,
|
| 371 |
+
"grad_norm": 0.15506014227867126,
|
| 372 |
+
"learning_rate": 6.2273288462559305e-06,
|
| 373 |
+
"loss": 2.2818,
|
| 374 |
+
"step": 520
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.9724770642201835,
|
| 378 |
+
"grad_norm": 0.1535671353340149,
|
| 379 |
+
"learning_rate": 2.361328992552314e-06,
|
| 380 |
+
"loss": 2.2846,
|
| 381 |
+
"step": 530
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.9908256880733946,
|
| 385 |
+
"grad_norm": 0.1600535362958908,
|
| 386 |
+
"learning_rate": 3.322867770044202e-07,
|
| 387 |
+
"loss": 2.2848,
|
| 388 |
+
"step": 540
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 1.0,
|
| 392 |
+
"step": 545,
|
| 393 |
+
"total_flos": 3.6873030393554534e+17,
|
| 394 |
+
"train_loss": 2.4202500964523455,
|
| 395 |
+
"train_runtime": 1367.5969,
|
| 396 |
+
"train_samples_per_second": 408.109,
|
| 397 |
+
"train_steps_per_second": 0.399
|
| 398 |
+
}
|
| 399 |
+
],
|
| 400 |
+
"logging_steps": 10,
|
| 401 |
+
"max_steps": 545,
|
| 402 |
+
"num_input_tokens_seen": 36058112,
|
| 403 |
+
"num_train_epochs": 1,
|
| 404 |
+
"save_steps": 500,
|
| 405 |
+
"stateful_callbacks": {
|
| 406 |
+
"TrainerControl": {
|
| 407 |
+
"args": {
|
| 408 |
+
"should_epoch_stop": false,
|
| 409 |
+
"should_evaluate": false,
|
| 410 |
+
"should_log": false,
|
| 411 |
+
"should_save": false,
|
| 412 |
+
"should_training_stop": false
|
| 413 |
+
},
|
| 414 |
+
"attributes": {}
|
| 415 |
+
}
|
| 416 |
+
},
|
| 417 |
+
"total_flos": 3.6873030393554534e+17,
|
| 418 |
+
"train_batch_size": 128,
|
| 419 |
+
"trial_name": null,
|
| 420 |
+
"trial_params": null
|
| 421 |
+
}
|
continuous/smollm2/mc1_b32_224_400m/finetune/chat_template.jinja
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{% for message in messages %}{% if loop.first and messages[0]['role'] != 'system' %}{{ '<|im_start|>system
|
| 2 |
+
You are a helpful AI assistant named SmolLM, trained by Hugging Face<|im_end|>
|
| 3 |
+
' }}{% endif %}{{'<|im_start|>' + message['role'] + '
|
| 4 |
+
' + message['content'] + '<|im_end|>' + '
|
| 5 |
+
'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant
|
| 6 |
+
' }}{% endif %}
|
continuous/smollm2/mc1_b32_224_400m/finetune/config.json
ADDED
|
@@ -0,0 +1,70 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_faster_video": false,
|
| 3 |
+
"add_time_instruction": false,
|
| 4 |
+
"architectures": [
|
| 5 |
+
"LlavaLlamaForCausalLM"
|
| 6 |
+
],
|
| 7 |
+
"attention_bias": false,
|
| 8 |
+
"attention_dropout": 0.0,
|
| 9 |
+
"bos_token_id": 1,
|
| 10 |
+
"dtype": "bfloat16",
|
| 11 |
+
"eos_token_id": 2,
|
| 12 |
+
"faster_token_stride": 10,
|
| 13 |
+
"force_sample": false,
|
| 14 |
+
"head_dim": 64,
|
| 15 |
+
"hidden_act": "silu",
|
| 16 |
+
"hidden_size": 2048,
|
| 17 |
+
"image_aspect_ratio": "square",
|
| 18 |
+
"image_crop_resolution": null,
|
| 19 |
+
"image_grid_pinpoints": null,
|
| 20 |
+
"image_split_resolution": null,
|
| 21 |
+
"initializer_range": 0.02,
|
| 22 |
+
"intermediate_size": 8192,
|
| 23 |
+
"max_position_embeddings": 8192,
|
| 24 |
+
"mlp_bias": false,
|
| 25 |
+
"mm_hidden_size": 768,
|
| 26 |
+
"mm_newline_position": "grid",
|
| 27 |
+
"mm_patch_merge_type": "flat",
|
| 28 |
+
"mm_projector_lr": 0.0001,
|
| 29 |
+
"mm_projector_type": "mlp2x_gelu",
|
| 30 |
+
"mm_resampler_type": null,
|
| 31 |
+
"mm_spatial_pool_mode": "bilinear",
|
| 32 |
+
"mm_spatial_pool_stride": null,
|
| 33 |
+
"mm_tunable_parts": "mm_mlp_adapter,mm_language_model",
|
| 34 |
+
"mm_use_im_patch_token": false,
|
| 35 |
+
"mm_use_im_start_end": false,
|
| 36 |
+
"mm_vision_select_feature": "patch",
|
| 37 |
+
"mm_vision_select_layer": -2,
|
| 38 |
+
"mm_vision_tower": "open_clip_hub:ViT-B-32",
|
| 39 |
+
"mm_vision_tower_lr": null,
|
| 40 |
+
"model_type": "llama",
|
| 41 |
+
"num_attention_heads": 32,
|
| 42 |
+
"num_hidden_layers": 24,
|
| 43 |
+
"num_key_value_heads": 32,
|
| 44 |
+
"pad_token_id": 0,
|
| 45 |
+
"pos_skipping_range": 4096,
|
| 46 |
+
"pretraining_tp": 1,
|
| 47 |
+
"rms_norm_eps": 1e-05,
|
| 48 |
+
"rope_scaling": null,
|
| 49 |
+
"rope_theta": 130000,
|
| 50 |
+
"tie_word_embeddings": true,
|
| 51 |
+
"tokenizer_model_max_length": 8192,
|
| 52 |
+
"tokenizer_padding_side": "right",
|
| 53 |
+
"transformers.js_config": {
|
| 54 |
+
"dtype": "q4",
|
| 55 |
+
"kv_cache_dtype": {
|
| 56 |
+
"fp16": "float16",
|
| 57 |
+
"q4f16": "float16"
|
| 58 |
+
},
|
| 59 |
+
"use_external_data_format": {
|
| 60 |
+
"model.onnx": true,
|
| 61 |
+
"model_fp16.onnx": true
|
| 62 |
+
}
|
| 63 |
+
},
|
| 64 |
+
"transformers_version": "4.57.6",
|
| 65 |
+
"use_cache": true,
|
| 66 |
+
"use_mm_proj": true,
|
| 67 |
+
"use_pos_skipping": false,
|
| 68 |
+
"vision_tower_pretrained": "/cache/ckpt/download/tokenizer/continuous/mc1/mc1_b32_224_400m.pt",
|
| 69 |
+
"vocab_size": 49152
|
| 70 |
+
}
|
continuous/smollm2/mc1_b32_224_400m/finetune/generation_config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 1,
|
| 4 |
+
"eos_token_id": [
|
| 5 |
+
2
|
| 6 |
+
],
|
| 7 |
+
"pad_token_id": 0,
|
| 8 |
+
"transformers_version": "4.57.6"
|
| 9 |
+
}
|
continuous/smollm2/mc1_b32_224_400m/finetune/merges.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
continuous/smollm2/mc1_b32_224_400m/finetune/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7d886ddcd9860f8efd9601829f1b579f8a52bb45ecd047c85f6a0113e133f75c
|
| 3 |
+
size 3822910424
|
continuous/smollm2/mc1_b32_224_400m/finetune/special_tokens_map.json
ADDED
|
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"additional_special_tokens": [
|
| 3 |
+
"<|im_start|>",
|
| 4 |
+
"<|im_end|>"
|
| 5 |
+
],
|
| 6 |
+
"bos_token": {
|
| 7 |
+
"content": "<|im_start|>",
|
| 8 |
+
"lstrip": false,
|
| 9 |
+
"normalized": false,
|
| 10 |
+
"rstrip": false,
|
| 11 |
+
"single_word": false
|
| 12 |
+
},
|
| 13 |
+
"eos_token": {
|
| 14 |
+
"content": "<|im_end|>",
|
| 15 |
+
"lstrip": false,
|
| 16 |
+
"normalized": false,
|
| 17 |
+
"rstrip": false,
|
| 18 |
+
"single_word": false
|
| 19 |
+
},
|
| 20 |
+
"pad_token": "<|endoftext|>",
|
| 21 |
+
"unk_token": {
|
| 22 |
+
"content": "<|endoftext|>",
|
| 23 |
+
"lstrip": false,
|
| 24 |
+
"normalized": false,
|
| 25 |
+
"rstrip": false,
|
| 26 |
+
"single_word": false
|
| 27 |
+
}
|
| 28 |
+
}
|
continuous/smollm2/mc1_b32_224_400m/finetune/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
continuous/smollm2/mc1_b32_224_400m/finetune/tokenizer_config.json
ADDED
|
@@ -0,0 +1,155 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_prefix_space": false,
|
| 3 |
+
"added_tokens_decoder": {
|
| 4 |
+
"0": {
|
| 5 |
+
"content": "<|endoftext|>",
|
| 6 |
+
"lstrip": false,
|
| 7 |
+
"normalized": false,
|
| 8 |
+
"rstrip": false,
|
| 9 |
+
"single_word": false,
|
| 10 |
+
"special": true
|
| 11 |
+
},
|
| 12 |
+
"1": {
|
| 13 |
+
"content": "<|im_start|>",
|
| 14 |
+
"lstrip": false,
|
| 15 |
+
"normalized": false,
|
| 16 |
+
"rstrip": false,
|
| 17 |
+
"single_word": false,
|
| 18 |
+
"special": true
|
| 19 |
+
},
|
| 20 |
+
"2": {
|
| 21 |
+
"content": "<|im_end|>",
|
| 22 |
+
"lstrip": false,
|
| 23 |
+
"normalized": false,
|
| 24 |
+
"rstrip": false,
|
| 25 |
+
"single_word": false,
|
| 26 |
+
"special": true
|
| 27 |
+
},
|
| 28 |
+
"3": {
|
| 29 |
+
"content": "<repo_name>",
|
| 30 |
+
"lstrip": false,
|
| 31 |
+
"normalized": false,
|
| 32 |
+
"rstrip": false,
|
| 33 |
+
"single_word": false,
|
| 34 |
+
"special": true
|
| 35 |
+
},
|
| 36 |
+
"4": {
|
| 37 |
+
"content": "<reponame>",
|
| 38 |
+
"lstrip": false,
|
| 39 |
+
"normalized": false,
|
| 40 |
+
"rstrip": false,
|
| 41 |
+
"single_word": false,
|
| 42 |
+
"special": true
|
| 43 |
+
},
|
| 44 |
+
"5": {
|
| 45 |
+
"content": "<file_sep>",
|
| 46 |
+
"lstrip": false,
|
| 47 |
+
"normalized": false,
|
| 48 |
+
"rstrip": false,
|
| 49 |
+
"single_word": false,
|
| 50 |
+
"special": true
|
| 51 |
+
},
|
| 52 |
+
"6": {
|
| 53 |
+
"content": "<filename>",
|
| 54 |
+
"lstrip": false,
|
| 55 |
+
"normalized": false,
|
| 56 |
+
"rstrip": false,
|
| 57 |
+
"single_word": false,
|
| 58 |
+
"special": true
|
| 59 |
+
},
|
| 60 |
+
"7": {
|
| 61 |
+
"content": "<gh_stars>",
|
| 62 |
+
"lstrip": false,
|
| 63 |
+
"normalized": false,
|
| 64 |
+
"rstrip": false,
|
| 65 |
+
"single_word": false,
|
| 66 |
+
"special": true
|
| 67 |
+
},
|
| 68 |
+
"8": {
|
| 69 |
+
"content": "<issue_start>",
|
| 70 |
+
"lstrip": false,
|
| 71 |
+
"normalized": false,
|
| 72 |
+
"rstrip": false,
|
| 73 |
+
"single_word": false,
|
| 74 |
+
"special": true
|
| 75 |
+
},
|
| 76 |
+
"9": {
|
| 77 |
+
"content": "<issue_comment>",
|
| 78 |
+
"lstrip": false,
|
| 79 |
+
"normalized": false,
|
| 80 |
+
"rstrip": false,
|
| 81 |
+
"single_word": false,
|
| 82 |
+
"special": true
|
| 83 |
+
},
|
| 84 |
+
"10": {
|
| 85 |
+
"content": "<issue_closed>",
|
| 86 |
+
"lstrip": false,
|
| 87 |
+
"normalized": false,
|
| 88 |
+
"rstrip": false,
|
| 89 |
+
"single_word": false,
|
| 90 |
+
"special": true
|
| 91 |
+
},
|
| 92 |
+
"11": {
|
| 93 |
+
"content": "<jupyter_start>",
|
| 94 |
+
"lstrip": false,
|
| 95 |
+
"normalized": false,
|
| 96 |
+
"rstrip": false,
|
| 97 |
+
"single_word": false,
|
| 98 |
+
"special": true
|
| 99 |
+
},
|
| 100 |
+
"12": {
|
| 101 |
+
"content": "<jupyter_text>",
|
| 102 |
+
"lstrip": false,
|
| 103 |
+
"normalized": false,
|
| 104 |
+
"rstrip": false,
|
| 105 |
+
"single_word": false,
|
| 106 |
+
"special": true
|
| 107 |
+
},
|
| 108 |
+
"13": {
|
| 109 |
+
"content": "<jupyter_code>",
|
| 110 |
+
"lstrip": false,
|
| 111 |
+
"normalized": false,
|
| 112 |
+
"rstrip": false,
|
| 113 |
+
"single_word": false,
|
| 114 |
+
"special": true
|
| 115 |
+
},
|
| 116 |
+
"14": {
|
| 117 |
+
"content": "<jupyter_output>",
|
| 118 |
+
"lstrip": false,
|
| 119 |
+
"normalized": false,
|
| 120 |
+
"rstrip": false,
|
| 121 |
+
"single_word": false,
|
| 122 |
+
"special": true
|
| 123 |
+
},
|
| 124 |
+
"15": {
|
| 125 |
+
"content": "<jupyter_script>",
|
| 126 |
+
"lstrip": false,
|
| 127 |
+
"normalized": false,
|
| 128 |
+
"rstrip": false,
|
| 129 |
+
"single_word": false,
|
| 130 |
+
"special": true
|
| 131 |
+
},
|
| 132 |
+
"16": {
|
| 133 |
+
"content": "<empty_output>",
|
| 134 |
+
"lstrip": false,
|
| 135 |
+
"normalized": false,
|
| 136 |
+
"rstrip": false,
|
| 137 |
+
"single_word": false,
|
| 138 |
+
"special": true
|
| 139 |
+
}
|
| 140 |
+
},
|
| 141 |
+
"additional_special_tokens": [
|
| 142 |
+
"<|im_start|>",
|
| 143 |
+
"<|im_end|>"
|
| 144 |
+
],
|
| 145 |
+
"bos_token": "<|im_start|>",
|
| 146 |
+
"clean_up_tokenization_spaces": false,
|
| 147 |
+
"eos_token": "<|im_end|>",
|
| 148 |
+
"extra_special_tokens": {},
|
| 149 |
+
"model_max_length": 8192,
|
| 150 |
+
"pad_token": "<|endoftext|>",
|
| 151 |
+
"padding_side": "right",
|
| 152 |
+
"tokenizer_class": "GPT2Tokenizer",
|
| 153 |
+
"unk_token": "<|endoftext|>",
|
| 154 |
+
"vocab_size": 49152
|
| 155 |
+
}
|
continuous/smollm2/mc1_b32_224_400m/finetune/trainer_state.json
ADDED
|
@@ -0,0 +1,1856 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 1.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 2597,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.003850596842510589,
|
| 14 |
+
"grad_norm": 3.595533847808838,
|
| 15 |
+
"learning_rate": 1.153846153846154e-06,
|
| 16 |
+
"loss": 1.3115,
|
| 17 |
+
"step": 10
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.007701193685021178,
