Commit
·
3cf1937
1
Parent(s):
712ca44
Refined (v5.3) update
Browse filesThis view is limited to 50 files because it contains too many changes.
See raw diff
- README.md +6 -4
- checkpoints/checkpoint-57000/trainer_state.json +0 -840
- checkpoints/checkpoint-57500/trainer_state.json +0 -847
- checkpoints/checkpoint-57726/trainer_state.json +0 -847
- checkpoints/{checkpoint-57000 → checkpoint-61500}/config.json +0 -0
- checkpoints/{checkpoint-57000 → checkpoint-61500}/generation_config.json +0 -0
- checkpoints/{checkpoint-57000 → checkpoint-61500}/model.safetensors +1 -1
- checkpoints/{checkpoint-57500 → checkpoint-61500}/optimizer.pt +1 -1
- checkpoints/{checkpoint-57726 → checkpoint-61500}/rng_state.pth +1 -1
- checkpoints/{checkpoint-57726 → checkpoint-61500}/scaler.pt +1 -1
- checkpoints/{checkpoint-57000 → checkpoint-61500}/scheduler.pt +1 -1
- checkpoints/{checkpoint-57000 → checkpoint-61500}/special_tokens_map.json +0 -0
- checkpoints/{checkpoint-57500 → checkpoint-61500}/spiece.model +2 -2
- checkpoints/{checkpoint-57000 → checkpoint-61500}/tokenizer.json +0 -0
- checkpoints/{checkpoint-57000 → checkpoint-61500}/tokenizer_config.json +0 -0
- checkpoints/checkpoint-61500/trainer_state.json +903 -0
- checkpoints/{checkpoint-57000 → checkpoint-61500}/training_args.bin +0 -0
- checkpoints/{checkpoint-57500 → checkpoint-62000}/config.json +0 -0
- checkpoints/{checkpoint-57500 → checkpoint-62000}/generation_config.json +0 -0
- checkpoints/{checkpoint-57500 → checkpoint-62000}/model.safetensors +1 -1
- checkpoints/{checkpoint-57000 → checkpoint-62000}/optimizer.pt +1 -1
- checkpoints/{checkpoint-57500 → checkpoint-62000}/rng_state.pth +1 -1
- checkpoints/{checkpoint-57500 → checkpoint-62000}/scaler.pt +1 -1
- checkpoints/{checkpoint-57726 → checkpoint-62000}/scheduler.pt +1 -1
- checkpoints/{checkpoint-57500 → checkpoint-62000}/special_tokens_map.json +0 -0
- checkpoints/{checkpoint-57000 → checkpoint-62000}/spiece.model +2 -2
- checkpoints/{checkpoint-57500 → checkpoint-62000}/tokenizer.json +0 -0
- checkpoints/{checkpoint-57500 → checkpoint-62000}/tokenizer_config.json +0 -0
- checkpoints/checkpoint-62000/trainer_state.json +910 -0
- checkpoints/{checkpoint-57500 → checkpoint-62000}/training_args.bin +0 -0
- checkpoints/{checkpoint-57726 → checkpoint-62164}/config.json +0 -0
- checkpoints/{checkpoint-57726 → checkpoint-62164}/generation_config.json +0 -0
- checkpoints/{checkpoint-57726 → checkpoint-62164}/model.safetensors +1 -1
- checkpoints/{checkpoint-57726 → checkpoint-62164}/optimizer.pt +1 -1
- checkpoints/{checkpoint-57000 → checkpoint-62164}/rng_state.pth +1 -1
- checkpoints/{checkpoint-57000 → checkpoint-62164}/scaler.pt +1 -1
- checkpoints/{checkpoint-57500 → checkpoint-62164}/scheduler.pt +1 -1
- checkpoints/{checkpoint-57726 → checkpoint-62164}/special_tokens_map.json +0 -0
- checkpoints/{checkpoint-57726 → checkpoint-62164}/spiece.model +2 -2
- checkpoints/{checkpoint-57726 → checkpoint-62164}/tokenizer.json +0 -0
- checkpoints/{checkpoint-57726 → checkpoint-62164}/tokenizer_config.json +0 -0
- checkpoints/checkpoint-62164/trainer_state.json +910 -0
- checkpoints/{checkpoint-57726 → checkpoint-62164}/training_args.bin +0 -0
- model.safetensors +1 -1
- spiece.model +2 -2
- src/data/generate_cyr_lat_pairs.py +1 -1
- src/tokeniser/dalat5_sp.model +2 -2
- src/tokeniser/dalat5_sp.vocab +0 -0
- src/tokeniser/spiece.model +2 -2
- src/train_tokeniser.py +1 -0
README.md
CHANGED
|
@@ -24,10 +24,10 @@ model-index:
|
|
| 24 |
metrics:
|
| 25 |
- name: Training Loss
|
| 26 |
type: loss
|
| 27 |
-
value: 0.
|
| 28 |
- name: Evaluation Loss
|
| 29 |
type: loss
|
| 30 |
-
value: 0.
|
| 31 |
---
|
| 32 |
# DalaT5 - T5 Fine-Tuned on Cyrillic-to-Latin Kazakh 🇰🇿
|
| 33 |
|
|
@@ -139,9 +139,11 @@ KazParC деректер жинағын жүктеп алу үшін сізге
|
|
| 139 |
|
| 140 |
* **DalaT5 v5**: 25 сәуірде дәл реттелген және сол күні қолжетімді болды. Қазақ кириллица және латын графикасын жақсырақ өңдеу үшін өзінің жеке токенизаторы бар ~1,9 миллион жазба (v4 сияқты) пайдаланылды / Fine-tuned on April 25 and made available on the same day. Used ~1.9 million records (like v4) with its own tokeniser to better handle the Kazakh Cyrillic and Latin scripts
|
| 141 |
|
| 142 |
-
* **DalaT5 v5.1**: 25 сәуірде (v5 нұсқасынан кейін бірден) дәл реттелген және сол күні қолжетімді болды. Жақсырақ жалпылауды қамтамасыз ету үшін
|
| 143 |
|
| 144 |
-
* **DalaT5 v5.2**:
|
|
|
|
|
|
|
| 145 |
|
| 146 |
---
|
| 147 |
|
|
|
|
| 24 |
metrics:
|
| 25 |
- name: Training Loss
|
| 26 |
type: loss
|
| 27 |
+
value: 0.6684
|
| 28 |
- name: Evaluation Loss
|
| 29 |
type: loss
|
| 30 |
+
value: 0.0886
|
| 31 |
---
|
| 32 |
# DalaT5 - T5 Fine-Tuned on Cyrillic-to-Latin Kazakh 🇰🇿
|
| 33 |
|
|
|
|
| 139 |
|
| 140 |
* **DalaT5 v5**: 25 сәуірде дәл реттелген және сол күні қолжетімді болды. Қазақ кириллица және латын графикасын жақсырақ өңдеу үшін өзінің жеке токенизаторы бар ~1,9 миллион жазба (v4 сияқты) пайдаланылды / Fine-tuned on April 25 and made available on the same day. Used ~1.9 million records (like v4) with its own tokeniser to better handle the Kazakh Cyrillic and Latin scripts
|
| 141 |
|
| 142 |
+
* **DalaT5 v5.1**: 25 сәуірде (v5 нұсқасынан кейін бірден) дәл реттелген және сол күні қолжетімді болды. Жақсырақ жалпылауды қамтамасыз ету үшін жаттығу үшін ~2,2 миллион жазба және токенизатор үшін 1 миллион жазба пайдаланылды. v5-пен салыстырғанда галлюцинациялар күрт төмендеп, семантикалық түсіну одан әрі жақсарды / Fine-tuned on April 25 (immediately after v5) and made available on the same day. Used ~2.2 million records for training and 1 million records for the tokeniser to ensure better generalisation. Hallucinations decreased drastically when compared to v5, and semantic understanding was further enhanced
|
| 143 |
|
| 144 |
+
* **DalaT5 v5.2**: 25 сәуірде (v5 нұсқасынан кейін бірден) дәл реттелген және сол күні қолжетімді болды. Жақсырақ жалпылауды қамтамасыз ету үшін жаттығу үшін ~2,2 миллион жазба және токенизатор үшін 1 миллион жазба пайдаланылды. v5-пен салыстырғанда галлюцинациялар күрт төмендеп, семантикалық түсіну одан әрі жақсарды / Fine-tuned on April 26 and made available on the same day. Used the same tokeniser structure overall, but leveraged ~2.4 million records for training and evaluation. The evaluation loss was also made available with this version. Overall, hallucinations that made it through v5.1 were almost completely eliminated
|
| 145 |
+
|
| 146 |
+
* **DalaT5 v5.3**: 28 сәуірде дәл реттелген және сол күні қолжетімді болды. Жалпы, токенизаторға арналған ұлғайтылған максималды сөйлем өлшемінен басқа (4192 орнына 8384), ол v5.2 сияқты құрылымды пайдаланды. Бұл нұсқа одан да жақсы жалпылауды қамтамасыз ету үшін оқыту және бағалау үшін ~2,6 миллион жазбаны пайдаланды. Галлюцинацияның одан әрі азаюы байқалды және модель қазір қазақ морфологиясын өңдеуге шебер болған сияқты. / Fine-tuned on April 28 and made available on the same day. Overall, other than an increased maximum sentence size for the tokeniser (8384 instead of 4192), it used the same structure as v5.2. This version leveraged ~2.6 million records for training and evaluation to ensure even better generalisation. Further reduction of hallucinations was observed, and the model now seems to have become adept at handling Kazakh morphology