|
| 21 |
+
"grad_norm": 2.3048336505889893,
|
| 22 |
+
"learning_rate": 2.435897435897436e-06,
|
| 23 |
+
"loss": 1.185,
|
| 24 |
+
"step": 20
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.011551790527531768,
|
| 28 |
+
"grad_norm": 2.0532186031341553,
|
| 29 |
+
"learning_rate": 3.7179487179487184e-06,
|
| 30 |
+
"loss": 1.0415,
|
| 31 |
+
"step": 30
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.015402387370042356,
|
| 35 |
+
"grad_norm": 1.4022222757339478,
|
| 36 |
+
"learning_rate": 5e-06,
|
| 37 |
+
"loss": 0.9853,
|
| 38 |
+
"step": 40
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.019252984212552945,
|
| 42 |
+
"grad_norm": 1.3086329698562622,
|
| 43 |
+
"learning_rate": 6.282051282051282e-06,
|
| 44 |
+
"loss": 0.9694,
|
| 45 |
+
"step": 50
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.023103581055063535,
|
| 49 |
+
"grad_norm": 0.34136155247688293,
|
| 50 |
+
"learning_rate": 7.564102564102564e-06,
|
| 51 |
+
"loss": 0.8983,
|
| 52 |
+
"step": 60
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.026954177897574125,
|
| 56 |
+
"grad_norm": 0.27412986755371094,
|
| 57 |
+
"learning_rate": 8.846153846153847e-06,
|
| 58 |
+
"loss": 0.8516,
|
| 59 |
+
"step": 70
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.030804774740084712,
|
| 63 |
+
"grad_norm": 0.2804144620895386,
|
| 64 |
+
"learning_rate": 9.999996111488723e-06,
|
| 65 |
+
"loss": 0.8633,
|
| 66 |
+
"step": 80
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.0346553715825953,
|
| 70 |
+
"grad_norm": 0.2501981258392334,
|
| 71 |
+
"learning_rate": 9.999529497453782e-06,
|
| 72 |
+
"loss": 0.8497,
|
| 73 |
+
"step": 90
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.03850596842510589,
|
| 77 |
+
"grad_norm": 0.24029213190078735,
|
| 78 |
+
"learning_rate": 9.998285264324193e-06,
|
| 79 |
+
"loss": 0.8195,
|
| 80 |
+
"step": 100
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.04235656526761648,
|
| 84 |
+
"grad_norm": 0.2280218005180359,
|
| 85 |
+
"learning_rate": 9.996263605626052e-06,
|
| 86 |
+
"loss": 0.8104,
|
| 87 |
+
"step": 110
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.04620716211012707,
|
| 91 |
+
"grad_norm": 0.22327172756195068,
|
| 92 |
+
"learning_rate": 9.99346483580503e-06,
|
| 93 |
+
"loss": 0.8302,
|
| 94 |
+
"step": 120
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.05005775895263766,
|
| 98 |
+
"grad_norm": 0.22010210156440735,
|
| 99 |
+
"learning_rate": 9.989889390177467e-06,
|
| 100 |
+
"loss": 0.8064,
|
| 101 |
+
"step": 130
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.05390835579514825,
|
| 105 |
+
"grad_norm": 0.23924896121025085,
|
| 106 |
+
"learning_rate": 9.985537824862647e-06,
|
| 107 |
+
"loss": 0.8194,
|
| 108 |
+
"step": 140
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.05775895263765884,
|
| 112 |
+
"grad_norm": 0.23586396872997284,
|
| 113 |
+
"learning_rate": 9.980410816696317e-06,
|
| 114 |
+
"loss": 0.7972,
|
| 115 |
+
"step": 150
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.061609549480169425,
|
| 119 |
+
"grad_norm": 0.23079901933670044,
|
| 120 |
+
"learning_rate": 9.97450916312541e-06,
|
| 121 |
+
"loss": 0.8027,
|
| 122 |
+
"step": 160
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.06546014632268002,
|
| 126 |
+
"grad_norm": 0.20307041704654694,
|
| 127 |
+
"learning_rate": 9.967833782084004e-06,
|
| 128 |
+
"loss": 0.8022,
|
| 129 |
+
"step": 170
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.0693107431651906,
|
| 133 |
+
"grad_norm": 0.20843824744224548,
|
| 134 |
+
"learning_rate": 9.960385711850554e-06,
|
| 135 |
+
"loss": 0.8332,
|
| 136 |
+
"step": 180
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.07316134000770119,
|
| 140 |
+
"grad_norm": 0.2243756204843521,
|
| 141 |
+
"learning_rate": 9.952166110886394e-06,
|
| 142 |
+
"loss": 0.8202,
|
| 143 |
+
"step": 190
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.07701193685021178,
|
| 147 |
+
"grad_norm": 0.22967787086963654,
|
| 148 |
+
"learning_rate": 9.943176257655567e-06,
|
| 149 |
+
"loss": 0.8178,
|
| 150 |
+
"step": 200
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.08086253369272237,
|
| 154 |
+
"grad_norm": 0.21346452832221985,
|
| 155 |
+
"learning_rate": 9.933417550425952e-06,
|
| 156 |
+
"loss": 0.8052,
|
| 157 |
+
"step": 210
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.08471313053523297,
|
| 161 |
+
"grad_norm": 0.2068740725517273,
|
| 162 |
+
"learning_rate": 9.9228915070518e-06,
|
| 163 |
+
"loss": 0.7963,
|
| 164 |
+
"step": 220
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.08856372737774355,
|
| 168 |
+
"grad_norm": 0.20086395740509033,
|
| 169 |
+
"learning_rate": 9.911599764737632e-06,
|
| 170 |
+
"loss": 0.7913,
|
| 171 |
+
"step": 230
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.09241432422025414,
|
| 175 |
+
"grad_norm": 0.2114594727754593,
|
| 176 |
+
"learning_rate": 9.899544079783606e-06,
|
| 177 |
+
"loss": 0.7849,
|
| 178 |
+
"step": 240
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.09626492106276473,
|
| 182 |
+
"grad_norm": 0.2043227106332779,
|
| 183 |
+
"learning_rate": 9.886726327312338e-06,
|
| 184 |
+
"loss": 0.8023,
|
| 185 |
+
"step": 250
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.10011551790527531,
|
| 189 |
+
"grad_norm": 0.21648824214935303,
|
| 190 |
+
"learning_rate": 9.87314850097724e-06,
|
| 191 |
+
"loss": 0.8009,
|
| 192 |
+
"step": 260
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.1039661147477859,
|
| 196 |
+
"grad_norm": 0.20135267078876495,
|
| 197 |
+
"learning_rate": 9.858812712652452e-06,
|
| 198 |
+
"loss": 0.7836,
|
| 199 |
+
"step": 270
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.1078167115902965,
|
| 203 |
+
"grad_norm": 0.18044470250606537,
|
| 204 |
+
"learning_rate": 9.843721192104341e-06,
|
| 205 |
+
"loss": 0.7809,
|
| 206 |
+
"step": 280
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.11166730843280709,
|
| 210 |
+
"grad_norm": 0.20001934468746185,
|
| 211 |
+
"learning_rate": 9.827876286644698e-06,
|
| 212 |
+
"loss": 0.8123,
|
| 213 |
+
"step": 290
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.11551790527531768,
|
| 217 |
+
"grad_norm": 0.189320370554924,
|
| 218 |
+
"learning_rate": 9.811280460765643e-06,
|
| 219 |
+
"loss": 0.7708,
|
| 220 |
+
"step": 300
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.11936850211782826,
|
| 224 |
+
"grad_norm": 0.19815504550933838,
|
| 225 |
+
"learning_rate": 9.793936295756292e-06,
|
| 226 |
+
"loss": 0.7757,
|
| 227 |
+
"step": 310
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.12321909896033885,
|
| 231 |
+
"grad_norm": 0.206741601228714,
|
| 232 |
+
"learning_rate": 9.775846489301279e-06,
|
| 233 |
+
"loss": 0.7845,
|
| 234 |
+
"step": 320
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.12706969580284944,
|
| 238 |
+
"grad_norm": 0.1894979476928711,
|
| 239 |
+
"learning_rate": 9.757013855061157e-06,
|
| 240 |
+
"loss": 0.787,
|
| 241 |
+
"step": 330
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.13092029264536004,
|
| 245 |
+
"grad_norm": 0.20134153962135315,
|
| 246 |
+
"learning_rate": 9.737441322234758e-06,
|
| 247 |
+
"loss": 0.7887,
|
| 248 |
+
"step": 340
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.1347708894878706,
|
| 252 |
+
"grad_norm": 0.2183346450328827,
|
| 253 |
+
"learning_rate": 9.717131935103605e-06,
|
| 254 |
+
"loss": 0.7959,
|
| 255 |
+
"step": 350
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.1386214863303812,
|
| 259 |
+
"grad_norm": 0.18161256611347198,
|
| 260 |
+
"learning_rate": 9.696088852558398e-06,
|
| 261 |
+
"loss": 0.7762,
|
| 262 |
+
"step": 360
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.1424720831728918,
|
| 266 |
+
"grad_norm": 0.20113760232925415,
|
| 267 |
+
"learning_rate": 9.674315347607686e-06,
|
| 268 |
+
"loss": 0.8043,
|
| 269 |
+
"step": 370
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.14632268001540238,
|
| 273 |
+
"grad_norm": 0.20212289690971375,
|
| 274 |
+
"learning_rate": 9.651814806868796e-06,
|
| 275 |
+
"loss": 0.7689,
|
| 276 |
+
"step": 380
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.15017327685791299,
|
| 280 |
+
"grad_norm": 0.20045772194862366,
|
| 281 |
+
"learning_rate": 9.62859073004107e-06,
|
| 282 |
+
"loss": 0.7718,
|
| 283 |
+
"step": 390
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.15402387370042356,
|
| 287 |
+
"grad_norm": 0.17607103288173676,
|
| 288 |
+
"learning_rate": 9.604646729361547e-06,
|
| 289 |
+
"loss": 0.78,
|
| 290 |
+
"step": 400
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.15787447054293416,
|
| 294 |
+
"grad_norm": 0.20219233632087708,
|
| 295 |
+
"learning_rate": 9.579986529043099e-06,
|
| 296 |
+
"loss": 0.7767,
|
| 297 |
+
"step": 410
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.16172506738544473,
|
| 301 |
+
"grad_norm": 0.19202937185764313,
|
| 302 |
+
"learning_rate": 9.554613964695189e-06,
|
| 303 |
+
"loss": 0.7909,
|
| 304 |
+
"step": 420
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 0.16557566422795533,
|
| 308 |
+
"grad_norm": 0.19343000650405884,
|
| 309 |
+
"learning_rate": 9.528532982727281e-06,
|
| 310 |
+
"loss": 0.7714,
|
| 311 |
+
"step": 430
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.16942626107046593,
|
| 315 |
+
"grad_norm": 0.18580776453018188,
|
| 316 |
+
"learning_rate": 9.501747639735018e-06,
|
| 317 |
+
"loss": 0.7797,
|
| 318 |
+
"step": 440
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.1732768579129765,
|
| 322 |
+
"grad_norm": 0.19166526198387146,
|
| 323 |
+
"learning_rate": 9.474262101869272e-06,
|
| 324 |
+
"loss": 0.7651,
|
| 325 |
+
"step": 450
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.1771274547554871,
|
| 329 |
+
"grad_norm": 0.21526563167572021,
|
| 330 |
+
"learning_rate": 9.446080644188141e-06,
|
| 331 |
+
"loss": 0.7739,
|
| 332 |
+
"step": 460
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.18097805159799768,
|
| 336 |
+
"grad_norm": 0.1975030153989792,
|
| 337 |
+
"learning_rate": 9.417207649992017e-06,
|
| 338 |
+
"loss": 0.7755,
|
| 339 |
+
"step": 470
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.18482864844050828,
|
| 343 |
+
"grad_norm": 0.20511198043823242,
|
| 344 |
+
"learning_rate": 9.387647610141816e-06,
|
| 345 |
+
"loss": 0.7764,
|
| 346 |
+
"step": 480
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.18867924528301888,
|
| 350 |
+
"grad_norm": 0.18795937299728394,
|
| 351 |
+
"learning_rate": 9.35740512236047e-06,
|
| 352 |
+
"loss": 0.7686,
|
| 353 |
+
"step": 490
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.19252984212552945,
|
| 357 |
+
"grad_norm": 0.18480385839939117,
|
| 358 |
+
"learning_rate": 9.326484890517809e-06,
|
| 359 |
+
"loss": 0.7724,
|
| 360 |
+
"step": 500
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.19638043896804006,
|
| 364 |
+
"grad_norm": 0.19882895052433014,
|
| 365 |
+
"learning_rate": 9.294891723898926e-06,
|
| 366 |
+
"loss": 0.762,
|
| 367 |
+
"step": 510
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.20023103581055063,
|
| 371 |
+
"grad_norm": 0.18767842650413513,
|
| 372 |
+
"learning_rate": 9.262630536456153e-06,
|
| 373 |
+
"loss": 0.7867,
|
| 374 |
+
"step": 520
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.20408163265306123,
|
| 378 |
+
"grad_norm": 0.19819511473178864,
|
| 379 |
+
"learning_rate": 9.229706346044749e-06,
|
| 380 |
+
"loss": 0.7684,
|
| 381 |
+
"step": 530
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.2079322294955718,
|
| 385 |
+
"grad_norm": 0.21290655434131622,
|
| 386 |
+
"learning_rate": 9.196124273642422e-06,
|
| 387 |
+
"loss": 0.7786,
|
| 388 |
+
"step": 540
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.2117828263380824,
|
| 392 |
+
"grad_norm": 0.19247612357139587,
|
| 393 |
+
"learning_rate": 9.161889542552842e-06,
|
| 394 |
+
"loss": 0.7845,
|
| 395 |
+
"step": 550
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.215633423180593,
|
| 399 |
+
"grad_norm": 0.20213238894939423,
|
| 400 |
+
"learning_rate": 9.127007477593193e-06,
|
| 401 |
+
"loss": 0.7866,
|
| 402 |
+
"step": 560
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.21948402002310358,
|
| 406 |
+
"grad_norm": 0.20958253741264343,
|
| 407 |
+
"learning_rate": 9.091483504265974e-06,
|
| 408 |
+
"loss": 0.8031,
|
| 409 |
+
"step": 570
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.22333461686561418,
|
| 413 |
+
"grad_norm": 0.2064582109451294,
|
| 414 |
+
"learning_rate": 9.055323147915116e-06,
|
| 415 |
+
"loss": 0.7694,
|
| 416 |
+
"step": 580
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.22718521370812475,
|
| 420 |
+
"grad_norm": 0.21315620839595795,
|
| 421 |
+
"learning_rate": 9.018532032866585e-06,
|
| 422 |
+
"loss": 0.7865,
|
| 423 |
+
"step": 590
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.23103581055063535,
|
| 427 |
+
"grad_norm": 0.1820932924747467,
|
| 428 |
+
"learning_rate": 8.981115881553579e-06,
|
| 429 |
+
"loss": 0.7828,
|
| 430 |
+
"step": 600
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 0.23488640739314592,
|
| 434 |
+
"grad_norm": 0.21416890621185303,
|
| 435 |
+
"learning_rate": 8.943080513626476e-06,
|
| 436 |
+
"loss": 0.8011,
|
| 437 |
+
"step": 610
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 0.23873700423565652,
|
| 441 |
+
"grad_norm": 0.1773897260427475,
|
| 442 |
+
"learning_rate": 8.904431845047641e-06,
|
| 443 |
+
"loss": 0.778,
|
| 444 |
+
"step": 620
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.24258760107816713,
|
| 448 |
+
"grad_norm": 0.18189263343811035,
|
| 449 |
+
"learning_rate": 8.86517588717128e-06,
|
| 450 |
+
"loss": 0.7909,
|
| 451 |
+
"step": 630
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 0.2464381979206777,
|
| 455 |
+
"grad_norm": 0.19914600253105164,
|
| 456 |
+
"learning_rate": 8.82531874580844e-06,
|
| 457 |
+
"loss": 0.7874,
|
| 458 |
+
"step": 640
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.25028879476318827,
|
| 462 |
+
"grad_norm": 0.17379818856716156,
|
| 463 |
+
"learning_rate": 8.784866620277314e-06,
|
| 464 |
+
"loss": 0.7499,
|
| 465 |
+
"step": 650
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.25413939160569887,
|
| 469 |
+
"grad_norm": 0.20106251537799835,
|
| 470 |
+
"learning_rate": 8.743825802439018e-06,
|
| 471 |
+
"loss": 0.7753,
|
| 472 |
+
"step": 660
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 0.2579899884482095,
|
| 476 |
+
"grad_norm": 0.18373987078666687,
|
| 477 |
+
"learning_rate": 8.702202675718956e-06,
|
| 478 |
+
"loss": 0.7831,
|
| 479 |
+
"step": 670
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.2618405852907201,
|
| 483 |
+
"grad_norm": 0.18779747188091278,
|
| 484 |
+
"learning_rate": 8.660003714113959e-06,
|
| 485 |
+
"loss": 0.7659,
|
| 486 |
+
"step": 680
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 0.2656911821332307,
|
| 490 |
+
"grad_norm": 0.1886136829853058,
|
| 491 |
+
"learning_rate": 8.617235481185323e-06,
|
| 492 |
+
"loss": 0.7751,
|
| 493 |
+
"step": 690
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 0.2695417789757412,
|
| 497 |
+
"grad_norm": 0.19052477180957794,
|
| 498 |
+
"learning_rate": 8.573904629037931e-06,
|
| 499 |
+
"loss": 0.7885,
|
| 500 |
+
"step": 700
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 0.2733923758182518,
|
| 504 |
+
"grad_norm": 0.18479368090629578,
|
| 505 |
+
"learning_rate": 8.530017897285585e-06,
|
| 506 |
+
"loss": 0.7691,
|
| 507 |
+
"step": 710
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.2772429726607624,
|
| 511 |
+
"grad_norm": 0.19049020111560822,
|
| 512 |
+
"learning_rate": 8.485582112002746e-06,
|
| 513 |
+
"loss": 0.7699,
|
| 514 |
+
"step": 720
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 0.281093569503273,
|
| 518 |
+
"grad_norm": 0.18384438753128052,
|
| 519 |
+
"learning_rate": 8.440604184662811e-06,
|
| 520 |
+
"loss": 0.7807,
|
| 521 |
+
"step": 730
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 0.2849441663457836,
|
| 525 |
+
"grad_norm": 0.18718111515045166,
|
| 526 |
+
"learning_rate": 8.395091111063118e-06,
|
| 527 |
+
"loss": 0.7654,
|
| 528 |
+
"step": 740
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 0.28879476318829417,
|
| 532 |
+
"grad_norm": 0.17094628512859344,
|
| 533 |
+
"learning_rate": 8.349049970236822e-06,
|
| 534 |
+
"loss": 0.7616,
|
| 535 |
+
"step": 750
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 0.29264536003080477,
|
| 539 |
+
"grad_norm": 0.18221288919448853,
|
| 540 |
+
"learning_rate": 8.302487923351842e-06,
|
| 541 |
+
"loss": 0.7602,
|
| 542 |
+
"step": 760
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 0.29649595687331537,
|
| 546 |
+
"grad_norm": 0.19451527297496796,
|
| 547 |
+
"learning_rate": 8.255412212597021e-06,
|
| 548 |
+
"loss": 0.7775,
|
| 549 |
+
"step": 770
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.30034655371582597,
|
| 553 |
+
"grad_norm": 0.26547208428382874,
|
| 554 |
+
"learning_rate": 8.207830160055688e-06,
|
| 555 |
+
"loss": 0.7729,
|
| 556 |
+
"step": 780
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 0.3041971505583365,
|
| 560 |
+
"grad_norm": 0.19694969058036804,
|
| 561 |
+
"learning_rate": 8.159749166566786e-06,
|
| 562 |
+
"loss": 0.7574,
|
| 563 |
+
"step": 790
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 0.3080477474008471,
|
| 567 |
+
"grad_norm": 0.19233858585357666,
|
| 568 |
+
"learning_rate": 8.11117671057377e-06,
|
| 569 |
+
"loss": 0.7788,
|
| 570 |
+
"step": 800
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 0.3118983442433577,
|
| 574 |
+
"grad_norm": 0.18351155519485474,
|
| 575 |
+
"learning_rate": 8.062120346961404e-06,
|
| 576 |
+
"loss": 0.7774,
|
| 577 |
+
"step": 810
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 0.3157489410858683,
|
| 581 |
+
"grad_norm": 0.18003684282302856,
|
| 582 |
+
"learning_rate": 8.012587705880705e-06,
|
| 583 |
+
"loss": 0.7751,
|
| 584 |
+
"step": 820
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 0.3195995379283789,
|
| 588 |
+
"grad_norm": 0.18485218286514282,
|
| 589 |
+
"learning_rate": 7.962586491562146e-06,
|
| 590 |
+
"loss": 0.7604,
|
| 591 |
+
"step": 830
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 0.32345013477088946,
|
| 595 |
+
"grad_norm": 0.18157586455345154,
|
| 596 |
+
"learning_rate": 7.912124481117357e-06,
|
| 597 |
+
"loss": 0.7604,
|
| 598 |
+
"step": 840
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 0.32730073161340006,
|
| 602 |
+
"grad_norm": 0.16809457540512085,
|
| 603 |
+
"learning_rate": 7.861209523329494e-06,
|
| 604 |
+
"loss": 0.7702,
|
| 605 |
+
"step": 850
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 0.33115132845591067,
|
| 609 |
+
"grad_norm": 0.19426926970481873,
|
| 610 |
+
"learning_rate": 7.809849537432432e-06,
|
| 611 |
+
"loss": 0.7624,
|
| 612 |
+
"step": 860
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 0.33500192529842127,
|
| 616 |
+
"grad_norm": 0.17366905510425568,
|
| 617 |
+
"learning_rate": 7.758052511879035e-06,
|
| 618 |
+
"loss": 0.7659,
|
| 619 |
+
"step": 870
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 0.33885252214093187,
|
| 623 |
+
"grad_norm": 0.18081077933311462,
|
| 624 |
+
"learning_rate": 7.705826503098642e-06,
|
| 625 |
+
"loss": 0.7692,
|
| 626 |
+
"step": 880
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 0.3427031189834424,
|
| 630 |
+
"grad_norm": 0.18441522121429443,
|
| 631 |
+
"learning_rate": 7.653179634243979e-06,
|
| 632 |
+
"loss": 0.7618,
|
| 633 |
+
"step": 890
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 0.346553715825953,
|
| 637 |
+
"grad_norm": 0.1794872134923935,
|
| 638 |
+
"learning_rate": 7.600120093927696e-06,
|
| 639 |
+
"loss": 0.7651,
|
| 640 |
+
"step": 900
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 0.3504043126684636,
|
| 644 |
+
"grad_norm": 0.2037183791399002,
|
| 645 |
+
"learning_rate": 7.546656134948729e-06,
|
| 646 |
+
"loss": 0.7551,
|
| 647 |
+
"step": 910
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 0.3542549095109742,
|
| 651 |
+
"grad_norm": 0.1703725904226303,
|
| 652 |
+
"learning_rate": 7.492796073008661e-06,
|
| 653 |
+
"loss": 0.7611,
|
| 654 |
+
"step": 920
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.3581055063534848,
|
| 658 |
+
"grad_norm": 0.19453218579292297,
|
| 659 |
+
"learning_rate": 7.438548285418326e-06,
|
| 660 |
+
"loss": 0.7427,
|
| 661 |
+
"step": 930
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.36195610319599536,
|
| 665 |
+
"grad_norm": 0.19357894361019135,
|
| 666 |
+
"learning_rate": 7.3839212097948e-06,
|
| 667 |
+
"loss": 0.7743,
|
| 668 |
+
"step": 940
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 0.36580670003850596,
|
| 672 |
+
"grad_norm": 0.19014203548431396,
|
| 673 |
+
"learning_rate": 7.328923342749039e-06,
|
| 674 |
+
"loss": 0.7911,
|
| 675 |
+
"step": 950
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.36965729688101656,
|
| 679 |
+
"grad_norm": 0.18132711946964264,
|
| 680 |
+
"learning_rate": 7.273563238564319e-06,
|
| 681 |
+
"loss": 0.786,
|
| 682 |
+
"step": 960
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 0.37350789372352716,
|
| 686 |
+
"grad_norm": 0.17752724885940552,
|
| 687 |
+
"learning_rate": 7.217849507865724e-06,
|
| 688 |
+
"loss": 0.7561,
|
| 689 |
+
"step": 970
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 0.37735849056603776,
|
| 693 |
+
"grad_norm": 0.17957748472690582,
|
| 694 |
+
"learning_rate": 7.161790816280858e-06,
|
| 695 |
+
"loss": 0.764,
|
| 696 |
+
"step": 980
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 0.3812090874085483,
|
| 700 |
+
"grad_norm": 0.1820351779460907,
|
| 701 |
+
"learning_rate": 7.1053958830920125e-06,
|
| 702 |
+
"loss": 0.7529,
|
| 703 |
+
"step": 990
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.3850596842510589,
|
| 707 |
+
"grad_norm": 0.182297021150589,
|
| 708 |
+
"learning_rate": 7.0486734798799715e-06,
|
| 709 |
+
"loss": 0.7632,
|
| 710 |
+
"step": 1000
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 0.3889102810935695,
|
| 714 |
+
"grad_norm": 0.17588675022125244,
|
| 715 |
+
"learning_rate": 6.991632429159708e-06,
|
| 716 |
+
"loss": 0.783,
|
| 717 |
+
"step": 1010
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 0.3927608779360801,
|
| 721 |
+
"grad_norm": 0.18616576492786407,
|
| 722 |
+
"learning_rate": 6.934281603008132e-06,
|
| 723 |
+
"loss": 0.7819,
|
| 724 |
+
"step": 1020
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"epoch": 0.39661147477859066,
|
| 728 |
+
"grad_norm": 0.1786801666021347,
|
| 729 |
+
"learning_rate": 6.876629921684143e-06,
|
| 730 |
+
"loss": 0.7775,
|
| 731 |
+
"step": 1030
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 0.40046207162110126,