|
| 147 |
|
| 148 |
---
|
| 149 |
|
checkpoints/checkpoint-57000/trainer_state.json
DELETED
|
@@ -1,840 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"best_global_step": null,
|
| 3 |
-
"best_metric": null,
|
| 4 |
-
"best_model_checkpoint": null,
|
| 5 |
-
"epoch": 1.9748466895333125,
|
| 6 |
-
"eval_steps": 500,
|
| 7 |
-
"global_step": 57000,
|
| 8 |
-
"is_hyper_param_search": false,
|
| 9 |
-
"is_local_process_zero": true,
|
| 10 |
-
"is_world_process_zero": true,
|
| 11 |
-
"log_history": [
|
| 12 |
-
{
|
| 13 |
-
"epoch": 0.01732321657485362,
|
| 14 |
-
"grad_norm": 0.5299897193908691,
|
| 15 |
-
"learning_rate": 4.956778574645741e-05,
|
| 16 |
-
"loss": 2.1733,
|
| 17 |
-
"step": 500
|
| 18 |
-
},
|
| 19 |
-
{
|
| 20 |
-
"epoch": 0.03464643314970724,
|
| 21 |
-
"grad_norm": 0.4744114279747009,
|
| 22 |
-
"learning_rate": 4.913470533208606e-05,
|
| 23 |
-
"loss": 1.936,
|
| 24 |
-
"step": 1000
|
| 25 |
-
},
|
| 26 |
-
{
|
| 27 |
-
"epoch": 0.051969649724560855,
|
| 28 |
-
"grad_norm": 0.5358501672744751,
|
| 29 |
-
"learning_rate": 4.870162491771472e-05,
|
| 30 |
-
"loss": 1.8605,
|
| 31 |
-
"step": 1500
|
| 32 |
-
},
|
| 33 |
-
{
|
| 34 |
-
"epoch": 0.06929286629941447,
|
| 35 |
-
"grad_norm": 0.5026484131813049,
|
| 36 |
-
"learning_rate": 4.826854450334338e-05,
|
| 37 |
-
"loss": 1.793,
|
| 38 |
-
"step": 2000
|
| 39 |
-
},
|
| 40 |
-
{
|
| 41 |
-
"epoch": 0.0866160828742681,
|
| 42 |
-
"grad_norm": 0.5730611681938171,
|
| 43 |
-
"learning_rate": 4.7835464088972044e-05,
|
| 44 |
-
"loss": 1.7309,
|
| 45 |
-
"step": 2500
|
| 46 |
-
},
|
| 47 |
-
{
|
| 48 |
-
"epoch": 0.10393929944912171,
|
| 49 |
-
"grad_norm": 0.47230127453804016,
|
| 50 |
-
"learning_rate": 4.74023836746007e-05,
|
| 51 |
-
"loss": 1.6938,
|
| 52 |
-
"step": 3000
|
| 53 |
-
},
|
| 54 |
-
{
|
| 55 |
-
"epoch": 0.12126251602397534,
|
| 56 |
-
"grad_norm": 0.6244428157806396,
|
| 57 |
-
"learning_rate": 4.6969303260229364e-05,
|
| 58 |
-
"loss": 1.6587,
|
| 59 |
-
"step": 3500
|
| 60 |
-
},
|
| 61 |
-
{
|
| 62 |
-
"epoch": 0.13858573259882895,
|
| 63 |
-
"grad_norm": 0.5069366693496704,
|
| 64 |
-
"learning_rate": 4.653622284585802e-05,
|
| 65 |
-
"loss": 1.6241,
|
| 66 |
-
"step": 4000
|
| 67 |
-
},
|
| 68 |
-
{
|
| 69 |
-
"epoch": 0.15590894917368256,
|
| 70 |
-
"grad_norm": 0.567051887512207,
|
| 71 |
-
"learning_rate": 4.610314243148668e-05,
|
| 72 |
-
"loss": 1.5789,
|
| 73 |
-
"step": 4500
|
| 74 |
-
},
|
| 75 |
-
{
|
| 76 |
-
"epoch": 0.1732321657485362,
|
| 77 |
-
"grad_norm": 0.5564538836479187,
|
| 78 |
-
"learning_rate": 4.567006201711534e-05,
|
| 79 |
-
"loss": 1.5619,
|
| 80 |
-
"step": 5000
|
| 81 |
-
},
|
| 82 |
-
{
|
| 83 |
-
"epoch": 0.1905553823233898,
|
| 84 |
-
"grad_norm": 0.7243916988372803,
|
| 85 |
-
"learning_rate": 4.5236981602744e-05,
|
| 86 |
-
"loss": 1.5358,
|
| 87 |
-
"step": 5500
|
| 88 |
-
},
|
| 89 |
-
{
|
| 90 |
-
"epoch": 0.20787859889824342,
|
| 91 |
-
"grad_norm": 0.6585371494293213,
|
| 92 |
-
"learning_rate": 4.480390118837266e-05,
|
| 93 |
-
"loss": 1.5396,
|
| 94 |
-
"step": 6000
|
| 95 |
-
},
|
| 96 |
-
{
|
| 97 |
-
"epoch": 0.22520181547309703,
|
| 98 |
-
"grad_norm": 0.5986710786819458,
|
| 99 |
-
"learning_rate": 4.437082077400132e-05,
|
| 100 |
-
"loss": 1.4991,
|
| 101 |
-
"step": 6500
|
| 102 |
-
},
|
| 103 |
-
{
|
| 104 |
-
"epoch": 0.24252503204795067,
|
| 105 |
-
"grad_norm": 0.6701385974884033,
|
| 106 |
-
"learning_rate": 4.393774035962998e-05,
|
| 107 |
-
"loss": 1.4823,
|
| 108 |
-
"step": 7000
|
| 109 |
-
},
|
| 110 |
-
{
|
| 111 |
-
"epoch": 0.25984824862280426,
|
| 112 |
-
"grad_norm": 0.6745362877845764,
|
| 113 |
-
"learning_rate": 4.350465994525863e-05,
|
| 114 |
-
"loss": 1.4715,
|
| 115 |
-
"step": 7500
|
| 116 |
-
},
|
| 117 |
-
{
|
| 118 |
-
"epoch": 0.2771714651976579,
|
| 119 |
-
"grad_norm": 0.6930545568466187,
|
| 120 |
-
"learning_rate": 4.30715795308873e-05,
|
| 121 |
-
"loss": 1.4354,
|
| 122 |
-
"step": 8000
|
| 123 |
-
},
|
| 124 |
-
{
|
| 125 |
-
"epoch": 0.29449468177251154,
|
| 126 |
-
"grad_norm": 0.6593695878982544,
|
| 127 |
-
"learning_rate": 4.2638499116515954e-05,
|
| 128 |
-
"loss": 1.432,
|
| 129 |
-
"step": 8500
|
| 130 |
-
},
|
| 131 |
-
{
|
| 132 |
-
"epoch": 0.3118178983473651,
|
| 133 |
-
"grad_norm": 0.6918864846229553,
|
| 134 |
-
"learning_rate": 4.220541870214462e-05,
|
| 135 |
-
"loss": 1.4209,
|
| 136 |
-
"step": 9000
|
| 137 |
-
},
|
| 138 |
-
{
|
| 139 |
-
"epoch": 0.32914111492221876,
|
| 140 |
-
"grad_norm": 0.7154921889305115,
|
| 141 |
-
"learning_rate": 4.1772338287773275e-05,
|
| 142 |
-
"loss": 1.391,
|
| 143 |
-
"step": 9500
|
| 144 |
-
},
|
| 145 |
-
{
|
| 146 |
-
"epoch": 0.3464643314970724,
|
| 147 |
-
"grad_norm": 0.8634065389633179,
|
| 148 |
-
"learning_rate": 4.133925787340194e-05,
|
| 149 |
-
"loss": 1.3843,
|
| 150 |
-
"step": 10000
|
| 151 |
-
},
|
| 152 |
-
{
|
| 153 |
-
"epoch": 0.363787548071926,
|
| 154 |
-
"grad_norm": 0.8820887804031372,
|
| 155 |
-
"learning_rate": 4.0906177459030595e-05,
|
| 156 |
-
"loss": 1.3612,
|
| 157 |
-
"step": 10500
|
| 158 |
-
},
|
| 159 |
-
{
|
| 160 |
-
"epoch": 0.3811107646467796,
|
| 161 |
-
"grad_norm": 0.9643932580947876,
|
| 162 |
-
"learning_rate": 4.047309704465925e-05,
|
| 163 |
-
"loss": 1.3326,
|
| 164 |
-
"step": 11000
|
| 165 |
-
},
|
| 166 |
-
{
|
| 167 |
-
"epoch": 0.39843398122163326,
|
| 168 |
-
"grad_norm": 0.9170373678207397,
|
| 169 |
-
"learning_rate": 4.0040016630287916e-05,
|
| 170 |
-
"loss": 1.3158,
|
| 171 |
-
"step": 11500
|
| 172 |
-
},
|
| 173 |
-
{
|
| 174 |
-
"epoch": 0.41575719779648684,
|
| 175 |
-
"grad_norm": 0.8911833167076111,
|
| 176 |
-
"learning_rate": 3.960693621591657e-05,
|
| 177 |
-
"loss": 1.2982,
|
| 178 |
-
"step": 12000
|
| 179 |
-
},
|
| 180 |
-
{
|
| 181 |
-
"epoch": 0.4330804143713405,
|
| 182 |
-
"grad_norm": 1.3023812770843506,
|
| 183 |
-
"learning_rate": 3.917472196237398e-05,
|
| 184 |
-
"loss": 1.2497,
|
| 185 |
-
"step": 12500
|
| 186 |
-
},
|
| 187 |
-
{
|
| 188 |
-
"epoch": 0.45040363094619407,
|
| 189 |
-
"grad_norm": 1.3367282152175903,
|
| 190 |
-
"learning_rate": 3.874164154800263e-05,
|
| 191 |
-
"loss": 1.2436,
|
| 192 |
-
"step": 13000
|
| 193 |
-
},
|
| 194 |
-
{
|
| 195 |
-
"epoch": 0.4677268475210477,
|
| 196 |
-
"grad_norm": 1.070709228515625,
|
| 197 |
-
"learning_rate": 3.8308561133631293e-05,
|
| 198 |
-
"loss": 1.2068,
|
| 199 |
-
"step": 13500
|
| 200 |
-
},
|
| 201 |
-
{
|
| 202 |
-
"epoch": 0.48505006409590135,
|
| 203 |
-
"grad_norm": 1.2574782371520996,
|
| 204 |
-
"learning_rate": 3.787548071925995e-05,
|
| 205 |
-
"loss": 1.1867,
|
| 206 |
-
"step": 14000
|
| 207 |
-
},
|
| 208 |
-
{
|
| 209 |
-
"epoch": 0.5023732806707549,
|
| 210 |
-
"grad_norm": 1.4889111518859863,
|
| 211 |
-
"learning_rate": 3.744326646571736e-05,
|
| 212 |
-
"loss": 1.1908,
|
| 213 |
-
"step": 14500
|
| 214 |
-
},
|
| 215 |
-
{
|
| 216 |
-
"epoch": 0.5196964972456085,
|
| 217 |
-
"grad_norm": 1.2556302547454834,