|
| 735 |
+
"grad_norm": 0.18849773705005646,
|
| 736 |
+
"learning_rate": 6.818686352241199e-06,
|
| 737 |
+
"loss": 0.7663,
|
| 738 |
+
"step": 1040
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 0.40431266846361186,
|
| 742 |
+
"grad_norm": 0.165129616856575,
|
| 743 |
+
"learning_rate": 6.760459907132573e-06,
|
| 744 |
+
"loss": 0.7619,
|
| 745 |
+
"step": 1050
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 0.40816326530612246,
|
| 749 |
+
"grad_norm": 0.16250041127204895,
|
| 750 |
+
"learning_rate": 6.701959642809589e-06,
|
| 751 |
+
"loss": 0.7678,
|
| 752 |
+
"step": 1060
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 0.41201386214863306,
|
| 756 |
+
"grad_norm": 0.1803434044122696,
|
| 757 |
+
"learning_rate": 6.643194658312987e-06,
|
| 758 |
+
"loss": 0.7597,
|
| 759 |
+
"step": 1070
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.4158644589911436,
|
| 763 |
+
"grad_norm": 0.1708746999502182,
|
| 764 |
+
"learning_rate": 6.584174093857676e-06,
|
| 765 |
+
"loss": 0.7846,
|
| 766 |
+
"step": 1080
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 0.4197150558336542,
|
| 770 |
+
"grad_norm": 0.18368251621723175,
|
| 771 |
+
"learning_rate": 6.5249071294110735e-06,
|
| 772 |
+
"loss": 0.7664,
|
| 773 |
+
"step": 1090
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 0.4235656526761648,
|
| 777 |
+
"grad_norm": 0.1844271719455719,
|
| 778 |
+
"learning_rate": 6.465402983265274e-06,
|
| 779 |
+
"loss": 0.7789,
|
| 780 |
+
"step": 1100
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 0.4274162495186754,
|
| 784 |
+
"grad_norm": 0.20799516141414642,
|
| 785 |
+
"learning_rate": 6.405670910603243e-06,
|
| 786 |
+
"loss": 0.7728,
|
| 787 |
+
"step": 1110
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 0.431266846361186,
|
| 791 |
+
"grad_norm": 0.21114841103553772,
|
| 792 |
+
"learning_rate": 6.345720202059291e-06,
|
| 793 |
+
"loss": 0.769,
|
| 794 |
+
"step": 1120
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 0.43511744320369655,
|
| 798 |
+
"grad_norm": 0.1883717030286789,
|
| 799 |
+
"learning_rate": 6.285560182274008e-06,
|
| 800 |
+
"loss": 0.75,
|
| 801 |
+
"step": 1130
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 0.43896804004620715,
|
| 805 |
+
"grad_norm": 0.18615327775478363,
|
| 806 |
+
"learning_rate": 6.225200208443936e-06,
|
| 807 |
+
"loss": 0.7741,
|
| 808 |
+
"step": 1140
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 0.44281863688871775,
|
| 812 |
+
"grad_norm": 0.17930544912815094,
|
| 813 |
+
"learning_rate": 6.16464966886616e-06,
|
| 814 |
+
"loss": 0.7597,
|
| 815 |
+
"step": 1150
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 0.44666923373122835,
|
| 819 |
+
"grad_norm": 0.21181921660900116,
|
| 820 |
+
"learning_rate": 6.103917981478065e-06,
|
| 821 |
+
"loss": 0.7885,
|
| 822 |
+
"step": 1160
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 0.45051983057373896,
|
| 826 |
+
"grad_norm": 0.1692444086074829,
|
| 827 |
+
"learning_rate": 6.043014592392489e-06,
|
| 828 |
+
"loss": 0.7949,
|
| 829 |
+
"step": 1170
|
| 830 |
+
},
|
| 831 |
+
{
|
| 832 |
+
"epoch": 0.4543704274162495,
|
| 833 |
+
"grad_norm": 0.17423470318317413,
|
| 834 |
+
"learning_rate": 5.981948974428485e-06,
|
| 835 |
+
"loss": 0.7752,
|
| 836 |
+
"step": 1180
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 0.4582210242587601,
|
| 840 |
+
"grad_norm": 0.17478615045547485,
|
| 841 |
+
"learning_rate": 5.920730625637934e-06,
|
| 842 |
+
"loss": 0.7526,
|
| 843 |
+
"step": 1190
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 0.4620716211012707,
|
| 847 |
+
"grad_norm": 0.18078437447547913,
|
| 848 |
+
"learning_rate": 5.8593690678282355e-06,
|
| 849 |
+
"loss": 0.7597,
|
| 850 |
+
"step": 1200
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 0.4659222179437813,
|
| 854 |
+
"grad_norm": 0.19151222705841064,
|
| 855 |
+
"learning_rate": 5.797873845081301e-06,
|
| 856 |
+
"loss": 0.7576,
|
| 857 |
+
"step": 1210
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 0.46977281478629185,
|
| 861 |
+
"grad_norm": 0.18590691685676575,
|
| 862 |
+
"learning_rate": 5.736254522269075e-06,
|
| 863 |
+
"loss": 0.7843,
|
| 864 |
+
"step": 1220
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 0.47362341162880245,
|
| 868 |
+
"grad_norm": 0.17336712777614594,
|
| 869 |
+
"learning_rate": 5.6745206835658386e-06,
|
| 870 |
+
"loss": 0.7792,
|
| 871 |
+
"step": 1230
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 0.47747400847131305,
|
| 875 |
+
"grad_norm": 0.18198688328266144,
|
| 876 |
+
"learning_rate": 5.612681930957495e-06,
|
| 877 |
+
"loss": 0.7577,
|
| 878 |
+
"step": 1240
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 0.48132460531382365,
|
| 882 |
+
"grad_norm": 0.17733630537986755,
|
| 883 |
+
"learning_rate": 5.550747882748095e-06,
|
| 884 |
+
"loss": 0.7521,
|
| 885 |
+
"step": 1250
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 0.48517520215633425,
|
| 889 |
+
"grad_norm": 0.19920699298381805,
|
| 890 |
+
"learning_rate": 5.4887281720638195e-06,
|
| 891 |
+
"loss": 0.7646,
|
| 892 |
+
"step": 1260
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 0.4890257989988448,
|
| 896 |
+
"grad_norm": 0.18769611418247223,
|
| 897 |
+
"learning_rate": 5.42663244535466e-06,
|
| 898 |
+
"loss": 0.7419,
|
| 899 |
+
"step": 1270
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 0.4928763958413554,
|
| 903 |
+
"grad_norm": 0.19103506207466125,
|
| 904 |
+
"learning_rate": 5.3644703608940205e-06,
|
| 905 |
+
"loss": 0.7576,
|
| 906 |
+
"step": 1280
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 0.496726992683866,
|
| 910 |
+
"grad_norm": 0.17808368802070618,
|
| 911 |
+
"learning_rate": 5.302251587276491e-06,
|
| 912 |
+
"loss": 0.7702,
|
| 913 |
+
"step": 1290
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 0.5005775895263765,
|
| 917 |
+
"grad_norm": 0.17964743077754974,
|
| 918 |
+
"learning_rate": 5.2399858019140005e-06,
|
| 919 |
+
"loss": 0.7611,
|
| 920 |
+
"step": 1300
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 0.5044281863688872,
|
| 924 |
+
"grad_norm": 0.18350589275360107,
|
| 925 |
+
"learning_rate": 5.177682689530615e-06,
|
| 926 |
+
"loss": 0.7794,
|
| 927 |
+
"step": 1310
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 0.5082787832113977,
|
| 931 |
+
"grad_norm": 0.1852412074804306,
|
| 932 |
+
"learning_rate": 5.11535194065619e-06,
|
| 933 |
+
"loss": 0.7509,
|
| 934 |
+
"step": 1320
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 0.5121293800539084,
|
| 938 |
+
"grad_norm": 0.1878136396408081,
|
| 939 |
+
"learning_rate": 5.053003250119113e-06,
|
| 940 |
+
"loss": 0.7593,
|
| 941 |
+
"step": 1330
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 0.515979976896419,
|
| 945 |
+
"grad_norm": 0.19463691115379333,
|
| 946 |
+
"learning_rate": 4.9906463155383925e-06,
|
| 947 |
+
"loss": 0.7645,
|
| 948 |
+
"step": 1340
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 0.5198305737389295,
|
| 952 |
+
"grad_norm": 0.18010522425174713,
|
| 953 |
+
"learning_rate": 4.9282908358153046e-06,
|
| 954 |
+
"loss": 0.7615,
|
| 955 |
+
"step": 1350
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 0.5236811705814401,
|
| 959 |
+
"grad_norm": 0.1946858912706375,
|
| 960 |
+
"learning_rate": 4.8659465096248345e-06,
|
| 961 |
+
"loss": 0.7779,
|
| 962 |
+
"step": 1360
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 0.5275317674239507,
|
| 966 |
+
"grad_norm": 0.1954675018787384,
|
| 967 |
+
"learning_rate": 4.803623033907169e-06,
|
| 968 |
+
"loss": 0.7706,
|
| 969 |
+
"step": 1370
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 0.5313823642664613,
|
| 973 |
+
"grad_norm": 0.176618292927742,
|
| 974 |
+
"learning_rate": 4.741330102359441e-06,
|
| 975 |
+
"loss": 0.7841,
|
| 976 |
+
"step": 1380
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 0.5352329611089719,
|
| 980 |
+
"grad_norm": 0.19782030582427979,
|
| 981 |
+
"learning_rate": 4.679077403927996e-06,
|
| 982 |
+
"loss": 0.7597,
|
| 983 |
+
"step": 1390
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 0.5390835579514824,
|
| 987 |
+
"grad_norm": 0.18390175700187683,
|
| 988 |
+
"learning_rate": 4.616874621301373e-06,
|
| 989 |
+
"loss": 0.7603,
|
| 990 |
+
"step": 1400
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 0.5429341547939931,
|
| 994 |
+
"grad_norm": 0.18650151789188385,
|
| 995 |
+
"learning_rate": 4.554731429404293e-06,
|
| 996 |
+
"loss": 0.7777,
|
| 997 |
+
"step": 1410
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 0.5467847516365036,
|
| 1001 |
+
"grad_norm": 0.19368036091327667,
|
| 1002 |
+
"learning_rate": 4.492657493892822e-06,
|
| 1003 |
+
"loss": 0.762,
|
| 1004 |
+
"step": 1420
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 0.5506353484790143,
|
| 1008 |
+
"grad_norm": 0.1843925565481186,
|
| 1009 |
+
"learning_rate": 4.430662469650999e-06,
|
| 1010 |
+
"loss": 0.7743,
|
| 1011 |
+
"step": 1430
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 0.5544859453215248,
|
| 1015 |
+
"grad_norm": 0.19267871975898743,
|
| 1016 |
+
"learning_rate": 4.368755999289118e-06,
|
| 1017 |
+
"loss": 0.7746,
|
| 1018 |
+
"step": 1440
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 0.5583365421640354,
|
| 1022 |
+
"grad_norm": 0.1849762201309204,
|
| 1023 |
+
"learning_rate": 4.306947711643949e-06,
|
| 1024 |
+
"loss": 0.7591,
|
| 1025 |
+
"step": 1450
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 0.562187139006546,
|
| 1029 |
+
"grad_norm": 0.18153931200504303,
|
| 1030 |
+
"learning_rate": 4.245247220281068e-06,
|
| 1031 |
+
"loss": 0.7651,
|
| 1032 |
+
"step": 1460
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 0.5660377358490566,
|
| 1036 |
+
"grad_norm": 0.18136726319789886,
|
| 1037 |
+
"learning_rate": 4.183664121999587e-06,
|
| 1038 |
+
"loss": 0.7613,
|
| 1039 |
+
"step": 1470
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"epoch": 0.5698883326915672,
|
| 1043 |
+
"grad_norm": 0.1842404007911682,
|
| 1044 |
+
"learning_rate": 4.122207995339474e-06,
|
| 1045 |
+
"loss": 0.7529,
|
| 1046 |
+
"step": 1480
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 0.5737389295340778,
|
| 1050 |
+
"grad_norm": 0.16961759328842163,
|
| 1051 |
+
"learning_rate": 4.060888399091734e-06,
|
| 1052 |
+
"loss": 0.7627,
|
| 1053 |
+
"step": 1490
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 0.5775895263765883,
|
| 1057 |
+
"grad_norm": 0.17846152186393738,
|
| 1058 |
+
"learning_rate": 3.999714870811639e-06,
|
| 1059 |
+
"loss": 0.7737,
|
| 1060 |
+
"step": 1500
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 0.581440123219099,
|
| 1064 |
+
"grad_norm": 0.1844281405210495,
|
| 1065 |
+
"learning_rate": 3.938696925335275e-06,
|
| 1066 |
+
"loss": 0.7674,
|
| 1067 |
+
"step": 1510
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 0.5852907200616095,
|
| 1071 |
+
"grad_norm": 0.17977382242679596,
|
| 1072 |
+
"learning_rate": 3.87784405329962e-06,
|
| 1073 |
+
"loss": 0.7752,
|
| 1074 |
+
"step": 1520
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 0.5891413169041202,
|
| 1078 |
+
"grad_norm": 0.16912299394607544,
|
| 1079 |
+
"learning_rate": 3.817165719666382e-06,
|
| 1080 |
+
"loss": 0.7754,
|
| 1081 |
+
"step": 1530
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 0.5929919137466307,
|
| 1085 |
+