|
| 218 |
-
"learning_rate": 3.7010186051346014e-05,
|
| 219 |
-
"loss": 1.1476,
|
| 220 |
-
"step": 15000
|
| 221 |
-
},
|
| 222 |
-
{
|
| 223 |
-
"epoch": 0.5370197138204622,
|
| 224 |
-
"grad_norm": 1.2582406997680664,
|
| 225 |
-
"learning_rate": 3.6577971797803415e-05,
|
| 226 |
-
"loss": 1.1268,
|
| 227 |
-
"step": 15500
|
| 228 |
-
},
|
| 229 |
-
{
|
| 230 |
-
"epoch": 0.5543429303953158,
|
| 231 |
-
"grad_norm": 1.4035013914108276,
|
| 232 |
-
"learning_rate": 3.614575754426082e-05,
|
| 233 |
-
"loss": 1.119,
|
| 234 |
-
"step": 16000
|
| 235 |
-
},
|
| 236 |
-
{
|
| 237 |
-
"epoch": 0.5716661469701694,
|
| 238 |
-
"grad_norm": 1.7117546796798706,
|
| 239 |
-
"learning_rate": 3.571267712988948e-05,
|
| 240 |
-
"loss": 1.106,
|
| 241 |
-
"step": 16500
|
| 242 |
-
},
|
| 243 |
-
{
|
| 244 |
-
"epoch": 0.5889893635450231,
|
| 245 |
-
"grad_norm": 1.503953218460083,
|
| 246 |
-
"learning_rate": 3.5279596715518135e-05,
|
| 247 |
-
"loss": 1.08,
|
| 248 |
-
"step": 17000
|
| 249 |
-
},
|
| 250 |
-
{
|
| 251 |
-
"epoch": 0.6063125801198767,
|
| 252 |
-
"grad_norm": 1.6753602027893066,
|
| 253 |
-
"learning_rate": 3.48465163011468e-05,
|
| 254 |
-
"loss": 1.055,
|
| 255 |
-
"step": 17500
|
| 256 |
-
},
|
| 257 |
-
{
|
| 258 |
-
"epoch": 0.6236357966947302,
|
| 259 |
-
"grad_norm": 1.5606533288955688,
|
| 260 |
-
"learning_rate": 3.4413435886775456e-05,
|
| 261 |
-
"loss": 1.043,
|
| 262 |
-
"step": 18000
|
| 263 |
-
},
|
| 264 |
-
{
|
| 265 |
-
"epoch": 0.6409590132695839,
|
| 266 |
-
"grad_norm": 2.1966652870178223,
|
| 267 |
-
"learning_rate": 3.398122163323286e-05,
|
| 268 |
-
"loss": 1.0294,
|
| 269 |
-
"step": 18500
|
| 270 |
-
},
|
| 271 |
-
{
|
| 272 |
-
"epoch": 0.6582822298444375,
|
| 273 |
-
"grad_norm": 2.0083296298980713,
|
| 274 |
-
"learning_rate": 3.354814121886152e-05,
|
| 275 |
-
"loss": 1.009,
|
| 276 |
-
"step": 19000
|
| 277 |
-
},
|
| 278 |
-
{
|
| 279 |
-
"epoch": 0.6756054464192911,
|
| 280 |
-
"grad_norm": 1.61182701587677,
|
| 281 |
-
"learning_rate": 3.311506080449018e-05,
|
| 282 |
-
"loss": 0.978,
|
| 283 |
-
"step": 19500
|
| 284 |
-
},
|
| 285 |
-
{
|
| 286 |
-
"epoch": 0.6929286629941448,
|
| 287 |
-
"grad_norm": 1.698030710220337,
|
| 288 |
-
"learning_rate": 3.268198039011884e-05,
|
| 289 |
-
"loss": 0.9798,
|
| 290 |
-
"step": 20000
|
| 291 |
-
},
|
| 292 |
-
{
|
| 293 |
-
"epoch": 0.7102518795689984,
|
| 294 |
-
"grad_norm": 1.810142993927002,
|
| 295 |
-
"learning_rate": 3.224976613657624e-05,
|
| 296 |
-
"loss": 0.9558,
|
| 297 |
-
"step": 20500
|
| 298 |
-
},
|
| 299 |
-
{
|
| 300 |
-
"epoch": 0.727575096143852,
|
| 301 |
-
"grad_norm": 4.462728023529053,
|
| 302 |
-
"learning_rate": 3.1816685722204904e-05,
|
| 303 |
-
"loss": 0.9337,
|
| 304 |
-
"step": 21000
|
| 305 |
-
},
|
| 306 |
-
{
|
| 307 |
-
"epoch": 0.7448983127187057,
|
| 308 |
-
"grad_norm": 3.5566582679748535,
|
| 309 |
-
"learning_rate": 3.1384471468662305e-05,
|
| 310 |
-
"loss": 0.9115,
|
| 311 |
-
"step": 21500
|
| 312 |
-
},
|
| 313 |
-
{
|
| 314 |
-
"epoch": 0.7622215292935592,
|
| 315 |
-
"grad_norm": 3.2824957370758057,
|
| 316 |
-
"learning_rate": 3.095139105429096e-05,
|
| 317 |
-
"loss": 0.8992,
|
| 318 |
-
"step": 22000
|
| 319 |
-
},
|
| 320 |
-
{
|
| 321 |
-
"epoch": 0.7795447458684128,
|
| 322 |
-
"grad_norm": 3.1051690578460693,
|
| 323 |
-
"learning_rate": 3.0518310639919625e-05,
|
| 324 |
-
"loss": 0.8945,
|
| 325 |
-
"step": 22500
|
| 326 |
-
},
|
| 327 |
-
{
|
| 328 |
-
"epoch": 0.7968679624432665,
|
| 329 |
-
"grad_norm": 5.406064510345459,
|
| 330 |
-
"learning_rate": 3.0085230225548282e-05,
|
| 331 |
-
"loss": 0.8544,
|
| 332 |
-
"step": 23000
|
| 333 |
-
},
|
| 334 |
-
{
|
| 335 |
-
"epoch": 0.8141911790181201,
|
| 336 |
-
"grad_norm": 3.3169949054718018,
|
| 337 |
-
"learning_rate": 2.9653015972005682e-05,
|
| 338 |
-
"loss": 0.8494,
|
| 339 |
-
"step": 23500
|
| 340 |
-
},
|
| 341 |
-
{
|
| 342 |
-
"epoch": 0.8315143955929737,
|
| 343 |
-
"grad_norm": 2.4768149852752686,
|
| 344 |
-
"learning_rate": 2.9219935557634343e-05,
|
| 345 |
-
"loss": 0.8145,
|
| 346 |
-
"step": 24000
|
| 347 |
-
},
|
| 348 |
-
{
|
| 349 |
-
"epoch": 0.8488376121678273,
|
| 350 |
-
"grad_norm": 2.6160104274749756,
|
| 351 |
-
"learning_rate": 2.8786855143263003e-05,
|
| 352 |
-
"loss": 0.7995,
|
| 353 |
-
"step": 24500
|
| 354 |
-
},
|
| 355 |
-
{
|
| 356 |
-
"epoch": 0.866160828742681,
|
| 357 |
-
"grad_norm": 2.7828731536865234,
|
| 358 |
-
"learning_rate": 2.8353774728891663e-05,
|
| 359 |
-
"loss": 0.7705,
|
| 360 |
-
"step": 25000
|
| 361 |
-
},
|
| 362 |
-
{
|
| 363 |
-
"epoch": 0.8834840453175345,
|
| 364 |
-
"grad_norm": 3.508859157562256,
|
| 365 |
-
"learning_rate": 2.7921560475349063e-05,
|
| 366 |
-
"loss": 0.7565,
|
| 367 |
-
"step": 25500
|
| 368 |
-
},
|
| 369 |
-
{
|
| 370 |
-
"epoch": 0.9008072618923881,
|
| 371 |
-
"grad_norm": 2.672107219696045,
|
| 372 |
-
"learning_rate": 2.7488480060977724e-05,
|
| 373 |
-
"loss": 0.7394,
|
| 374 |
-
"step": 26000
|
| 375 |
-
},
|
| 376 |
-
{
|
| 377 |
-
"epoch": 0.9181304784672418,
|
| 378 |
-
"grad_norm": 2.712963819503784,
|
| 379 |
-
"learning_rate": 2.7055399646606384e-05,
|
| 380 |
-
"loss": 0.7144,
|
| 381 |
-
"step": 26500
|
| 382 |
-
},
|
| 383 |
-
{
|
| 384 |
-
"epoch": 0.9354536950420954,
|
| 385 |
-
"grad_norm": 4.505606651306152,
|
| 386 |
-
"learning_rate": 2.6622319232235044e-05,
|
| 387 |
-
"loss": 0.6964,
|
| 388 |
-
"step": 27000
|
| 389 |
-
},
|
| 390 |
-
{
|
| 391 |
-
"epoch": 0.952776911616949,
|
| 392 |
-
"grad_norm": 1.9717707633972168,
|
| 393 |
-
"learning_rate": 2.619010497869244e-05,
|
| 394 |
-
"loss": 0.6642,
|
| 395 |
-
"step": 27500
|
| 396 |
-
},
|
| 397 |
-
{
|
| 398 |
-
"epoch": 0.9701001281918027,
|
| 399 |
-
"grad_norm": 2.501847743988037,
|
| 400 |
-
"learning_rate": 2.57570245643211e-05,
|
| 401 |
-
"loss": 0.6542,
|
| 402 |
-
"step": 28000
|
| 403 |
-
},
|
| 404 |
-
{
|
| 405 |
-
"epoch": 0.9874233447666563,
|
| 406 |
-
"grad_norm": 4.659865379333496,
|
| 407 |
-
"learning_rate": 2.532481031077851e-05,
|
| 408 |
-
"loss": 0.622,
|
| 409 |
-
"step": 28500
|
| 410 |
-
},
|
| 411 |
-
{
|
| 412 |
-
"epoch": 1.0,
|
| 413 |
-
"eval_loss": 0.3914269506931305,
|
| 414 |
-
"eval_runtime": 1756.0029,
|
| 415 |
-
"eval_samples_per_second": 350.648,
|
| 416 |
-
"eval_steps_per_second": 43.831,
|
| 417 |
-
"step": 28863
|
| 418 |
-
},
|
| 419 |
-
{
|
| 420 |
-
"epoch": 1.0047465613415099,
|
| 421 |
-
"grad_norm": 1.5512120723724365,
|
| 422 |
-
"learning_rate": 2.4891729896407165e-05,
|
| 423 |
-
"loss": 0.6,
|
| 424 |
-
"step": 29000
|
| 425 |
-
},
|
| 426 |
-
{
|
| 427 |
-
"epoch": 1.0220697779163634,
|
| 428 |
-
"grad_norm": 2.7224926948547363,
|
| 429 |
-
"learning_rate": 2.4458649482035826e-05,
|
| 430 |
-
"loss": 0.5733,
|
| 431 |
-
"step": 29500
|
| 432 |
-
},
|
| 433 |
-
{
|
| 434 |
-
"epoch": 1.039392994491217,
|
| 435 |
-
"grad_norm": 2.0893383026123047,
|
| 436 |
-
"learning_rate": 2.4025569067664486e-05,
|
| 437 |
-
"loss": 0.5517,
|
| 438 |
-
"step": 30000
|
| 439 |
-
},
|
| 440 |
-
{
|
| 441 |
-
"epoch": 1.0567162110660708,
|
| 442 |
-
"grad_norm": 2.235480785369873,
|
| 443 |