"grad_norm": 0.18019342422485352,
|
| 1086 |
+
"learning_rate": 3.756671362249841e-06,
|
| 1087 |
+
"loss": 0.7767,
|
| 1088 |
+
"step": 1540
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 0.5968425105891413,
|
| 1092 |
+
"grad_norm": 0.17296551167964935,
|
| 1093 |
+
"learning_rate": 3.6963703902488907e-06,
|
| 1094 |
+
"loss": 0.7646,
|
| 1095 |
+
"step": 1550
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 0.6006931074316519,
|
| 1099 |
+
"grad_norm": 0.17508560419082642,
|
| 1100 |
+
"learning_rate": 3.6362721827835655e-06,
|
| 1101 |
+
"loss": 0.7786,
|
| 1102 |
+
"step": 1560
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 0.6045437042741625,
|
| 1106 |
+
"grad_norm": 0.19022397696971893,
|
| 1107 |
+
"learning_rate": 3.576386087436212e-06,
|
| 1108 |
+
"loss": 0.781,
|
| 1109 |
+
"step": 1570
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 0.608394301116673,
|
| 1113 |
+
"grad_norm": 0.19272534549236298,
|
| 1114 |
+
"learning_rate": 3.5167214187975913e-06,
|
| 1115 |
+
"loss": 0.7747,
|
| 1116 |
+
"step": 1580
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 0.6122448979591837,
|
| 1120 |
+
"grad_norm": 0.15803010761737823,
|
| 1121 |
+
"learning_rate": 3.457287457018086e-06,
|
| 1122 |
+
"loss": 0.7694,
|
| 1123 |
+
"step": 1590
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 0.6160954948016942,
|
| 1127 |
+
"grad_norm": 0.17525801062583923,
|
| 1128 |
+
"learning_rate": 3.3980934463643003e-06,
|
| 1129 |
+
"loss": 0.765,
|
| 1130 |
+
"step": 1600
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 0.6199460916442049,
|
| 1134 |
+
"grad_norm": 0.16603899002075195,
|
| 1135 |
+
"learning_rate": 3.3391485937812053e-06,
|
| 1136 |
+
"loss": 0.7424,
|
| 1137 |
+
"step": 1610
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 0.6237966884867154,
|
| 1141 |
+
"grad_norm": 0.17662331461906433,
|
| 1142 |
+
"learning_rate": 3.280462067460115e-06,
|
| 1143 |
+
"loss": 0.7449,
|
| 1144 |
+
"step": 1620
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 0.627647285329226,
|
| 1148 |
+
"grad_norm": 0.17884600162506104,
|
| 1149 |
+
"learning_rate": 3.222042995412669e-06,
|
| 1150 |
+
"loss": 0.7704,
|
| 1151 |
+
"step": 1630
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 0.6314978821717366,
|
| 1155 |
+
"grad_norm": 0.16368567943572998,
|
| 1156 |
+
"learning_rate": 3.163900464051089e-06,
|
| 1157 |
+
"loss": 0.7613,
|
| 1158 |
+
"step": 1640
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"epoch": 0.6353484790142472,
|
| 1162 |
+
"grad_norm": 0.18273191154003143,
|
| 1163 |
+
"learning_rate": 3.1060435167748826e-06,
|
| 1164 |
+
"loss": 0.776,
|
| 1165 |
+
"step": 1650
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 0.6391990758567578,
|
| 1169 |
+
"grad_norm": 0.18097540736198425,
|
| 1170 |
+
"learning_rate": 3.048481152564253e-06,
|
| 1171 |
+
"loss": 0.7524,
|
| 1172 |
+
"step": 1660
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 0.6430496726992684,
|
| 1176 |
+
"grad_norm": 0.18518507480621338,
|
| 1177 |
+
"learning_rate": 2.9912223245804017e-06,
|
| 1178 |
+
"loss": 0.771,
|
| 1179 |
+
"step": 1670
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 0.6469002695417789,
|
| 1183 |
+
"grad_norm": 0.1912342756986618,
|
| 1184 |
+
"learning_rate": 2.934275938772981e-06,
|
| 1185 |
+
"loss": 0.7635,
|
| 1186 |
+
"step": 1680
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 0.6507508663842896,
|
| 1190 |
+
"grad_norm": 0.1883031576871872,
|
| 1191 |
+
"learning_rate": 2.8776508524948676e-06,
|
| 1192 |
+
"loss": 0.7586,
|
| 1193 |
+
"step": 1690
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 0.6546014632268001,
|
| 1197 |
+
"grad_norm": 0.16946600377559662,
|
| 1198 |
+
"learning_rate": 2.821355873124503e-06,
|
| 1199 |
+
"loss": 0.7613,
|
| 1200 |
+
"step": 1700
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 0.6584520600693108,
|
| 1204 |
+
"grad_norm": 0.19103579223155975,
|
| 1205 |
+
"learning_rate": 2.7653997566960144e-06,
|
| 1206 |
+
"loss": 0.7652,
|
| 1207 |
+
"step": 1710
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 0.6623026569118213,
|
| 1211 |
+
"grad_norm": 0.19570384919643402,
|
| 1212 |
+
"learning_rate": 2.709791206537309e-06,
|
| 1213 |
+
"loss": 0.7585,
|
| 1214 |
+
"step": 1720
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 0.6661532537543319,
|
| 1218 |
+
"grad_norm": 0.18786662817001343,
|
| 1219 |
+
"learning_rate": 2.6545388719163747e-06,
|
| 1220 |
+
"loss": 0.7662,
|
| 1221 |
+
"step": 1730
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 0.6700038505968425,
|
| 1225 |
+
"grad_norm": 0.1833648532629013,
|
| 1226 |
+
"learning_rate": 2.599651346695979e-06,
|
| 1227 |
+
"loss": 0.7736,
|
| 1228 |
+
"step": 1740
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 0.6738544474393531,
|
| 1232 |
+
"grad_norm": 0.17637798190116882,
|
| 1233 |
+
"learning_rate": 2.545137167997007e-06,
|
| 1234 |
+
"loss": 0.7655,
|
| 1235 |
+
"step": 1750
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 0.6777050442818637,
|
| 1239 |
+
"grad_norm": 0.18396182358264923,
|
| 1240 |
+
"learning_rate": 2.4910048148705972e-06,
|
| 1241 |
+
"loss": 0.7609,
|
| 1242 |
+
"step": 1760
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 0.6815556411243743,
|
| 1246 |
+
"grad_norm": 0.18467749655246735,
|
| 1247 |
+
"learning_rate": 2.4372627069793327e-06,
|
| 1248 |
+
"loss": 0.7947,
|
| 1249 |
+
"step": 1770
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 0.6854062379668848,
|
| 1253 |
+
"grad_norm": 0.17588379979133606,
|
| 1254 |
+
"learning_rate": 2.3839192032876486e-06,
|
| 1255 |
+
"loss": 0.7716,
|
| 1256 |
+
"step": 1780
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 0.6892568348093955,
|
| 1260 |
+
"grad_norm": 0.18606217205524445,
|
| 1261 |
+
"learning_rate": 2.3309826007617047e-06,
|
| 1262 |
+
"loss": 0.7875,
|
| 1263 |
+
"step": 1790
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 0.693107431651906,
|
| 1267 |
+
"grad_norm": 0.17875328660011292,
|
| 1268 |
+
"learning_rate": 2.278461133078881e-06,
|
| 1269 |
+
"loss": 0.7553,
|
| 1270 |
+
"step": 1800
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 0.6969580284944167,
|
| 1274 |
+
"grad_norm": 0.18708276748657227,
|
| 1275 |
+
"learning_rate": 2.2263629693471258e-06,
|
| 1276 |
+
"loss": 0.7494,
|
| 1277 |
+
"step": 1810
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 0.7008086253369272,
|
| 1281 |
+
"grad_norm": 0.17505298554897308,
|
| 1282 |
+
"learning_rate": 2.1746962128343414e-06,
|
| 1283 |
+
"loss": 0.7732,
|
| 1284 |
+
"step": 1820
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 0.7046592221794378,
|
| 1288 |
+
"grad_norm": 0.17824627459049225,
|
| 1289 |
+
"learning_rate": 2.1234688997080183e-06,
|
| 1290 |
+
"loss": 0.756,
|
| 1291 |
+
"step": 1830
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 0.7085098190219484,
|
| 1295 |
+
"grad_norm": 0.18059316277503967,
|
| 1296 |
+
"learning_rate": 2.0726889977853053e-06,
|
| 1297 |
+
"loss": 0.7612,
|
| 1298 |
+
"step": 1840
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 0.712360415864459,
|
| 1302 |
+
"grad_norm": 0.17944778501987457,
|
| 1303 |
+
"learning_rate": 2.022364405293703e-06,
|
| 1304 |
+
"loss": 0.7751,
|
| 1305 |
+
"step": 1850
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 0.7162110127069696,
|
| 1309 |
+
"grad_norm": 0.17491385340690613,
|
| 1310 |
+
"learning_rate": 1.9725029496425814e-06,
|
| 1311 |
+
"loss": 0.7728,
|
| 1312 |
+
"step": 1860
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 0.7200616095494802,
|
| 1316 |
+
"grad_norm": 0.18272943794727325,
|
| 1317 |
+
"learning_rate": 1.923112386205727e-06,
|
| 1318 |
+
"loss": 0.7636,
|
| 1319 |
+
"step": 1870
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 0.7239122063919907,
|
| 1323 |
+
"grad_norm": 0.17867524921894073,
|
| 1324 |
+
"learning_rate": 1.874200397115079e-06,
|
| 1325 |
+
"loss": 0.7746,
|
| 1326 |
+
"step": 1880
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 0.7277628032345014,
|
| 1330 |
+
"grad_norm": 0.1796174943447113,
|
| 1331 |
+
"learning_rate": 1.825774590065857e-06,
|
| 1332 |
+
"loss": 0.7609,
|
| 1333 |
+
"step": 1890
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 0.7316134000770119,
|
| 1337 |
+
"grad_norm": 0.19310863316059113,
|
| 1338 |
+
"learning_rate": 1.7778424971332813e-06,
|
| 1339 |
+
"loss": 0.7707,
|
| 1340 |
+
"step": 1900
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 0.7354639969195226,
|
| 1344 |
+
"grad_norm": 0.17040301859378815,
|
| 1345 |
+
"learning_rate": 1.7304115736010407e-06,
|
| 1346 |
+
"loss": 0.77,
|
| 1347 |
+
"step": 1910
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 0.7393145937620331,
|
| 1351 |
+
"grad_norm": 0.1800745129585266,
|
| 1352 |
+
"learning_rate": 1.6834891968017108e-06,
|
| 1353 |
+
"loss": 0.7586,
|
| 1354 |
+
"step": 1920
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 0.7431651906045437,
|
| 1358 |
+
"grad_norm": 0.18720895051956177,
|
| 1359 |
+
"learning_rate": 1.6370826649692884e-06,
|
| 1360 |
+
"loss": 0.7783,
|
| 1361 |
+
"step": 1930
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 0.7470157874470543,
|
| 1365 |
+
"grad_norm": 0.17494580149650574,
|
| 1366 |
+
"learning_rate": 1.5911991961040463e-06,
|
| 1367 |
+
"loss": 0.7749,
|
| 1368 |
+
"step": 1940
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 0.7508663842895649,
|
| 1372 |
+
"grad_norm": 0.18061463534832,
|
| 1373 |
+
"learning_rate": 1.545845926849851e-06,
|
| 1374 |
+
"loss": 0.7569,
|
| 1375 |
+
"step": 1950
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 0.7547169811320755,
|
| 1379 |
+
"grad_norm": 0.19259820878505707,
|
| 1380 |
+
"learning_rate": 1.5010299113841397e-06,
|
| 1381 |
+
"loss": 0.7828,
|
| 1382 |
+
"step": 1960
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 0.7585675779745861,
|
| 1386 |
+
"grad_norm": 0.17582263052463531,
|
| 1387 |
+
"learning_rate": 1.4567581203207238e-06,
|
| 1388 |
+
"loss": 0.7744,
|
| 1389 |
+
"step": 1970
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 0.7624181748170966,
|
| 1393 |
+
"grad_norm": 0.20081813633441925,
|
| 1394 |
+
"learning_rate": 1.413037439625593e-06,
|
| 1395 |
+
"loss": 0.7719,
|
| 1396 |
+
"step": 1980
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 0.7662687716596073,
|
| 1400 |
+
"grad_norm": 0.2015567570924759,
|
| 1401 |
+
"learning_rate": 1.3698746695458847e-06,
|
| 1402 |
+
"loss": 0.7604,
|
| 1403 |
+
"step": 1990
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 0.7701193685021178,
|
| 1407 |
+
"grad_norm": 0.17931988835334778,
|
| 1408 |
+
"learning_rate": 1.3272765235521745e-06,
|
| 1409 |
+
"loss": 0.7626,
|
| 1410 |
+
"step": 2000
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 0.7739699653446284,
|
| 1414 |
+
"grad_norm": 0.17461808025836945,
|
| 1415 |
+
"learning_rate": 1.2852496272942877e-06,
|
| 1416 |
+
"loss": 0.7681,
|
| 1417 |
+
"step": 2010
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 0.777820562187139,
|
| 1421 |
+
"grad_norm": 0.16841629147529602,
|
| 1422 |
+
"learning_rate": 1.2438005175707396e-06,
|
| 1423 |
+
"loss": 0.7509,
|
| 1424 |
+
"step": 2020
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 0.7816711590296496,
|
| 1428 |
+
"grad_norm": 0.17277155816555023,
|
| 1429 |
+
"learning_rate": 1.2029356413120297e-06,
|
| 1430 |
+
"loss": 0.7461,