-
"learning_rate": 2.3592488653293143e-05,
|
| 444 |
-
"loss": 0.5389,
|
| 445 |
-
"step": 30500
|
| 446 |
-
},
|
| 447 |
-
{
|
| 448 |
-
"epoch": 1.0740394276409244,
|
| 449 |
-
"grad_norm": 2.4926114082336426,
|
| 450 |
-
"learning_rate": 2.3159408238921803e-05,
|
| 451 |
-
"loss": 0.5282,
|
| 452 |
-
"step": 31000
|
| 453 |
-
},
|
| 454 |
-
{
|
| 455 |
-
"epoch": 1.091362644215778,
|
| 456 |
-
"grad_norm": 1.9329782724380493,
|
| 457 |
-
"learning_rate": 2.2726327824550464e-05,
|
| 458 |
-
"loss": 0.5076,
|
| 459 |
-
"step": 31500
|
| 460 |
-
},
|
| 461 |
-
{
|
| 462 |
-
"epoch": 1.1086858607906316,
|
| 463 |
-
"grad_norm": 1.8705673217773438,
|
| 464 |
-
"learning_rate": 2.2293247410179124e-05,
|
| 465 |
-
"loss": 0.489,
|
| 466 |
-
"step": 32000
|
| 467 |
-
},
|
| 468 |
-
{
|
| 469 |
-
"epoch": 1.1260090773654852,
|
| 470 |
-
"grad_norm": 1.8372488021850586,
|
| 471 |
-
"learning_rate": 2.186016699580778e-05,
|
| 472 |
-
"loss": 0.4701,
|
| 473 |
-
"step": 32500
|
| 474 |
-
},
|
| 475 |
-
{
|
| 476 |
-
"epoch": 1.1433322939403388,
|
| 477 |
-
"grad_norm": 1.7131813764572144,
|
| 478 |
-
"learning_rate": 2.142708658143644e-05,
|
| 479 |
-
"loss": 0.4592,
|
| 480 |
-
"step": 33000
|
| 481 |
-
},
|
| 482 |
-
{
|
| 483 |
-
"epoch": 1.1606555105151926,
|
| 484 |
-
"grad_norm": 3.1449172496795654,
|
| 485 |
-
"learning_rate": 2.0994872327893845e-05,
|
| 486 |
-
"loss": 0.4408,
|
| 487 |
-
"step": 33500
|
| 488 |
-
},
|
| 489 |
-
{
|
| 490 |
-
"epoch": 1.1779787270900461,
|
| 491 |
-
"grad_norm": 2.199096918106079,
|
| 492 |
-
"learning_rate": 2.0561791913522505e-05,
|
| 493 |
-
"loss": 0.4278,
|
| 494 |
-
"step": 34000
|
| 495 |
-
},
|
| 496 |
-
{
|
| 497 |
-
"epoch": 1.1953019436648997,
|
| 498 |
-
"grad_norm": 1.6359717845916748,
|
| 499 |
-
"learning_rate": 2.0128711499151162e-05,
|
| 500 |
-
"loss": 0.4238,
|
| 501 |
-
"step": 34500
|
| 502 |
-
},
|
| 503 |
-
{
|
| 504 |
-
"epoch": 1.2126251602397533,
|
| 505 |
-
"grad_norm": 1.7747199535369873,
|
| 506 |
-
"learning_rate": 1.9695631084779822e-05,
|
| 507 |
-
"loss": 0.4024,
|
| 508 |
-
"step": 35000
|
| 509 |
-
},
|
| 510 |
-
{
|
| 511 |
-
"epoch": 1.229948376814607,
|
| 512 |
-
"grad_norm": 2.073904037475586,
|
| 513 |
-
"learning_rate": 1.9263416831237226e-05,
|
| 514 |
-
"loss": 0.39,
|
| 515 |
-
"step": 35500
|
| 516 |
-
},
|
| 517 |
-
{
|
| 518 |
-
"epoch": 1.2472715933894605,
|
| 519 |
-
"grad_norm": 1.9012521505355835,
|
| 520 |
-
"learning_rate": 1.8830336416865886e-05,
|
| 521 |
-
"loss": 0.3856,
|
| 522 |
-
"step": 36000
|
| 523 |
-
},
|
| 524 |
-
{
|
| 525 |
-
"epoch": 1.264594809964314,
|
| 526 |
-
"grad_norm": 2.2919228076934814,
|
| 527 |
-
"learning_rate": 1.8397256002494546e-05,
|
| 528 |
-
"loss": 0.3729,
|
| 529 |
-
"step": 36500
|
| 530 |
-
},
|
| 531 |
-
{
|
| 532 |
-
"epoch": 1.2819180265391679,
|
| 533 |
-
"grad_norm": 1.933583378791809,
|
| 534 |
-
"learning_rate": 1.7964175588123203e-05,
|
| 535 |
-
"loss": 0.3584,
|
| 536 |
-
"step": 37000
|
| 537 |
-
},
|
| 538 |
-
{
|
| 539 |
-
"epoch": 1.2992412431140214,
|
| 540 |
-
"grad_norm": 2.1168253421783447,
|
| 541 |
-
"learning_rate": 1.7531095173751864e-05,
|
| 542 |
-
"loss": 0.356,
|
| 543 |
-
"step": 37500
|
| 544 |
-
},
|
| 545 |
-
{
|
| 546 |
-
"epoch": 1.316564459688875,
|
| 547 |
-
"grad_norm": 1.7472686767578125,
|
| 548 |
-
"learning_rate": 1.7098014759380524e-05,
|
| 549 |
-
"loss": 0.339,
|
| 550 |
-
"step": 38000
|
| 551 |
-
},
|
| 552 |
-
{
|
| 553 |
-
"epoch": 1.3338876762637286,
|
| 554 |
-
"grad_norm": 1.9178825616836548,
|
| 555 |
-
"learning_rate": 1.666493434500918e-05,
|
| 556 |
-
"loss": 0.3294,
|
| 557 |
-
"step": 38500
|
| 558 |
-
},
|
| 559 |
-
{
|
| 560 |
-
"epoch": 1.3512108928385822,
|
| 561 |
-
"grad_norm": 1.4763795137405396,
|
| 562 |
-
"learning_rate": 1.623185393063784e-05,
|
| 563 |
-
"loss": 0.3299,
|
| 564 |
-
"step": 39000
|
| 565 |
-
},
|
| 566 |
-
{
|
| 567 |
-
"epoch": 1.368534109413436,
|
| 568 |
-
"grad_norm": 1.974442481994629,
|
| 569 |
-
"learning_rate": 1.57987735162665e-05,
|
| 570 |
-
"loss": 0.3206,
|
| 571 |
-
"step": 39500
|
| 572 |
-
},
|
| 573 |
-
{
|
| 574 |
-
"epoch": 1.3858573259882894,
|
| 575 |
-
"grad_norm": 1.315453052520752,
|
| 576 |
-
"learning_rate": 1.5366559262723905e-05,
|
| 577 |
-
"loss": 0.313,
|
| 578 |
-
"step": 40000
|
| 579 |
-
},
|
| 580 |
-
{
|
| 581 |
-
"epoch": 1.4031805425631432,
|
| 582 |
-
"grad_norm": 1.605042815208435,
|
| 583 |
-
"learning_rate": 1.4934345009181305e-05,
|
| 584 |
-
"loss": 0.2998,
|
| 585 |
-
"step": 40500
|
| 586 |
-
},
|
| 587 |
-
{
|
| 588 |
-
"epoch": 1.4205037591379968,
|
| 589 |
-
"grad_norm": 1.5163339376449585,
|
| 590 |
-
"learning_rate": 1.4501264594809966e-05,
|
| 591 |
-
"loss": 0.3031,
|
| 592 |
-
"step": 41000
|
| 593 |
-
},
|
| 594 |
-
{
|
| 595 |
-
"epoch": 1.4378269757128503,
|
| 596 |
-
"grad_norm": 1.9643980264663696,
|
| 597 |
-
"learning_rate": 1.4068184180438623e-05,
|
| 598 |
-
"loss": 0.2922,
|
| 599 |
-
"step": 41500
|
| 600 |
-
},
|
| 601 |
-
{
|
| 602 |
-
"epoch": 1.455150192287704,
|
| 603 |
-
"grad_norm": 1.4320045709609985,
|
| 604 |
-
"learning_rate": 1.3635103766067284e-05,
|
| 605 |
-
"loss": 0.2858,
|
| 606 |
-
"step": 42000
|
| 607 |
-
},
|
| 608 |
-
{
|
| 609 |
-
"epoch": 1.4724734088625575,
|
| 610 |
-
"grad_norm": 1.4415267705917358,
|
| 611 |
-
"learning_rate": 1.3202889512524685e-05,
|
| 612 |
-
"loss": 0.2807,
|
| 613 |
-
"step": 42500
|
| 614 |
-
},
|
| 615 |
-
{
|
| 616 |
-
"epoch": 1.4897966254374113,
|
| 617 |
-
"grad_norm": 1.1676790714263916,
|
| 618 |
-
"learning_rate": 1.2769809098153345e-05,
|
| 619 |
-
"loss": 0.279,
|
| 620 |
-
"step": 43000
|
| 621 |
-
},
|
| 622 |
-
{
|
| 623 |
-
"epoch": 1.5071198420122647,
|
| 624 |
-
"grad_norm": 1.2914575338363647,
|
| 625 |
-
"learning_rate": 1.2336728683782005e-05,
|
| 626 |
-
"loss": 0.2658,
|
| 627 |
-
"step": 43500
|
| 628 |
-
},
|
| 629 |
-
{
|
| 630 |
-
"epoch": 1.5244430585871185,
|
| 631 |
-
"grad_norm": 1.8342537879943848,
|
| 632 |
-
"learning_rate": 1.1903648269410666e-05,
|
| 633 |
-
"loss": 0.2669,
|
| 634 |
-
"step": 44000
|
| 635 |
-
},
|
| 636 |
-
{
|
| 637 |
-
"epoch": 1.541766275161972,
|
| 638 |
-
"grad_norm": 1.6250979900360107,
|
| 639 |
-
"learning_rate": 1.1471434015868068e-05,
|
| 640 |
-
"loss": 0.2615,
|
| 641 |
-
"step": 44500
|
| 642 |
-
},
|
| 643 |
-
{
|
| 644 |
-
"epoch": 1.5590894917368257,
|
| 645 |
-
"grad_norm": 1.8935564756393433,
|
| 646 |
-
"learning_rate": 1.1038353601496726e-05,
|
| 647 |
-
"loss": 0.2513,
|
| 648 |
-
"step": 45000
|
| 649 |
-
},
|
| 650 |
-
{
|
| 651 |
-
"epoch": 1.5764127083116795,
|
| 652 |
-
"grad_norm": 1.7190736532211304,
|
| 653 |
-
"learning_rate": 1.0605273187125386e-05,
|
| 654 |
-
"loss": 0.2488,
|
| 655 |
-
"step": 45500
|
| 656 |
-
},
|
| 657 |
-
{
|
| 658 |
-
"epoch": 1.5937359248865328,
|
| 659 |
-
"grad_norm": 1.5352853536605835,
|
| 660 |
-
"learning_rate": 1.0172192772754045e-05,
|
| 661 |
-
"loss": 0.2502,
|
| 662 |
-
"step": 46000
|
| 663 |
-
},
|
| 664 |
-
{
|
| 665 |
-
"epoch": 1.6110591414613866,
|
| 666 |
-
"grad_norm": 1.452185869216919,
|
| 667 |
-
"learning_rate": 9.739112358382704e-06,