|
| 1431 |
+
"step": 2030
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 0.7855217558721602,
|
| 1435 |
+
"grad_norm": 0.18536098301410675,
|
| 1436 |
+
"learning_rate": 1.1626613545778786e-06,
|
| 1437 |
+
"loss": 0.7612,
|
| 1438 |
+
"step": 2040
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 0.7893723527146708,
|
| 1442 |
+
"grad_norm": 0.1674879640340805,
|
| 1443 |
+
"learning_rate": 1.1229839215686294e-06,
|
| 1444 |
+
"loss": 0.7636,
|
| 1445 |
+
"step": 2050
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 0.7932229495571813,
|
| 1449 |
+
"grad_norm": 0.16451922059059143,
|
| 1450 |
+
"learning_rate": 1.0839095136509188e-06,
|
| 1451 |
+
"loss": 0.7488,
|
| 1452 |
+
"step": 2060
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 0.797073546399692,
|
| 1456 |
+
"grad_norm": 0.18709605932235718,
|
| 1457 |
+
"learning_rate": 1.045444208397791e-06,
|
| 1458 |
+
"loss": 0.7665,
|
| 1459 |
+
"step": 2070
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 0.8009241432422025,
|
| 1463 |
+
"grad_norm": 0.1725429743528366,
|
| 1464 |
+
"learning_rate": 1.007593988643402e-06,
|
| 1465 |
+
"loss": 0.7427,
|
| 1466 |
+
"step": 2080
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 0.8047747400847132,
|
| 1470 |
+
"grad_norm": 0.16911064088344574,
|
| 1471 |
+
"learning_rate": 9.703647415524613e-07,
|
| 1472 |
+
"loss": 0.7689,
|
| 1473 |
+
"step": 2090
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 0.8086253369272237,
|
| 1477 |
+
"grad_norm": 0.17104431986808777,
|
| 1478 |
+
"learning_rate": 9.337622577045485e-07,
|
| 1479 |
+
"loss": 0.7621,
|
| 1480 |
+
"step": 2100
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"epoch": 0.8124759337697343,
|
| 1484 |
+
"grad_norm": 0.17540930211544037,
|
| 1485 |
+
"learning_rate": 8.977922301934588e-07,
|
| 1486 |
+
"loss": 0.7633,
|
| 1487 |
+
"step": 2110
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 0.8163265306122449,
|
| 1491 |
+
"grad_norm": 0.1884843111038208,
|
| 1492 |
+
"learning_rate": 8.624602537417009e-07,
|
| 1493 |
+
"loss": 0.7779,
|
| 1494 |
+
"step": 2120
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 0.8201771274547555,
|
| 1498 |
+
"grad_norm": 0.17875108122825623,
|
| 1499 |
+
"learning_rate": 8.277718238303123e-07,
|
| 1500 |
+
"loss": 0.7598,
|
| 1501 |
+
"step": 2130
|
| 1502 |
+
},
|
| 1503 |
+
{
|
| 1504 |
+
"epoch": 0.8240277242972661,
|
| 1505 |
+
"grad_norm": 0.17374736070632935,
|
| 1506 |
+
"learning_rate": 7.937323358440935e-07,
|
| 1507 |
+
"loss": 0.7619,
|
| 1508 |
+
"step": 2140
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 0.8278783211397767,
|
| 1512 |
+
"grad_norm": 0.18400409817695618,
|
| 1513 |
+
"learning_rate": 7.603470842324223e-07,
|
| 1514 |
+
"loss": 0.7512,
|
| 1515 |
+
"step": 2150
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"epoch": 0.8317289179822872,
|
| 1519 |
+
"grad_norm": 0.17466437816619873,
|
| 1520 |
+
"learning_rate": 7.276212616857575e-07,
|
| 1521 |
+
"loss": 0.7722,
|
| 1522 |
+
"step": 2160
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"epoch": 0.8355795148247979,
|
| 1526 |
+
"grad_norm": 0.17787925899028778,
|
| 1527 |
+
"learning_rate": 6.955599583279837e-07,
|
| 1528 |
+
"loss": 0.766,
|
| 1529 |
+
"step": 2170
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 0.8394301116673084,
|
| 1533 |
+
"grad_norm": 0.17496667802333832,
|
| 1534 |
+
"learning_rate": 6.641681609246981e-07,
|
| 1535 |
+
"loss": 0.7609,
|
| 1536 |
+
"step": 2180
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 0.8432807085098191,
|
| 1540 |
+
"grad_norm": 0.18895865976810455,
|
| 1541 |
+
"learning_rate": 6.334507521075711e-07,
|
| 1542 |
+
"loss": 0.7631,
|
| 1543 |
+
"step": 2190
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 0.8471313053523296,
|
| 1547 |
+
"grad_norm": 0.19027301669120789,
|
| 1548 |
+
"learning_rate": 6.034125096149179e-07,
|
| 1549 |
+
"loss": 0.7651,
|
| 1550 |
+
"step": 2200
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 0.8509819021948402,
|
| 1554 |
+
"grad_norm": 0.187191903591156,
|
| 1555 |
+
"learning_rate": 5.740581055485722e-07,
|
| 1556 |
+
"loss": 0.7357,
|
| 1557 |
+
"step": 2210
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 0.8548324990373508,
|
| 1561 |
+
"grad_norm": 0.17203845083713531,
|
| 1562 |
+
"learning_rate": 5.453921056471961e-07,
|
| 1563 |
+
"loss": 0.7693,
|
| 1564 |
+
"step": 2220
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 0.8586830958798614,
|
| 1568 |
+
"grad_norm": 0.190839022397995,
|
| 1569 |
+
"learning_rate": 5.174189685761288e-07,
|
| 1570 |
+
"loss": 0.7619,
|
| 1571 |
+
"step": 2230
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 0.862533692722372,
|
| 1575 |
+
"grad_norm": 0.16851457953453064,
|
| 1576 |
+
"learning_rate": 4.901430452338979e-07,
|
| 1577 |
+
"loss": 0.7733,
|
| 1578 |
+
"step": 2240
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 0.8663842895648826,
|
| 1582 |
+
"grad_norm": 0.1746158003807068,
|
| 1583 |
+
"learning_rate": 4.6356857807548286e-07,
|
| 1584 |
+
"loss": 0.7609,
|
| 1585 |
+
"step": 2250
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 0.8702348864073931,
|
| 1589 |
+
"grad_norm": 0.2034357190132141,
|
| 1590 |
+
"learning_rate": 4.376997004524541e-07,
|
| 1591 |
+
"loss": 0.7438,
|
| 1592 |
+
"step": 2260
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 0.8740854832499038,
|
| 1596 |
+
"grad_norm": 0.16974875330924988,
|
| 1597 |
+
"learning_rate": 4.12540435970073e-07,
|
| 1598 |
+
"loss": 0.7631,
|
| 1599 |
+
"step": 2270
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 0.8779360800924143,
|
| 1603 |
+
"grad_norm": 0.1701684445142746,
|
| 1604 |
+
"learning_rate": 3.880946978614703e-07,
|
| 1605 |
+
"loss": 0.7668,
|
| 1606 |
+
"step": 2280
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"epoch": 0.881786676934925,
|
| 1610 |
+
"grad_norm": 0.1747182458639145,
|
| 1611 |
+
"learning_rate": 3.643662883789878e-07,
|
| 1612 |
+
"loss": 0.7726,
|
| 1613 |
+
"step": 2290
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 0.8856372737774355,
|
| 1617 |
+
"grad_norm": 0.16895291209220886,
|
| 1618 |
+
"learning_rate": 3.41358898202781e-07,
|
| 1619 |
+
"loss": 0.7525,
|
| 1620 |
+
"step": 2300
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 0.889487870619946,
|
| 1624 |
+
"grad_norm": 0.17260903120040894,
|
| 1625 |
+
"learning_rate": 3.190761058667741e-07,
|
| 1626 |
+
"loss": 0.7639,
|
| 1627 |
+
"step": 2310
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 0.8933384674624567,
|
| 1631 |
+
"grad_norm": 0.16682130098342896,
|
| 1632 |
+
"learning_rate": 2.9752137720206463e-07,
|
| 1633 |
+
"loss": 0.7665,
|
| 1634 |
+
"step": 2320
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 0.8971890643049673,
|
| 1638 |
+
"grad_norm": 0.17828993499279022,
|
| 1639 |
+
"learning_rate": 2.7669806479785067e-07,
|
| 1640 |
+
"loss": 0.7762,
|
| 1641 |
+
"step": 2330
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 0.9010396611474779,
|
| 1645 |
+
"grad_norm": 0.1745750457048416,
|
| 1646 |
+
"learning_rate": 2.566094074799763e-07,
|
| 1647 |
+
"loss": 0.7675,
|
| 1648 |
+
"step": 2340
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 0.9048902579899885,
|
| 1652 |
+
"grad_norm": 0.17699599266052246,
|
| 1653 |
+
"learning_rate": 2.3725852980716557e-07,
|
| 1654 |
+
"loss": 0.7473,
|
| 1655 |
+
"step": 2350
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 0.908740854832499,
|
| 1659 |
+
"grad_norm": 0.1686660647392273,
|
| 1660 |
+
"learning_rate": 2.1864844158503983e-07,
|
| 1661 |
+
"loss": 0.7639,
|
| 1662 |
+
"step": 2360
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 0.9125914516750097,
|
| 1666 |
+
"grad_norm": 0.1927846074104309,
|
| 1667 |
+
"learning_rate": 2.007820373979702e-07,
|
| 1668 |
+
"loss": 0.7566,
|
| 1669 |
+
"step": 2370
|
| 1670 |
+
},
|
| 1671 |
+
{
|
| 1672 |
+
"epoch": 0.9164420485175202,
|
| 1673 |
+
"grad_norm": 0.17016184329986572,
|
| 1674 |
+
"learning_rate": 1.8366209615886532e-07,
|
| 1675 |
+
"loss": 0.766,
|
| 1676 |
+
"step": 2380
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 0.9202926453600309,
|
| 1680 |
+
"grad_norm": 0.17298370599746704,
|
| 1681 |
+
"learning_rate": 1.6729128067693613e-07,
|
| 1682 |
+
"loss": 0.778,
|
| 1683 |
+
"step": 2390
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 0.9241432422025414,
|
| 1687 |
+
"grad_norm": 0.1942056119441986,
|
| 1688 |
+
"learning_rate": 1.5167213724353426e-07,
|
| 1689 |
+
"loss": 0.7717,
|
| 1690 |
+
"step": 2400
|
| 1691 |
+
},
|
| 1692 |
+
{
|
| 1693 |
+
"epoch": 0.927993839045052,
|
| 1694 |
+
"grad_norm": 0.1704406589269638,
|
| 1695 |
+
"learning_rate": 1.3680709523610058e-07,
|
| 1696 |
+
"loss": 0.7501,
|
| 1697 |
+
"step": 2410
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"epoch": 0.9318444358875626,
|
| 1701 |
+
"grad_norm": 0.16945718228816986,
|
| 1702 |
+
"learning_rate": 1.2269846674030418e-07,
|
| 1703 |
+
"loss": 0.7811,
|
| 1704 |
+
"step": 2420
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"epoch": 0.9356950327300732,
|
| 1708 |
+
"grad_norm": 0.17950332164764404,
|
| 1709 |
+
"learning_rate": 1.0934844619042595e-07,
|
| 1710 |
+
"loss": 0.7737,
|
| 1711 |
+
"step": 2430
|
| 1712 |
+
},
|
| 1713 |
+
{
|
| 1714 |
+
"epoch": 0.9395456295725837,
|
| 1715 |
+
"grad_norm": 0.1754617542028427,
|
| 1716 |
+
"learning_rate": 9.675911002803961e-08,
|
| 1717 |
+
"loss": 0.7643,
|
| 1718 |
+
"step": 2440
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 0.9433962264150944,
|
| 1722 |
+
"grad_norm": 0.1852453202009201,
|
| 1723 |
+
"learning_rate": 8.493241637904381e-08,
|
| 1724 |
+
"loss": 0.752,
|
| 1725 |
+
"step": 2450
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 0.9472468232576049,
|
| 1729 |
+
"grad_norm": 0.2000853419303894,
|
| 1730 |
+
"learning_rate": 7.387020474909689e-08,
|
| 1731 |
+
"loss": 0.7795,
|
| 1732 |
+
"step": 2460
|
| 1733 |
+
},
|
| 1734 |
+
{
|
| 1735 |
+
"epoch": 0.9510974201001156,
|
| 1736 |
+
"grad_norm": 0.21519646048545837,
|
| 1737 |
+
"learning_rate": 6.357419573750856e-08,
|
| 1738 |
+
"loss": 0.774,
|
| 1739 |
+
"step": 2470
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 0.9549480169426261,
|
| 1743 |
+
"grad_norm": 0.17473869025707245,
|
| 1744 |
+
"learning_rate": 5.4045990769613365e-08,
|
| 1745 |
+
"loss": 0.7603,
|
| 1746 |
+
"step": 2480
|
| 1747 |
+
},
|
| 1748 |
+
{
|
| 1749 |
+
"epoch": 0.9587986137851366,
|
| 1750 |
+
"grad_norm": 0.1719547063112259,
|
| 1751 |
+
"learning_rate": 4.5287071847694434e-08,
|
| 1752 |
+
"loss": 0.7617,
|
| 1753 |
+
"step": 2490
|
| 1754 |
+
},
|
| 1755 |
+
{
|
| 1756 |
+
"epoch": 0.9626492106276473,
|
| 1757 |
+
"grad_norm": 0.16891182959079742,
|
| 1758 |
+
"learning_rate": 3.7298801320467817e-08,
|
| 1759 |
+
"loss": 0.7629,
|
| 1760 |
+
"step": 2500
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 0.9664998074701578,
|
| 1764 |
+
"grad_norm": 0.17253223061561584,
|
| 1765 |
+
"learning_rate": 3.0082421671192576e-08,
|
| 1766 |
+
"loss": 0.7617,
|
| 1767 |
+
"step": 2510
|
| 1768 |
+
},
|
| 1769 |
+
{
|
| 1770 |
+
"epoch": 0.9703504043126685,
|
| 1771 |
+
"grad_norm": 0.18243873119354248,
|
| 1772 |
+
"learning_rate": 2.3639055324410355e-08,
|
| 1773 |
+
"loss": 0.7756,
|
| 1774 |
+
"step": 2520
|
| 1775 |
+
},
|
| 1776 |
+
{
|
| 1777 |
+
"epoch": 0.974201001155179,