|
| 668 |
-
"loss": 0.2441,
|
| 669 |
-
"step": 46500
|
| 670 |
-
},
|
| 671 |
-
{
|
| 672 |
-
"epoch": 1.6283823580362402,
|
| 673 |
-
"grad_norm": 1.2263524532318115,
|
| 674 |
-
"learning_rate": 9.306031944011364e-06,
|
| 675 |
-
"loss": 0.2366,
|
| 676 |
-
"step": 47000
|
| 677 |
-
},
|
| 678 |
-
{
|
| 679 |
-
"epoch": 1.6457055746110938,
|
| 680 |
-
"grad_norm": 1.7002058029174805,
|
| 681 |
-
"learning_rate": 8.872951529640024e-06,
|
| 682 |
-
"loss": 0.2361,
|
| 683 |
-
"step": 47500
|
| 684 |
-
},
|
| 685 |
-
{
|
| 686 |
-
"epoch": 1.6630287911859474,
|
| 687 |
-
"grad_norm": 0.8775345087051392,
|
| 688 |
-
"learning_rate": 8.439871115268685e-06,
|
| 689 |
-
"loss": 0.2314,
|
| 690 |
-
"step": 48000
|
| 691 |
-
},
|
| 692 |
-
{
|
| 693 |
-
"epoch": 1.680352007760801,
|
| 694 |
-
"grad_norm": 1.68264639377594,
|
| 695 |
-
"learning_rate": 8.007656861726086e-06,
|
| 696 |
-
"loss": 0.2341,
|
| 697 |
-
"step": 48500
|
| 698 |
-
},
|
| 699 |
-
{
|
| 700 |
-
"epoch": 1.6976752243356548,
|
| 701 |
-
"grad_norm": 1.4811515808105469,
|
| 702 |
-
"learning_rate": 7.574576447354745e-06,
|
| 703 |
-
"loss": 0.2238,
|
| 704 |
-
"step": 49000
|
| 705 |
-
},
|
| 706 |
-
{
|
| 707 |
-
"epoch": 1.7149984409105081,
|
| 708 |
-
"grad_norm": 1.5231512784957886,
|
| 709 |
-
"learning_rate": 7.142362193812148e-06,
|
| 710 |
-
"loss": 0.2229,
|
| 711 |
-
"step": 49500
|
| 712 |
-
},
|
| 713 |
-
{
|
| 714 |
-
"epoch": 1.732321657485362,
|
| 715 |
-
"grad_norm": 1.870348334312439,
|
| 716 |
-
"learning_rate": 6.709281779440807e-06,
|
| 717 |
-
"loss": 0.2211,
|
| 718 |
-
"step": 50000
|
| 719 |
-
},
|
| 720 |
-
{
|
| 721 |
-
"epoch": 1.7496448740602155,
|
| 722 |
-
"grad_norm": 1.2438400983810425,
|
| 723 |
-
"learning_rate": 6.276201365069466e-06,
|
| 724 |
-
"loss": 0.2205,
|
| 725 |
-
"step": 50500
|
| 726 |
-
},
|
| 727 |
-
{
|
| 728 |
-
"epoch": 1.766968090635069,
|
| 729 |
-
"grad_norm": 1.426330327987671,
|
| 730 |
-
"learning_rate": 5.843120950698125e-06,
|
| 731 |
-
"loss": 0.2189,
|
| 732 |
-
"step": 51000
|
| 733 |
-
},
|
| 734 |
-
{
|
| 735 |
-
"epoch": 1.7842913072099227,
|
| 736 |
-
"grad_norm": 0.9976692199707031,
|
| 737 |
-
"learning_rate": 5.410040536326786e-06,
|
| 738 |
-
"loss": 0.2168,
|
| 739 |
-
"step": 51500
|
| 740 |
-
},
|
| 741 |
-
{
|
| 742 |
-
"epoch": 1.8016145237847763,
|
| 743 |
-
"grad_norm": 0.7628383040428162,
|
| 744 |
-
"learning_rate": 4.976960121955445e-06,
|
| 745 |
-
"loss": 0.2137,
|
| 746 |
-
"step": 52000
|
| 747 |
-
},
|
| 748 |
-
{
|
| 749 |
-
"epoch": 1.81893774035963,
|
| 750 |
-
"grad_norm": 1.1943764686584473,
|
| 751 |
-
"learning_rate": 4.543879707584105e-06,
|
| 752 |
-
"loss": 0.2108,
|
| 753 |
-
"step": 52500
|
| 754 |
-
},
|
| 755 |
-
{
|
| 756 |
-
"epoch": 1.8362609569344834,
|
| 757 |
-
"grad_norm": 1.6423786878585815,
|
| 758 |
-
"learning_rate": 4.110799293212764e-06,
|
| 759 |
-
"loss": 0.207,
|
| 760 |
-
"step": 53000
|
| 761 |
-
},
|
| 762 |
-
{
|
| 763 |
-
"epoch": 1.8535841735093372,
|
| 764 |
-
"grad_norm": 1.1439303159713745,
|
| 765 |
-
"learning_rate": 3.678585039670166e-06,
|
| 766 |
-
"loss": 0.2102,
|
| 767 |
-
"step": 53500
|
| 768 |
-
},
|
| 769 |
-
{
|
| 770 |
-
"epoch": 1.8709073900841908,
|
| 771 |
-
"grad_norm": 1.1228655576705933,
|
| 772 |
-
"learning_rate": 3.2463707861275687e-06,
|
| 773 |
-
"loss": 0.2078,
|
| 774 |
-
"step": 54000
|
| 775 |
-
},
|
| 776 |
-
{
|
| 777 |
-
"epoch": 1.8882306066590444,
|
| 778 |
-
"grad_norm": 1.6855107545852661,
|
| 779 |
-
"learning_rate": 2.8132903717562278e-06,
|
| 780 |
-
"loss": 0.2095,
|
| 781 |
-
"step": 54500
|
| 782 |
-
},
|
| 783 |
-
{
|
| 784 |
-
"epoch": 1.9055538232338982,
|
| 785 |
-
"grad_norm": 1.0492366552352905,
|
| 786 |
-
"learning_rate": 2.3802099573848872e-06,
|
| 787 |
-
"loss": 0.2059,
|
| 788 |
-
"step": 55000
|
| 789 |
-
},
|
| 790 |
-
{
|
| 791 |
-
"epoch": 1.9228770398087516,
|
| 792 |
-
"grad_norm": 1.1293083429336548,
|
| 793 |
-
"learning_rate": 1.947129543013547e-06,
|
| 794 |
-
"loss": 0.2084,
|
| 795 |
-
"step": 55500
|
| 796 |
-
},
|
| 797 |
-
{
|
| 798 |
-
"epoch": 1.9402002563836054,
|
| 799 |
-
"grad_norm": 0.6616119742393494,
|
| 800 |
-
"learning_rate": 1.5140491286422063e-06,
|
| 801 |
-
"loss": 0.2004,
|
| 802 |
-
"step": 56000
|
| 803 |
-
},
|
| 804 |
-
{
|
| 805 |
-
"epoch": 1.957523472958459,
|
| 806 |
-
"grad_norm": 1.435498595237732,
|
| 807 |
-
"learning_rate": 1.0809687142708658e-06,
|
| 808 |
-
"loss": 0.2,
|
| 809 |
-
"step": 56500
|
| 810 |
-
},
|
| 811 |
-
{
|
| 812 |
-
"epoch": 1.9748466895333125,
|
| 813 |
-
"grad_norm": 1.5348039865493774,
|
| 814 |
-
"learning_rate": 6.478882998995254e-07,
|
| 815 |
-
"loss": 0.203,
|
| 816 |
-
"step": 57000
|
| 817 |
-
}
|
| 818 |
-
],
|
| 819 |
-
"logging_steps": 500,
|
| 820 |
-
"max_steps": 57726,
|
| 821 |
-
"num_input_tokens_seen": 0,
|
| 822 |
-
"num_train_epochs": 2,
|
| 823 |
-
"save_steps": 500,
|
| 824 |
-
"stateful_callbacks": {
|
| 825 |
-
"TrainerControl": {
|
| 826 |
-
"args": {
|
| 827 |
-
"should_epoch_stop": false,
|
| 828 |
-
"should_evaluate": false,
|
| 829 |
-
"should_log": false,
|
| 830 |
-
"should_save": true,
|
| 831 |
-
"should_training_stop": false
|
| 832 |
-
},
|
| 833 |
-
"attributes": {}
|
| 834 |
-
}
|
| 835 |
-
},
|
| 836 |
-
"total_flos": 1.2343118157525811e+17,
|
| 837 |
-
"train_batch_size": 32,
|
| 838 |
-
"trial_name": null,
|
| 839 |
-
"trial_params": null
|
| 840 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
checkpoints/checkpoint-57500/trainer_state.json
DELETED
|
@@ -1,847 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"best_global_step": null,
|
| 3 |
-
"best_metric": null,
|
| 4 |
-
"best_model_checkpoint": null,
|
| 5 |
-
"epoch": 1.9921699061081661,
|
| 6 |
-
"eval_steps": 500,
|
| 7 |
-
"global_step": 57500,
|
| 8 |
-
"is_hyper_param_search": false,
|
| 9 |
-
"is_local_process_zero": true,
|
| 10 |
-
"is_world_process_zero": true,
|
| 11 |
-
"log_history": [
|
| 12 |
-
{
|
| 13 |
-
"epoch": 0.01732321657485362,
|
| 14 |
-
"grad_norm": 0.5299897193908691,
|
| 15 |
-
"learning_rate": 4.956778574645741e-05,
|
| 16 |
-
"loss": 2.1733,
|
| 17 |
-
"step": 500
|
| 18 |
-
},
|
| 19 |
-
{
|
| 20 |
-
"epoch": 0.03464643314970724,
|
| 21 |
-
"grad_norm": 0.4744114279747009,
|
| 22 |
-
"learning_rate": 4.913470533208606e-05,
|
| 23 |
-
"loss": 1.936,
|
| 24 |
-
"step": 1000
|
| 25 |
-
},
|
| 26 |
-
{
|
| 27 |
-
"epoch": 0.051969649724560855,
|
| 28 |
-
"grad_norm": 0.5358501672744751,
|
| 29 |
-
"learning_rate": 4.870162491771472e-05,
|
| 30 |
-
"loss": 1.8605,
|
| 31 |
-
"step": 1500
|
| 32 |
-
},
|
| 33 |
-
{
|
| 34 |
-
"epoch": 0.06929286629941447,
|
| 35 |
-
"grad_norm": 0.5026484131813049,
|
| 36 |
-
"learning_rate": 4.826854450334338e-05,
|
| 37 |
-
"loss": 1.793,
|
| 38 |
-
"step": 2000
|
| 39 |
-
},
|
| 40 |
-
{
|
| 41 |
-
"epoch": 0.0866160828742681,
|
| 42 |
-
"grad_norm": 0.5730611681938171,
|
| 43 |
-
"learning_rate": 4.7835464088972044e-05,
|
| 44 |
-
"loss": 1.7309,
|
| 45 |
-
"step": 2500
|
| 46 |
-
},
|
| 47 |
-
{
|
| 48 |
-
"epoch": 0.10393929944912171,
|
| 49 |
-
"grad_norm": 0.47230127453804016,
|
| 50 |
-
"learning_rate": 4.74023836746007e-05,
|
| 51 |
-
"loss": 1.6938,
|
| 52 |
-
"step": 3000
|
| 53 |
-
},
|
| 54 |
-
{
|
| 55 |
-
"epoch": 0.12126251602397534,
|
| 56 |
-
"grad_norm": 0.6244428157806396,
|
| 57 |
-
"learning_rate": 4.6969303260229364e-05,
|
| 58 |
-
"loss": 1.6587,
|
| 59 |
-
"step": 3500
|
| 60 |
-
},
|
| 61 |
-
{
|
| 62 |
-
"epoch": 0.13858573259882895,
|
| 63 |
-
"grad_norm": 0.5069366693496704,
|
| 64 |
-
"learning_rate": 4.653622284585802e-05,
|
| 65 |
-
"loss": 1.6241,
|
| 66 |
-
"step": 4000
|
| 67 |
-
},
|
| 68 |
-
{
|
| 69 |
-
"epoch": 0.15590894917368256,
|
| 70 |
-
"grad_norm": 0.567051887512207,
|
| 71 |
-
"learning_rate": 4.610314243148668e-05,
|
| 72 |
-
"loss": 1.5789,
|
| 73 |
-
"step": 4500
|
| 74 |
-
},
|
| 75 |
-
{
|
| 76 |
-
"epoch": 0.1732321657485362,
|
| 77 |
-
"grad_norm": 0.5564538836479187,
|
| 78 |
-
"learning_rate": 4.567006201711534e-05,
|
| 79 |
-
"loss": 1.5619,
|
| 80 |
-
"step": 5000
|
| 81 |
-
},
|
| 82 |
-
{
|
| 83 |
-
"epoch": 0.1905553823233898,
|
| 84 |
-
"grad_norm": 0.7243916988372803,
|
| 85 |
-
"learning_rate": 4.5236981602744e-05,
|
| 86 |
-
"loss": 1.5358,
|
| 87 |
-
"step": 5500
|
| 88 |
-
},
|
| 89 |
-
{
|
| 90 |
-
"epoch": 0.20787859889824342,
|
| 91 |
-
"grad_norm": 0.6585371494293213,
|
| 92 |
-
"learning_rate": 4.480390118837266e-05,
|
| 93 |
-
"loss": 1.5396,
|
| 94 |
-
"step": 6000
|
| 95 |
-
},
|
| 96 |
-
{
|
| 97 |
-
"epoch": 0.22520181547309703,
|
| 98 |
-
"grad_norm": 0.5986710786819458,
|
| 99 |
-
"learning_rate": 4.437082077400132e-05,
|
| 100 |
-
"loss": 1.4991,
|
| 101 |
-
"step": 6500
|
| 102 |
-
},
|
| 103 |
-
{
|
| 104 |
-
"epoch": 0.24252503204795067,
|
| 105 |
-
"grad_norm": 0.6701385974884033,
|
| 106 |
-
"learning_rate": 4.393774035962998e-05,
|
| 107 |
-
"loss": 1.4823,
|
| 108 |
-
"step": 7000
|
| 109 |
-
},
|
| 110 |
-
{
|
| 111 |
-
"epoch": 0.25984824862280426,
|
| 112 |
-
"grad_norm": 0.6745362877845764,
|
| 113 |
-
"learning_rate": 4.350465994525863e-05,
|
| 114 |
-
"loss": 1.4715,
|
| 115 |
-
"step": 7500
|
| 116 |
-
},
|
| 117 |
-
{
|
| 118 |
-
"epoch": 0.2771714651976579,
|
| 119 |
-
"grad_norm": 0.6930545568466187,
|
| 120 |
-
"learning_rate": 4.30715795308873e-05,
|
| 121 |
-
"loss": 1.4354,
|
| 122 |
-
"step": 8000
|
| 123 |
-
},
|
| 124 |
-
{
|
| 125 |
-
"epoch": 0.29449468177251154,
|
| 126 |
-
"grad_norm": 0.6593695878982544,
|
| 127 |
-
"learning_rate": 4.2638499116515954e-05,
|
| 128 |
-
"loss": 1.432,
|
| 129 |
-
"step": 8500
|
| 130 |
-
},
|
| 131 |
-
{
|
| 132 |
-
"epoch": 0.3118178983473651,
|
| 133 |
-
"grad_norm": 0.6918864846229553,
|
| 134 |
-
"learning_rate": 4.220541870214462e-05,
|
| 135 |
-
"loss": 1.4209,
|
| 136 |
-
"step": 9000
|
| 137 |
-
},
|
| 138 |
-
{
|
| 139 |
-
"epoch": 0.32914111492221876,
|
| 140 |
-
"grad_norm": 0.7154921889305115,
|
| 141 |
-
"learning_rate": 4.1772338287773275e-05,
|
| 142 |
-
"loss": 1.391,
|
| 143 |
-
"step": 9500
|
| 144 |
-
},
|
| 145 |
-
{
|
| 146 |
-
"epoch": 0.3464643314970724,
|
| 147 |
-
"grad_norm": 0.8634065389633179,
|
| 148 |
-
"learning_rate": 4.133925787340194e-05,
|
| 149 |
-
"loss": 1.3843,
|
| 150 |
-
"step": 10000
|
| 151 |
-
},
|
| 152 |
-
{
|
| 153 |
-
"epoch": 0.363787548071926,
|
| 154 |
-
"grad_norm": 0.8820887804031372,
|
| 155 |
-
"learning_rate": 4.0906177459030595e-05,
|
| 156 |
-
"loss": 1.3612,
|
| 157 |
-
"step": 10500
|
| 158 |
-
},
|
| 159 |
-
{
|
| 160 |
-
"epoch": 0.3811107646467796,
|
| 161 |
-
"grad_norm": 0.9643932580947876,
|
| 162 |
-
"learning_rate": 4.047309704465925e-05,
|
| 163 |
-
"loss": 1.3326,
|
| 164 |
-
"step": 11000
|
| 165 |
-
},
|
| 166 |
-
{
|
| 167 |
-
"epoch": 0.39843398122163326,
|
| 168 |
-
"grad_norm": 0.9170373678207397,
|
| 169 |
-
"learning_rate": 4.0040016630287916e-05,
|
| 170 |
-
"loss": 1.3158,
|
| 171 |
-
"step": 11500
|
| 172 |
-
},
|
| 173 |
-
{
|
| 174 |
-
"epoch": 0.41575719779648684,
|
| 175 |
-
"grad_norm": 0.8911833167076111,
|
| 176 |
-
"learning_rate": 3.960693621591657e-05,
|
| 177 |
-
"loss": 1.2982,
|
| 178 |
-
"step": 12000
|
| 179 |
-
},
|
| 180 |
-
{
|
| 181 |
-
"epoch": 0.4330804143713405,
|
| 182 |
-
"grad_norm": 1.3023812770843506,
|
| 183 |
-
"learning_rate": 3.917472196237398e-05,
|
| 184 |
-
"loss": 1.2497,
|
| 185 |
-
"step": 12500
|
| 186 |
-
},
|
| 187 |
-
{
|
| 188 |
-
"epoch": 0.45040363094619407,
|
| 189 |
-
"grad_norm": 1.3367282152175903,
|
| 190 |
-
"learning_rate": 3.874164154800263e-05,
|
| 191 |
-
"loss": 1.2436,
|
| 192 |
-
"step": 13000
|
| 193 |
-
},
|
| 194 |
-
{
|
| 195 |
-
"epoch": 0.4677268475210477,
|
| 196 |
-
"grad_norm": 1.070709228515625,
|
| 197 |
-
"learning_rate": 3.8308561133631293e-05,
|
| 198 |
-
"loss": 1.2068,
|
| 199 |
-
"step": 13500
|
| 200 |
-
},
|
| 201 |
-
{
|
| 202 |
-
"epoch": 0.48505006409590135,
|
| 203 |
-
"grad_norm": 1.2574782371520996,
|
| 204 |
-
"learning_rate": 3.787548071925995e-05,
|
| 205 |
-
"loss": 1.1867,
|
| 206 |
-
"step": 14000
|
| 207 |
-
},
|
| 208 |
-
{
|
| 209 |
-
"epoch": 0.5023732806707549,
|
| 210 |
-
"grad_norm": 1.4889111518859863,
|
| 211 |
-
"learning_rate": 3.744326646571736e-05,
|
| 212 |
-
"loss": 1.1908,
|
| 213 |
-
"step": 14500
|
| 214 |
-
},
|
| 215 |
-
{
|
| 216 |
-
"epoch": 0.5196964972456085,
|
| 217 |
-
"grad_norm": 1.2556302547454834,
|
| 218 |
-
"learning_rate": 3.7010186051346014e-05,
|
| 219 |
-
"loss": 1.1476,
|
| 220 |
-
"step": 15000
|
| 221 |
-
},
|
| 222 |
-
{
|
| 223 |
-
"epoch": 0.5370197138204622,
|
| 224 |
-
"grad_norm": 1.2582406997680664,
|
| 225 |
-
"learning_rate": 3.6577971797803415e-05,
|
| 226 |
-
"loss": 1.1268,
|
| 227 |
-
"step": 15500
|
| 228 |
-
},
|
| 229 |
-
{
|
| 230 |
-
"epoch": 0.5543429303953158,
|
| 231 |
-
"grad_norm": 1.4035013914108276,
|
| 232 |
-
"learning_rate": 3.614575754426082e-05,
|
| 233 |
-
"loss": 1.119,
|
| 234 |
-
"step": 16000
|
| 235 |
-
},
|
| 236 |
-
{
|
| 237 |
-
"epoch": 0.5716661469701694,
|
| 238 |
-
"grad_norm": 1.7117546796798706,
|
| 239 |
-
"learning_rate": 3.571267712988948e-05,
|
| 240 |
-
"loss": 1.106,
|
| 241 |
-
"step": 16500
|
| 242 |
-
},
|
| 243 |
-
{
|
| 244 |
-
"epoch": 0.5889893635450231,
|
| 245 |
-
"grad_norm": 1.503953218460083,
|
| 246 |
-
"learning_rate": 3.5279596715518135e-05,
|
| 247 |
-
"loss": 1.08,
|
| 248 |
-
"step": 17000
|
| 249 |
-
},
|
| 250 |
-
{
|
| 251 |
-
"epoch": 0.6063125801198767,
|
| 252 |
-
"grad_norm": 1.6753602027893066,
|
| 253 |
-
"learning_rate": 3.48465163011468e-05,
|
| 254 |
-
"loss": 1.055,
|
| 255 |
-
"step": 17500
|
| 256 |
-
},
|
| 257 |
-
{
|
| 258 |
-
"epoch": 0.6236357966947302,
|
| 259 |
-
"grad_norm": 1.5606533288955688,
|
| 260 |
-
"learning_rate": 3.4413435886775456e-05,
|
| 261 |
-
"loss": 1.043,
|
| 262 |
-
"step": 18000
|
| 263 |
-
},
|
| 264 |
-
{
|
| 265 |
-
"epoch": 0.6409590132695839,
|
| 266 |
-
"grad_norm": 2.1966652870178223,
|
| 267 |
-
"learning_rate": 3.398122163323286e-05,
|
| 268 |
-
"loss": 1.0294,
|
| 269 |
-
"step": 18500
|
| 270 |
-
},
|
| 271 |
-
{
|
| 272 |
-
"epoch": 0.6582822298444375,
|
| 273 |
-
"grad_norm": 2.0083296298980713,
|
| 274 |
-
"learning_rate": 3.354814121886152e-05,
|
| 275 |
-
"loss": 1.009,
|
| 276 |
-
"step": 19000
|
| 277 |
-
},
|