|
| 1778 |
+
"grad_norm": 0.17357204854488373,
|
| 1779 |
+
"learning_rate": 1.7969704471369477e-08,
|
| 1780 |
+
"loss": 0.7568,
|
| 1781 |
+
"step": 2530
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 0.9780515979976896,
|
| 1785 |
+
"grad_norm": 0.170266255736351,
|
| 1786 |
+
"learning_rate": 1.3075250914142413e-08,
|
| 1787 |
+
"loss": 0.7565,
|
| 1788 |
+
"step": 2540
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"epoch": 0.9819021948402002,
|
| 1792 |
+
"grad_norm": 0.1690375953912735,
|
| 1793 |
+
"learning_rate": 8.956455928476049e-09,
|
| 1794 |
+
"loss": 0.7725,
|
| 1795 |
+
"step": 2550
|
| 1796 |
+
},
|
| 1797 |
+
{
|
| 1798 |
+
"epoch": 0.9857527916827108,
|
| 1799 |
+
"grad_norm": 0.16802605986595154,
|
| 1800 |
+
"learning_rate": 5.6139601453786275e-09,
|
| 1801 |
+
"loss": 0.7579,
|
| 1802 |
+
"step": 2560
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 0.9896033885252215,
|
| 1806 |
+
"grad_norm": 0.17523124814033508,
|
| 1807 |
+
"learning_rate": 3.0482834514805602e-09,
|
| 1808 |
+
"loss": 0.7534,
|
| 1809 |
+
"step": 2570
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 0.993453985367732,
|
| 1813 |
+
"grad_norm": 0.16995380818843842,
|
| 1814 |
+
"learning_rate": 1.2598249081713366e-09,
|
| 1815 |
+
"loss": 0.7449,
|
| 1816 |
+
"step": 2580
|
| 1817 |
+
},
|
| 1818 |
+
{
|
| 1819 |
+
"epoch": 0.9973045822102425,
|
| 1820 |
+
"grad_norm": 0.1810821294784546,
|
| 1821 |
+
"learning_rate": 2.488626895302826e-10,
|
| 1822 |
+
"loss": 0.7764,
|
| 1823 |
+
"step": 2590
|
| 1824 |
+
},
|
| 1825 |
+
{
|
| 1826 |
+
"epoch": 1.0,
|
| 1827 |
+
"step": 2597,
|
| 1828 |
+
"total_flos": 5.975227850884645e+18,
|
| 1829 |
+
"train_loss": 0.7788908106113884,
|
| 1830 |
+
"train_runtime": 14429.7522,
|
| 1831 |
+
"train_samples_per_second": 46.079,
|
| 1832 |
+
"train_steps_per_second": 0.18
|
| 1833 |
+
}
|
| 1834 |
+
],
|
| 1835 |
+
"logging_steps": 10,
|
| 1836 |
+
"max_steps": 2597,
|
| 1837 |
+
"num_input_tokens_seen": 584317136,
|
| 1838 |
+
"num_train_epochs": 1,
|
| 1839 |
+
"save_steps": 500,
|
| 1840 |
+
"stateful_callbacks": {
|
| 1841 |
+
"TrainerControl": {
|
| 1842 |
+
"args": {
|
| 1843 |
+
"should_epoch_stop": false,
|
| 1844 |
+
"should_evaluate": false,
|
| 1845 |
+
"should_log": false,
|
| 1846 |
+
"should_save": true,
|
| 1847 |
+
"should_training_stop": true
|
| 1848 |
+
},
|
| 1849 |
+
"attributes": {}
|
| 1850 |
+
}
|
| 1851 |
+
},
|
| 1852 |
+
"total_flos": 5.975227850884645e+18,
|
| 1853 |
+
"train_batch_size": 16,
|
| 1854 |
+
"trial_name": null,
|
| 1855 |
+
"trial_params": null
|
| 1856 |
+
}
|
continuous/smollm2/mc1_b32_224_400m/finetune/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:feabacf737e8598a82575a862f7a426900b130f7e9ab0dee9b2b9120415acf40
|
| 3 |
+
size 6264
|
continuous/smollm2/mc1_b32_224_400m/finetune/vocab.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
continuous/smollm2/mc1_b32_224_400m/pretrain/config.json
ADDED
|
@@ -0,0 +1,70 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_faster_video": false,
|
| 3 |
+
"add_time_instruction": false,
|
| 4 |
+
"architectures": [
|
| 5 |
+
"LlamaForCausalLM"
|
| 6 |
+
],
|
| 7 |
+
"attention_bias": false,
|
| 8 |
+
"attention_dropout": 0.0,
|
| 9 |
+
"bos_token_id": 1,
|
| 10 |
+
"dtype": "bfloat16",
|
| 11 |
+
"eos_token_id": 2,
|
| 12 |
+
"faster_token_stride": 10,
|
| 13 |
+
"force_sample": false,
|
| 14 |
+
"head_dim": 64,
|
| 15 |
+
"hidden_act": "silu",
|
| 16 |
+
"hidden_size": 2048,
|
| 17 |
+
"image_aspect_ratio": "square",
|
| 18 |
+
"image_crop_resolution": null,
|
| 19 |
+
"image_grid_pinpoints": null,
|
| 20 |
+
"image_split_resolution": null,
|
| 21 |
+
"initializer_range": 0.02,
|
| 22 |
+
"intermediate_size": 8192,
|
| 23 |
+
"max_position_embeddings": 8192,
|
| 24 |
+
"mlp_bias": false,
|
| 25 |
+
"mm_hidden_size": 768,
|
| 26 |
+
"mm_newline_position": "grid",
|
| 27 |
+
"mm_patch_merge_type": "flat",
|
| 28 |
+
"mm_projector_lr": null,
|
| 29 |
+
"mm_projector_type": "mlp2x_gelu",
|
| 30 |
+
"mm_resampler_type": null,
|
| 31 |
+
"mm_spatial_pool_mode": "bilinear",
|
| 32 |
+
"mm_spatial_pool_stride": null,
|
| 33 |
+
"mm_tunable_parts": "mm_mlp_adapter",
|
| 34 |
+
"mm_use_im_patch_token": false,
|
| 35 |
+
"mm_use_im_start_end": false,
|
| 36 |
+
"mm_vision_select_feature": "patch",
|
| 37 |
+
"mm_vision_select_layer": -2,
|
| 38 |
+
"mm_vision_tower": "open_clip_hub:ViT-B-32",
|
| 39 |
+
"mm_vision_tower_lr": null,
|
| 40 |
+
"model_type": "llama",
|
| 41 |
+
"num_attention_heads": 32,
|
| 42 |
+
"num_hidden_layers": 24,
|
| 43 |
+
"num_key_value_heads": 32,
|
| 44 |
+
"pad_token_id": 0,
|
| 45 |
+
"pos_skipping_range": 4096,
|
| 46 |
+
"pretraining_tp": 1,
|
| 47 |
+
"rms_norm_eps": 1e-05,
|
| 48 |
+
"rope_scaling": null,
|
| 49 |
+
"rope_theta": 130000,
|
| 50 |
+
"tie_word_embeddings": true,
|
| 51 |
+
"tokenizer_model_max_length": 8192,
|
| 52 |
+
"tokenizer_padding_side": "right",
|
| 53 |
+
"transformers.js_config": {
|
| 54 |
+
"dtype": "q4",
|
| 55 |
+
"kv_cache_dtype": {
|
| 56 |
+
"fp16": "float16",
|
| 57 |
+
"q4f16": "float16"
|
| 58 |
+
},
|
| 59 |
+
"use_external_data_format": {
|
| 60 |
+
"model.onnx": true,
|
| 61 |
+
"model_fp16.onnx": true
|
| 62 |
+
}
|
| 63 |
+
},
|
| 64 |
+
"transformers_version": "4.57.6",
|
| 65 |
+
"use_cache": true,
|
| 66 |
+
"use_mm_proj": true,
|
| 67 |
+
"use_pos_skipping": false,
|
| 68 |
+
"vision_tower_pretrained": "/cache/ckpt/download/tokenizer/continuous/mc1/mc1_b32_224_400m.pt",
|
| 69 |
+
"vocab_size": 49152
|
| 70 |
+
}
|
continuous/smollm2/mc1_b32_224_400m/pretrain/mm_projector.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0e86f85c16d1911f527d30fbe5144ad03ccb11bc6b774238d5eda88a2b783839
|
| 3 |
+
size 23087160
|
continuous/smollm2/mc1_b32_224_400m/pretrain/trainer_state.json
ADDED
|
@@ -0,0 +1,421 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 1.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 545,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.01834862385321101,
|
| 14 |
+
"grad_norm": 9.987513542175293,
|
| 15 |
+
"learning_rate": 0.00032142857142857147,
|
| 16 |
+
"loss": 3.6183,
|
| 17 |
+
"step": 10
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.03669724770642202,
|
| 21 |
+
"grad_norm": 0.7574322819709778,
|
| 22 |
+
"learning_rate": 0.0006785714285714287,
|
| 23 |
+
"loss": 2.9392,
|
| 24 |
+
"step": 20
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.05504587155963303,
|
| 28 |
+
"grad_norm": 0.6218893527984619,
|
| 29 |
+
"learning_rate": 0.0009999907688176173,
|
| 30 |
+
"loss": 2.8127,
|
| 31 |
+
"step": 30
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.07339449541284404,
|
| 35 |
+
"grad_norm": 0.20031295716762543,
|
| 36 |
+
"learning_rate": 0.0009988834393115766,
|
| 37 |
+
"loss": 2.7413,
|
| 38 |
+
"step": 40
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.09174311926605505,
|
| 42 |
+
"grad_norm": 0.16991810500621796,
|
| 43 |
+
"learning_rate": 0.0009959345572940772,
|
| 44 |
+
"loss": 2.6842,
|
| 45 |
+
"step": 50
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.11009174311926606,
|
| 49 |
+
"grad_norm": 0.2222665697336197,
|
| 50 |
+
"learning_rate": 0.0009911550081155983,
|
| 51 |
+
"loss": 2.6301,
|
| 52 |
+
"step": 60
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.12844036697247707,
|
| 56 |
+
"grad_norm": 0.20103929936885834,
|
| 57 |
+
"learning_rate": 0.0009845624347566433,
|
| 58 |
+
"loss": 2.5924,
|
| 59 |
+
"step": 70
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.14678899082568808,
|
| 63 |
+
"grad_norm": 0.25329989194869995,
|
| 64 |
+
"learning_rate": 0.0009761811727013547,
|
| 65 |
+
"loss": 2.5603,
|
| 66 |
+
"step": 80
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.1651376146788991,
|
| 70 |
+
"grad_norm": 0.23735052347183228,
|
| 71 |
+
"learning_rate": 0.0009660421601067666,
|
| 72 |
+
"loss": 2.5374,
|
| 73 |
+
"step": 90
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.1834862385321101,
|
| 77 |
+
"grad_norm": 0.3392362892627716,
|
| 78 |
+
"learning_rate": 0.0009541828235992926,
|
| 79 |
+
"loss": 2.5224,
|
| 80 |
+
"step": 100
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.2018348623853211,
|
| 84 |
+
"grad_norm": 0.23204679787158966,
|
| 85 |
+
"learning_rate": 0.0009406469401200151,
|
| 86 |
+
"loss": 2.5013,
|
| 87 |
+
"step": 110
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.22018348623853212,
|
| 91 |
+
"grad_norm": 0.26075395941734314,
|
| 92 |
+
"learning_rate": 0.0009254844753287492,
|
| 93 |
+
"loss": 2.493,
|
| 94 |
+
"step": 120
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.23853211009174313,
|
| 98 |
+
"grad_norm": 0.19990012049674988,
|
| 99 |
+
"learning_rate": 0.0009087513991633924,
|
| 100 |
+
"loss": 2.4675,
|
| 101 |
+
"step": 130
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.25688073394495414,
|
| 105 |
+
"grad_norm": 0.29057177901268005,
|
| 106 |
+
"learning_rate": 0.0008905094792353917,
|
| 107 |
+
"loss": 2.4526,
|
| 108 |
+
"step": 140
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.27522935779816515,
|
| 112 |
+
"grad_norm": 0.22137577831745148,
|
| 113 |
+
"learning_rate": 0.0008708260528239789,
|
| 114 |
+
"loss": 2.4514,
|
| 115 |
+
"step": 150
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.29357798165137616,
|
| 119 |
+
"grad_norm": 0.20905187726020813,
|
| 120 |
+
"learning_rate": 0.0008497737783108237,
|
| 121 |
+
"loss": 2.4318,
|
| 122 |
+
"step": 160
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.3119266055045872,
|
| 126 |
+
"grad_norm": 0.2661164700984955,
|
| 127 |
+
"learning_rate": 0.0008274303669726426,
|
| 128 |
+
"loss": 2.4199,
|
| 129 |
+
"step": 170
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.3302752293577982,
|
| 133 |
+
"grad_norm": 0.18506155908107758,
|
| 134 |
+
"learning_rate": 0.0008038782961218137,
|
| 135 |
+
"loss": 2.4033,
|
| 136 |
+
"step": 180
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.3486238532110092,
|
| 140 |
+
"grad_norm": 0.20867474377155304,
|
| 141 |
+
"learning_rate": 0.000779204504653894,
|
| 142 |
+
"loss": 2.391,
|
| 143 |
+
"step": 190
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.3669724770642202,
|
| 147 |
+
"grad_norm": 0.24325165152549744,
|
| 148 |
+
"learning_rate": 0.0007535000721258791,
|
| 149 |
+
"loss": 2.3671,
|
| 150 |
+
"step": 200
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.3853211009174312,
|
| 154 |
+
"grad_norm": 0.2229505330324173,
|
| 155 |
+
"learning_rate": 0.00072685988254984,
|
| 156 |
+
"loss": 2.3652,
|
| 157 |
+
"step": 210
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.4036697247706422,
|
| 161 |
+
"grad_norm": 0.26211461424827576,
|
| 162 |
+
"learning_rate": 0.0006993822741429906,
|
| 163 |
+
"loss": 2.3724,
|
| 164 |
+
"step": 220
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.42201834862385323,
|
| 168 |
+
"grad_norm": 0.19598594307899475,
|
| 169 |
+
"learning_rate": 0.0006711686763270818,
|
| 170 |
+
"loss": 2.3613,
|
| 171 |
+
"step": 230
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.44036697247706424,
|
| 175 |
+
"grad_norm": 0.17058943212032318,
|
| 176 |
+
"learning_rate": 0.0006423232353170831,
|
| 177 |
+
"loss": 2.3508,
|
| 178 |
+
"step": 240
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.45871559633027525,
|
| 182 |
+
"grad_norm": 0.21554125845432281,
|
| 183 |