| 278 |
-
{
|
| 279 |
-
"epoch": 0.6756054464192911,
|
| 280 |
-
"grad_norm": 1.61182701587677,
|
| 281 |
-
"learning_rate": 3.311506080449018e-05,
|
| 282 |
-
"loss": 0.978,
|
| 283 |
-
"step": 19500
|
| 284 |
-
},
|
| 285 |
-
{
|
| 286 |
-
"epoch": 0.6929286629941448,
|
| 287 |
-
"grad_norm": 1.698030710220337,
|
| 288 |
-
"learning_rate": 3.268198039011884e-05,
|
| 289 |
-
"loss": 0.9798,
|
| 290 |
-
"step": 20000
|
| 291 |
-
},
|
| 292 |
-
{
|
| 293 |
-
"epoch": 0.7102518795689984,
|
| 294 |
-
"grad_norm": 1.810142993927002,
|
| 295 |
-
"learning_rate": 3.224976613657624e-05,
|
| 296 |
-
"loss": 0.9558,
|
| 297 |
-
"step": 20500
|
| 298 |
-
},
|
| 299 |
-
{
|
| 300 |
-
"epoch": 0.727575096143852,
|
| 301 |
-
"grad_norm": 4.462728023529053,
|
| 302 |
-
"learning_rate": 3.1816685722204904e-05,
|
| 303 |
-
"loss": 0.9337,
|
| 304 |
-
"step": 21000
|
| 305 |
-
},
|
| 306 |
-
{
|
| 307 |
-
"epoch": 0.7448983127187057,
|
| 308 |
-
"grad_norm": 3.5566582679748535,
|
| 309 |
-
"learning_rate": 3.1384471468662305e-05,
|
| 310 |
-
"loss": 0.9115,
|
| 311 |
-
"step": 21500
|
| 312 |
-
},
|
| 313 |
-
{
|
| 314 |
-
"epoch": 0.7622215292935592,
|
| 315 |
-
"grad_norm": 3.2824957370758057,
|
| 316 |
-
"learning_rate": 3.095139105429096e-05,
|
| 317 |
-
"loss": 0.8992,
|
| 318 |
-
"step": 22000
|
| 319 |
-
},
|
| 320 |
-
{
|
| 321 |
-
"epoch": 0.7795447458684128,
|
| 322 |
-
"grad_norm": 3.1051690578460693,
|
| 323 |
-
"learning_rate": 3.0518310639919625e-05,
|
| 324 |
-
"loss": 0.8945,
|
| 325 |
-
"step": 22500
|
| 326 |
-
},
|
| 327 |
-
{
|
| 328 |
-
"epoch": 0.7968679624432665,
|
| 329 |
-
"grad_norm": 5.406064510345459,
|
| 330 |
-
"learning_rate": 3.0085230225548282e-05,
|
| 331 |
-
"loss": 0.8544,
|
| 332 |
-
"step": 23000
|
| 333 |
-
},
|
| 334 |
-
{
|
| 335 |
-
"epoch": 0.8141911790181201,
|
| 336 |
-
"grad_norm": 3.3169949054718018,
|
| 337 |
-
"learning_rate": 2.9653015972005682e-05,
|
| 338 |
-
"loss": 0.8494,
|
| 339 |
-
"step": 23500
|
| 340 |
-
},
|
| 341 |
-
{
|
| 342 |
-
"epoch": 0.8315143955929737,
|
| 343 |
-
"grad_norm": 2.4768149852752686,
|
| 344 |
-
"learning_rate": 2.9219935557634343e-05,
|
| 345 |
-
"loss": 0.8145,
|
| 346 |
-
"step": 24000
|
| 347 |
-
},
|
| 348 |
-
{
|
| 349 |
-
"epoch": 0.8488376121678273,
|
| 350 |
-
"grad_norm": 2.6160104274749756,
|
| 351 |
-
"learning_rate": 2.8786855143263003e-05,
|
| 352 |
-
"loss": 0.7995,
|
| 353 |
-
"step": 24500
|
| 354 |
-
},
|
| 355 |
-
{
|
| 356 |
-
"epoch": 0.866160828742681,
|
| 357 |
-
"grad_norm": 2.7828731536865234,
|
| 358 |
-
"learning_rate": 2.8353774728891663e-05,
|
| 359 |
-
"loss": 0.7705,
|
| 360 |
-
"step": 25000
|
| 361 |
-
},
|
| 362 |
-
{
|
| 363 |
-
"epoch": 0.8834840453175345,
|
| 364 |
-
"grad_norm": 3.508859157562256,
|
| 365 |
-
"learning_rate": 2.7921560475349063e-05,
|
| 366 |
-
"loss": 0.7565,
|
| 367 |
-
"step": 25500
|
| 368 |
-
},
|
| 369 |
-
{
|
| 370 |
-
"epoch": 0.9008072618923881,
|
| 371 |
-
"grad_norm": 2.672107219696045,
|
| 372 |
-
"learning_rate": 2.7488480060977724e-05,
|
| 373 |
-
"loss": 0.7394,
|
| 374 |
-
"step": 26000
|
| 375 |
-
},
|
| 376 |
-
{
|
| 377 |
-
"epoch": 0.9181304784672418,
|
| 378 |
-
"grad_norm": 2.712963819503784,
|
| 379 |
-
"learning_rate": 2.7055399646606384e-05,
|
| 380 |
-
"loss": 0.7144,
|
| 381 |
-
"step": 26500
|
| 382 |
-
},
|
| 383 |
-
{
|
| 384 |
-
"epoch": 0.9354536950420954,
|
| 385 |
-
"grad_norm": 4.505606651306152,
|
| 386 |
-
"learning_rate": 2.6622319232235044e-05,
|
| 387 |
-
"loss": 0.6964,
|
| 388 |
-
"step": 27000
|
| 389 |
-
},
|
| 390 |
-
{
|
| 391 |
-
"epoch": 0.952776911616949,
|
| 392 |
-
"grad_norm": 1.9717707633972168,
|
| 393 |
-
"learning_rate": 2.619010497869244e-05,
|
| 394 |
-
"loss": 0.6642,
|
| 395 |
-
"step": 27500
|
| 396 |
-
},
|
| 397 |
-
{
|
| 398 |
-
"epoch": 0.9701001281918027,
|
| 399 |
-
"grad_norm": 2.501847743988037,
|
| 400 |
-
"learning_rate": 2.57570245643211e-05,
|
| 401 |
-
"loss": 0.6542,
|
| 402 |
-
"step": 28000
|
| 403 |
-
},
|
| 404 |
-
{
|
| 405 |
-
"epoch": 0.9874233447666563,
|
| 406 |
-
"grad_norm": 4.659865379333496,
|
| 407 |
-
"learning_rate": 2.532481031077851e-05,
|
| 408 |
-
"loss": 0.622,
|
| 409 |
-
"step": 28500
|
| 410 |
-
},
|
| 411 |
-
{
|
| 412 |
-
"epoch": 1.0,
|
| 413 |
-
"eval_loss": 0.3914269506931305,
|
| 414 |
-
"eval_runtime": 1756.0029,
|
| 415 |
-
"eval_samples_per_second": 350.648,
|
| 416 |
-
"eval_steps_per_second": 43.831,
|
| 417 |
-
"step": 28863
|
| 418 |
-
},
|
| 419 |
-
{
|
| 420 |
-
"epoch": 1.0047465613415099,
|
| 421 |
-
"grad_norm": 1.5512120723724365,
|
| 422 |
-
"learning_rate": 2.4891729896407165e-05,
|
| 423 |
-
"loss": 0.6,
|
| 424 |
-
"step": 29000
|
| 425 |
-
},
|
| 426 |
-
{
|
| 427 |
-
"epoch": 1.0220697779163634,
|
| 428 |
-
"grad_norm": 2.7224926948547363,
|
| 429 |
-
"learning_rate": 2.4458649482035826e-05,
|
| 430 |
-
"loss": 0.5733,
|
| 431 |
-
"step": 29500
|
| 432 |
-
},
|
| 433 |
-
{
|
| 434 |
-
"epoch": 1.039392994491217,
|
| 435 |
-
"grad_norm": 2.0893383026123047,
|
| 436 |
-
"learning_rate": 2.4025569067664486e-05,
|
| 437 |
-
"loss": 0.5517,
|
| 438 |
-
"step": 30000
|
| 439 |
-
},
|
| 440 |
-
{
|
| 441 |
-
"epoch": 1.0567162110660708,
|
| 442 |
-
"grad_norm": 2.235480785369873,
|
| 443 |
-
"learning_rate": 2.3592488653293143e-05,
|
| 444 |
-
"loss": 0.5389,
|
| 445 |
-
"step": 30500
|
| 446 |
-
},
|
| 447 |
-
{
|
| 448 |
-
"epoch": 1.0740394276409244,
|
| 449 |
-
"grad_norm": 2.4926114082336426,
|
| 450 |
-
"learning_rate": 2.3159408238921803e-05,
|
| 451 |
-
"loss": 0.5282,
|
| 452 |
-
"step": 31000
|
| 453 |
-
},
|
| 454 |
-
{
|
| 455 |
-
"epoch": 1.091362644215778,
|
| 456 |
-
"grad_norm": 1.9329782724380493,
|
| 457 |
-
"learning_rate": 2.2726327824550464e-05,
|
| 458 |
-
"loss": 0.5076,
|
| 459 |
-
"step": 31500
|
| 460 |
-
},
|
| 461 |
-
{
|
| 462 |
-
"epoch": 1.1086858607906316,
|
| 463 |
-
"grad_norm": 1.8705673217773438,
|
| 464 |
-
"learning_rate": 2.2293247410179124e-05,
|
| 465 |
-
"loss": 0.489,
|
| 466 |
-
"step": 32000
|
| 467 |
-
},
|
| 468 |
-
{
|
| 469 |
-
"epoch": 1.1260090773654852,
|
| 470 |
-
"grad_norm": 1.8372488021850586,
|
| 471 |
-
"learning_rate": 2.186016699580778e-05,
|
| 472 |
-
"loss": 0.4701,
|
| 473 |
-
"step": 32500
|
| 474 |
-
},
|
| 475 |
-
{
|
| 476 |
-
"epoch": 1.1433322939403388,
|
| 477 |
-
"grad_norm": 1.7131813764572144,
|
| 478 |
-
"learning_rate": 2.142708658143644e-05,
|
| 479 |
-
"loss": 0.4592,
|
| 480 |
-
"step": 33000
|
| 481 |
-
},
|
| 482 |
-
{
|
| 483 |
-
"epoch": 1.1606555105151926,
|
| 484 |
-
"grad_norm": 3.1449172496795654,
|
| 485 |
-
"learning_rate": 2.0994872327893845e-05,
|
| 486 |
-
"loss": 0.4408,
|
| 487 |
-
"step": 33500
|
| 488 |
-
},
|
| 489 |
-
{
|
| 490 |
-
"epoch": 1.1779787270900461,
|
| 491 |
-
"grad_norm": 2.199096918106079,
|
| 492 |
-
"learning_rate": 2.0561791913522505e-05,
|
| 493 |
-
"loss": 0.4278,
|
| 494 |
-
"step": 34000
|
| 495 |
-
},
|
| 496 |
-
{
|
| 497 |
-
"epoch": 1.1953019436648997,
|
| 498 |
-
"grad_norm": 1.6359717845916748,
|
| 499 |
-
"learning_rate": 2.0128711499151162e-05,
|
| 500 |
-
"loss": 0.4238,
|
| 501 |
-
"step": 34500
|
| 502 |
-
},
|
| 503 |
-