+
"learning_rate": 0.0006129524296812335,
|
| 184 |
+
"loss": 2.3468,
|
| 185 |
+
"step": 250
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.47706422018348627,
|
| 189 |
+
"grad_norm": 0.17847643792629242,
|
| 190 |
+
"learning_rate": 0.0005831646772915651,
|
| 191 |
+
"loss": 2.3537,
|
| 192 |
+
"step": 260
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.4954128440366973,
|
| 196 |
+
"grad_norm": 0.20648759603500366,
|
| 197 |
+
"learning_rate": 0.0005530699351157809,
|
| 198 |
+
"loss": 2.3297,
|
| 199 |
+
"step": 270
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.5137614678899083,
|
| 203 |
+
"grad_norm": 0.2298634946346283,
|
| 204 |
+
"learning_rate": 0.0005227792933277943,
|
| 205 |
+
"loss": 2.3381,
|
| 206 |
+
"step": 280
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.5321100917431193,
|
| 210 |
+
"grad_norm": 0.1709970086812973,
|
| 211 |
+
"learning_rate": 0.0004924045652352092,
|
| 212 |
+
"loss": 2.3211,
|
| 213 |
+
"step": 290
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.5504587155963303,
|
| 217 |
+
"grad_norm": 0.18010194599628448,
|
| 218 |
+
"learning_rate": 0.0004620578745374633,
|
| 219 |
+
"loss": 2.3237,
|
| 220 |
+
"step": 300
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.5688073394495413,
|
| 224 |
+
"grad_norm": 0.22244006395339966,
|
| 225 |
+
"learning_rate": 0.0004318512414382059,
|
| 226 |
+
"loss": 2.332,
|
| 227 |
+
"step": 310
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.5871559633027523,
|
| 231 |
+
"grad_norm": 0.18778839707374573,
|
| 232 |
+
"learning_rate": 0.0004018961691397148,
|
| 233 |
+
"loss": 2.3309,
|
| 234 |
+
"step": 320
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.6055045871559633,
|
| 238 |
+
"grad_norm": 0.16418741643428802,
|
| 239 |
+
"learning_rate": 0.0003723032322457458,
|
| 240 |
+
"loss": 2.3137,
|
| 241 |
+
"step": 330
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.6238532110091743,
|
| 245 |
+
"grad_norm": 0.24950625002384186,
|
| 246 |
+
"learning_rate": 0.0003431816685921625,
|
| 247 |
+
"loss": 2.3209,
|
| 248 |
+
"step": 340
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.6422018348623854,
|
| 252 |
+
"grad_norm": 0.16960828006267548,
|
| 253 |
+
"learning_rate": 0.0003146389760120416,
|
| 254 |
+
"loss": 2.32,
|
| 255 |
+
"step": 350
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.6605504587155964,
|
| 259 |
+
"grad_norm": 0.1736227124929428,
|
| 260 |
+
"learning_rate": 0.00028678051552373485,
|
| 261 |
+
"loss": 2.3218,
|
| 262 |
+
"step": 360
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.6788990825688074,
|
| 266 |
+
"grad_norm": 0.16135239601135254,
|
| 267 |
+
"learning_rate": 0.0002597091224066581,
|
| 268 |
+
"loss": 2.3166,
|
| 269 |
+
"step": 370
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.6972477064220184,
|
| 273 |
+
"grad_norm": 0.1582319736480713,
|
| 274 |
+
"learning_rate": 0.0002335247266004629,
|
| 275 |
+
"loss": 2.3223,
|
| 276 |
+
"step": 380
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.7155963302752294,
|
| 280 |
+
"grad_norm": 0.15470121800899506,
|
| 281 |
+
"learning_rate": 0.000208323983828828,
|
| 282 |
+
"loss": 2.3123,
|
| 283 |
+
"step": 390
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.7339449541284404,
|
| 287 |
+
"grad_norm": 0.16077810525894165,
|
| 288 |
+
"learning_rate": 0.0001841999188095224,
|
| 289 |
+
"loss": 2.3057,
|
| 290 |
+
"step": 400
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.7522935779816514,
|
| 294 |
+
"grad_norm": 0.1593514084815979,
|
| 295 |
+
"learning_rate": 0.00016124158186777676,
|
| 296 |
+
"loss": 2.3081,
|
| 297 |
+
"step": 410
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.7706422018348624,
|
| 301 |
+
"grad_norm": 0.1685478240251541,
|
| 302 |
+
"learning_rate": 0.0001395337202205211,
|
| 303 |
+
"loss": 2.309,
|
| 304 |
+
"step": 420
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 0.7889908256880734,
|
| 308 |
+
"grad_norm": 0.15101832151412964,
|
| 309 |
+
"learning_rate": 0.00011915646514489292,
|
| 310 |
+
"loss": 2.305,
|
| 311 |
+
"step": 430
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.8073394495412844,
|
| 315 |
+
"grad_norm": 0.17798583209514618,
|
| 316 |
+
"learning_rate": 0.00010018503618578818,
|
| 317 |
+
"loss": 2.3113,
|
| 318 |
+
"step": 440
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.8256880733944955,
|
| 322 |
+
"grad_norm": 0.16483835875988007,
|
| 323 |
+
"learning_rate": 8.268946349432583e-05,
|
| 324 |
+
"loss": 2.3105,
|
| 325 |
+
"step": 450
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.8440366972477065,
|
| 329 |
+
"grad_norm": 0.14816823601722717,
|
| 330 |
+
"learning_rate": 6.673432932217117e-05,
|
| 331 |
+
"loss": 2.2989,
|
| 332 |
+
"step": 460
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.8623853211009175,
|
| 336 |
+
"grad_norm": 0.16976268589496613,
|
| 337 |
+
"learning_rate": 5.2378529625954684e-05,
|
| 338 |
+
"loss": 2.2939,
|
| 339 |
+
"step": 470
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.8807339449541285,
|
| 343 |
+
"grad_norm": 0.14344191551208496,
|
| 344 |
+
"learning_rate": 3.9675056661785556e-05,
|
| 345 |
+
"loss": 2.2967,
|
| 346 |
+
"step": 480
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.8990825688073395,
|
| 350 |
+
"grad_norm": 0.14687775075435638,
|
| 351 |
+
"learning_rate": 2.8670803372379538e-05,
|
| 352 |
+
"loss": 2.3033,
|
| 353 |
+
"step": 490
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.9174311926605505,
|
| 357 |
+
"grad_norm": 0.14416512846946716,
|
| 358 |
+
"learning_rate": 1.9406390288876586e-05,
|
| 359 |
+
"loss": 2.3054,
|
| 360 |
+
"step": 500
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.9357798165137615,
|
| 364 |
+
"grad_norm": 0.14302416145801544,
|
| 365 |
+
"learning_rate": 1.191601558631461e-05,
|
| 366 |
+
"loss": 2.2969,
|
| 367 |
+
"step": 510
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.9541284403669725,
|
| 371 |
+
"grad_norm": 0.14250217378139496,
|
| 372 |
+
"learning_rate": 6.2273288462559305e-06,
|
| 373 |
+
"loss": 2.2932,
|
| 374 |
+
"step": 520
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.9724770642201835,
|
| 378 |
+
"grad_norm": 0.14332719147205353,
|
| 379 |
+
"learning_rate": 2.361328992552314e-06,
|
| 380 |
+
"loss": 2.2968,
|
| 381 |
+
"step": 530
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.9908256880733946,
|
| 385 |
+
"grad_norm": 0.14881865680217743,
|
| 386 |
+
"learning_rate": 3.322867770044202e-07,
|
| 387 |
+
"loss": 2.2948,
|
| 388 |
+
"step": 540
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 1.0,
|
| 392 |
+
"step": 545,
|
| 393 |
+
"total_flos": 3.6873030393554534e+17,
|
| 394 |
+
"train_loss": 2.422908702465372,
|
| 395 |
+
"train_runtime": 1370.887,
|
| 396 |
+
"train_samples_per_second": 407.129,
|
| 397 |
+
"train_steps_per_second": 0.398
|
| 398 |
+
}
|
| 399 |
+
],
|
| 400 |
+
"logging_steps": 10,
|
| 401 |
+
"max_steps": 545,
|
| 402 |
+
"num_input_tokens_seen": 36058112,
|
| 403 |
+
"num_train_epochs": 1,
|
| 404 |
+
"save_steps": 500,
|
| 405 |
+
"stateful_callbacks": {
|
| 406 |
+
"TrainerControl": {
|
| 407 |
+
"args": {
|
| 408 |
+
"should_epoch_stop": false,
|
| 409 |
+
"should_evaluate": false,
|
| 410 |
+
"should_log": false,
|
| 411 |
+
"should_save": false,
|
| 412 |
+
"should_training_stop": false
|
| 413 |
+
},
|
| 414 |
+
"attributes": {}
|
| 415 |
+
}
|
| 416 |
+
},
|
| 417 |
+
"total_flos": 3.6873030393554534e+17,
|
| 418 |
+
"train_batch_size": 128,
|
| 419 |
+
"trial_name": null,
|
| 420 |
+
"trial_params": null
|
| 421 |
+
}
|
continuous/smollm2/mc1_g14_224_2.5b/finetune/chat_template.jinja
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{% for message in messages %}{% if loop.first and messages[0]['role'] != 'system' %}{{ '<|im_start|>system
|
| 2 |
+
You are a helpful AI assistant named SmolLM, trained by Hugging Face<|im_end|>
|
| 3 |
+
' }}{% endif %}{{'<|im_start|>' + message['role'] + '
|
| 4 |
+
' + message['content'] + '<|im_end|>' + '
|
| 5 |
+
'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant
|
| 6 |
+
' }}{% endif %}
|
continuous/smollm2/mc1_g14_224_2.5b/finetune/config.json
ADDED
|
@@ -0,0 +1,70 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_faster_video": false,
|
| 3 |
+
"add_time_instruction": false,
|
| 4 |
+
"architectures": [
|
| 5 |
+
"LlavaLlamaForCausalLM"
|
| 6 |
+
],
|
| 7 |
+
"attention_bias": false,
|
| 8 |
+
"attention_dropout": 0.0,
|
| 9 |
+
"bos_token_id": 1,
|
| 10 |
+
"dtype": "bfloat16",
|
| 11 |
+
"eos_token_id": 2,
|
| 12 |
+
"faster_token_stride": 10,
|
| 13 |
+
"force_sample": false,
|
| 14 |
+
"head_dim": 64,
|
| 15 |
+
"hidden_act": "silu",
|
| 16 |
+
"hidden_size": 2048,
|
| 17 |
+
"image_aspect_ratio": "square",
|
| 18 |
+
"image_crop_resolution": null,
|
| 19 |
+
"image_grid_pinpoints": null,
|
| 20 |
+
"image_split_resolution": null,
|
| 21 |
+
"initializer_range": 0.02,
|
| 22 |
+
"intermediate_size": 8192,
|
| 23 |
+
"max_position_embeddings": 8192,
|
| 24 |
+
"mlp_bias": false,
|
| 25 |
+
"mm_hidden_size": 1664,
|
| 26 |
+
"mm_newline_position": "grid",
|
| 27 |
+
"mm_patch_merge_type": "flat",
|
| 28 |
+
"mm_projector_lr": 0.0001,
|
| 29 |
+
"mm_projector_type": "mlp2x_gelu",
|
| 30 |
+
"mm_resampler_type": null,
|
| 31 |
+
"mm_spatial_pool_mode": "bilinear",
|
| 32 |
+
"mm_spatial_pool_stride": null,
|
| 33 |
+
"mm_tunable_parts": "mm_mlp_adapter,mm_language_model",
|
| 34 |
+
"mm_use_im_patch_token": false,
|
| 35 |
+
"mm_use_im_start_end": false,
|
| 36 |
+
"mm_vision_select_feature": "patch",
|
| 37 |
+
"mm_vision_select_layer": -2,
|
| 38 |
+
"mm_vision_tower": "open_clip_hub:ViT-bigG-14-quickgelu",
|
| 39 |
+
"mm_vision_tower_lr": null,
|
| 40 |
+
"model_type": "llama",
|
| 41 |
+
"num_attention_heads": 32,
|
| 42 |
+
"num_hidden_layers": 24,
|
| 43 |
+
"num_key_value_heads": 32,
|
| 44 |
+
"pad_token_id": 0,
|
| 45 |
+
"pos_skipping_range": 4096,
|
| 46 |
+
"pretraining_tp": 1,
|
| 47 |
+
"rms_norm_eps": 1e-05,
|
| 48 |
+
"rope_scaling": null,
|
| 49 |
+
"rope_theta": 130000,
|
| 50 |
+
"tie_word_embeddings": true,
|
| 51 |
+
"tokenizer_model_max_length": 8192,
|
| 52 |
+
"tokenizer_padding_side": "right",
|
| 53 |
+
"transformers.js_config": {
|
| 54 |
+
"dtype": "q4",
|
| 55 |
+
"kv_cache_dtype": {
|
| 56 |
+
"fp16": "float16",
|
| 57 |
+
"q4f16": "float16"
|
| 58 |
+
},
|
| 59 |
+
"use_external_data_format": {
|
| 60 |
+
"model.onnx": true,
|
| 61 |
+
"model_fp16.onnx": true
|
| 62 |
+
}
|
| 63 |
+
},
|
| 64 |
+
"transformers_version": "4.57.6",
|
| 65 |
+
"use_cache": true,
|
| 66 |
+
"use_mm_proj": true,
|
| 67 |
+
"use_pos_skipping": false,
|
| 68 |
+
"vision_tower_pretrained": "/cache/ckpt/download/tokenizer/continuous/mc1/mc1_g14_224_2.5b.pt",
|
| 69 |
+
"vocab_size": 49152
|
| 70 |
+
}
|
continuous/smollm2/mc1_g14_224_2.5b/finetune/generation_config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 1,
|
| 4 |
+
"eos_token_id": [
|
| 5 |
+
2
|
| 6 |
+
],
|
| 7 |
+
"pad_token_id": 0,
|
| 8 |
+
"transformers_version": "4.57.6"
|
| 9 |
+
}
|
continuous/smollm2/mc1_g14_224_2.5b/finetune/merges.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
continuous/smollm2/mc1_g14_224_2.5b/finetune/model-00001-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:28339ca6b496b28d835eecd2016491a8b45fb8ce060c941edb85b8adc6020f5c
|
| 3 |
+
size 4986543952
|