{
|
| 504 |
-
"epoch": 1.2126251602397533,
|
| 505 |
-
"grad_norm": 1.7747199535369873,
|
| 506 |
-
"learning_rate": 1.9695631084779822e-05,
|
| 507 |
-
"loss": 0.4024,
|
| 508 |
-
"step": 35000
|
| 509 |
-
},
|
| 510 |
-
{
|
| 511 |
-
"epoch": 1.229948376814607,
|
| 512 |
-
"grad_norm": 2.073904037475586,
|
| 513 |
-
"learning_rate": 1.9263416831237226e-05,
|
| 514 |
-
"loss": 0.39,
|
| 515 |
-
"step": 35500
|
| 516 |
-
},
|
| 517 |
-
{
|
| 518 |
-
"epoch": 1.2472715933894605,
|
| 519 |
-
"grad_norm": 1.9012521505355835,
|
| 520 |
-
"learning_rate": 1.8830336416865886e-05,
|
| 521 |
-
"loss": 0.3856,
|
| 522 |
-
"step": 36000
|
| 523 |
-
},
|
| 524 |
-
{
|
| 525 |
-
"epoch": 1.264594809964314,
|
| 526 |
-
"grad_norm": 2.2919228076934814,
|
| 527 |
-
"learning_rate": 1.8397256002494546e-05,
|
| 528 |
-
"loss": 0.3729,
|
| 529 |
-
"step": 36500
|
| 530 |
-
},
|
| 531 |
-
{
|
| 532 |
-
"epoch": 1.2819180265391679,
|
| 533 |
-
"grad_norm": 1.933583378791809,
|
| 534 |
-
"learning_rate": 1.7964175588123203e-05,
|
| 535 |
-
"loss": 0.3584,
|
| 536 |
-
"step": 37000
|
| 537 |
-
},
|
| 538 |
-
{
|
| 539 |
-
"epoch": 1.2992412431140214,
|
| 540 |
-
"grad_norm": 2.1168253421783447,
|
| 541 |
-
"learning_rate": 1.7531095173751864e-05,
|
| 542 |
-
"loss": 0.356,
|
| 543 |
-
"step": 37500
|
| 544 |
-
},
|
| 545 |
-
{
|
| 546 |
-
"epoch": 1.316564459688875,
|
| 547 |
-
"grad_norm": 1.7472686767578125,
|
| 548 |
-
"learning_rate": 1.7098014759380524e-05,
|
| 549 |
-
"loss": 0.339,
|
| 550 |
-
"step": 38000
|
| 551 |
-
},
|
| 552 |
-
{
|
| 553 |
-
"epoch": 1.3338876762637286,
|
| 554 |
-
"grad_norm": 1.9178825616836548,
|
| 555 |
-
"learning_rate": 1.666493434500918e-05,
|
| 556 |
-
"loss": 0.3294,
|
| 557 |
-
"step": 38500
|
| 558 |
-
},
|
| 559 |
-
{
|
| 560 |
-
"epoch": 1.3512108928385822,
|
| 561 |
-
"grad_norm": 1.4763795137405396,
|
| 562 |
-
"learning_rate": 1.623185393063784e-05,
|
| 563 |
-
"loss": 0.3299,
|
| 564 |
-
"step": 39000
|
| 565 |
-
},
|
| 566 |
-
{
|
| 567 |
-
"epoch": 1.368534109413436,
|
| 568 |
-
"grad_norm": 1.974442481994629,
|
| 569 |
-
"learning_rate": 1.57987735162665e-05,
|
| 570 |
-
"loss": 0.3206,
|
| 571 |
-
"step": 39500
|
| 572 |
-
},
|
| 573 |
-
{
|
| 574 |
-
"epoch": 1.3858573259882894,
|
| 575 |
-
"grad_norm": 1.315453052520752,
|
| 576 |
-
"learning_rate": 1.5366559262723905e-05,
|
| 577 |
-
"loss": 0.313,
|
| 578 |
-
"step": 40000
|
| 579 |
-
},
|
| 580 |
-
{
|
| 581 |
-
"epoch": 1.4031805425631432,
|
| 582 |
-
"grad_norm": 1.605042815208435,
|
| 583 |
-
"learning_rate": 1.4934345009181305e-05,
|
| 584 |
-
"loss": 0.2998,
|
| 585 |
-
"step": 40500
|
| 586 |
-
},
|
| 587 |
-
{
|
| 588 |
-
"epoch": 1.4205037591379968,
|
| 589 |
-
"grad_norm": 1.5163339376449585,
|
| 590 |
-
"learning_rate": 1.4501264594809966e-05,
|
| 591 |
-
"loss": 0.3031,
|
| 592 |
-
"step": 41000
|
| 593 |
-
},
|
| 594 |
-
{
|
| 595 |
-
"epoch": 1.4378269757128503,
|
| 596 |
-
"grad_norm": 1.9643980264663696,
|
| 597 |
-
"learning_rate": 1.4068184180438623e-05,
|
| 598 |
-
"loss": 0.2922,
|
| 599 |
-
"step": 41500
|
| 600 |
-
},
|
| 601 |
-
{
|
| 602 |
-
"epoch": 1.455150192287704,
|
| 603 |
-
"grad_norm": 1.4320045709609985,
|
| 604 |
-
"learning_rate": 1.3635103766067284e-05,
|
| 605 |
-
"loss": 0.2858,
|
| 606 |
-
"step": 42000
|
| 607 |
-
},
|
| 608 |
-
{
|
| 609 |
-
"epoch": 1.4724734088625575,
|
| 610 |
-
"grad_norm": 1.4415267705917358,
|
| 611 |
-
"learning_rate": 1.3202889512524685e-05,
|
| 612 |
-
"loss": 0.2807,
|
| 613 |
-
"step": 42500
|
| 614 |
-
},
|
| 615 |
-
{
|
| 616 |
-
"epoch": 1.4897966254374113,
|
| 617 |
-
"grad_norm": 1.1676790714263916,
|
| 618 |
-
"learning_rate": 1.2769809098153345e-05,
|
| 619 |
-
"loss": 0.279,
|
| 620 |
-
"step": 43000
|
| 621 |
-
},
|
| 622 |
-
{
|
| 623 |
-
"epoch": 1.5071198420122647,
|
| 624 |
-
"grad_norm": 1.2914575338363647,
|
| 625 |
-
"learning_rate": 1.2336728683782005e-05,
|
| 626 |
-
"loss": 0.2658,
|
| 627 |
-
"step": 43500
|
| 628 |
-
},
|
| 629 |
-
{
|
| 630 |
-
"epoch": 1.5244430585871185,
|
| 631 |
-
"grad_norm": 1.8342537879943848,
|
| 632 |
-
"learning_rate": 1.1903648269410666e-05,
|
| 633 |
-
"loss": 0.2669,
|
| 634 |
-
"step": 44000
|
| 635 |
-
},
|
| 636 |
-
{
|
| 637 |
-
"epoch": 1.541766275161972,
|
| 638 |
-
"grad_norm": 1.6250979900360107,
|
| 639 |
-
"learning_rate": 1.1471434015868068e-05,
|
| 640 |
-
"loss": 0.2615,
|
| 641 |
-
"step": 44500
|
| 642 |
-
},
|
| 643 |
-
{
|
| 644 |
-
"epoch": 1.5590894917368257,
|
| 645 |
-
"grad_norm": 1.8935564756393433,
|
| 646 |
-
"learning_rate": 1.1038353601496726e-05,
|
| 647 |
-
"loss": 0.2513,
|
| 648 |
-
"step": 45000
|
| 649 |
-
},
|
| 650 |
-
{
|
| 651 |
-
"epoch": 1.5764127083116795,
|
| 652 |
-
"grad_norm": 1.7190736532211304,
|
| 653 |
-
"learning_rate": 1.0605273187125386e-05,
|
| 654 |
-
"loss": 0.2488,
|
| 655 |
-
"step": 45500
|
| 656 |
-
},
|
| 657 |
-
{
|
| 658 |
-
"epoch": 1.5937359248865328,
|
| 659 |
-
"grad_norm": 1.5352853536605835,
|
| 660 |
-
"learning_rate": 1.0172192772754045e-05,
|
| 661 |
-
"loss": 0.2502,
|
| 662 |
-
"step": 46000
|
| 663 |
-
},
|
| 664 |
-
{
|
| 665 |
-
"epoch": 1.6110591414613866,
|
| 666 |
-
"grad_norm": 1.452185869216919,
|
| 667 |
-
"learning_rate": 9.739112358382704e-06,
|
| 668 |
-
"loss": 0.2441,
|
| 669 |
-
"step": 46500
|
| 670 |
-
},
|
| 671 |
-
{
|
| 672 |
-
"epoch": 1.6283823580362402,
|
| 673 |
-
"grad_norm": 1.2263524532318115,
|
| 674 |
-
"learning_rate": 9.306031944011364e-06,
|
| 675 |
-
"loss": 0.2366,
|
| 676 |
-
"step": 47000
|
| 677 |
-
},
|
| 678 |
-
{
|
| 679 |
-
"epoch": 1.6457055746110938,
|
| 680 |
-
"grad_norm": 1.7002058029174805,
|
| 681 |
-
"learning_rate": 8.872951529640024e-06,
|
| 682 |
-
"loss": 0.2361,
|
| 683 |
-
"step": 47500
|
| 684 |
-
},
|
| 685 |
-
{
|
| 686 |
-
"epoch": 1.6630287911859474,
|
| 687 |
-
"grad_norm": 0.8775345087051392,
|
| 688 |
-
"learning_rate": 8.439871115268685e-06,
|
| 689 |
-
"loss": 0.2314,
|
| 690 |
-
"step": 48000
|
| 691 |
-
},
|
| 692 |
-
{
|
| 693 |
-
"epoch": 1.680352007760801,
|
| 694 |
-
"grad_norm": 1.68264639377594,
|
| 695 |
-
"learning_rate": 8.007656861726086e-06,
|
| 696 |
-
"loss": 0.2341,
|
| 697 |
-
"step": 48500
|
| 698 |
-
},
|
| 699 |
-
{
|
| 700 |
-
"epoch": 1.6976752243356548,
|
| 701 |
-
"grad_norm": 1.4811515808105469,
|
| 702 |
-
"learning_rate": 7.574576447354745e-06,
|
| 703 |
-
"loss": 0.2238,
|
| 704 |
-
"step": 49000
|
| 705 |
-
},
|
| 706 |
-
{
|
| 707 |
-
"epoch": 1.7149984409105081,
|
| 708 |
-
"grad_norm": 1.5231512784957886,
|
| 709 |
-
"learning_rate": 7.142362193812148e-06,
|
| 710 |
-
"loss": 0.2229,
|
| 711 |
-
"step": 49500
|
| 712 |
-
},
|
| 713 |
-
{
|
| 714 |
-
"epoch": 1.732321657485362,
|
| 715 |
-
"grad_norm": 1.870348334312439,
|
| 716 |
-
"learning_rate": 6.709281779440807e-06,
|
| 717 |
-
"loss": 0.2211,
|
| 718 |
-
"step": 50000
|
| 719 |
-
},
|
| 720 |
-
{
|
| 721 |
-
"epoch": 1.7496448740602155,
|
| 722 |
-
"grad_norm": 1.2438400983810425,
|
| 723 |
-
"learning_rate": 6.276201365069466e-06,
|
| 724 |
-
"loss": 0.2205,
|
| 725 |
-
"step": 50500
|
| 726 |
-
},
|
| 727 |
-
{
|
| 728 |
-
"epoch": 1.766968090635069,
|
| 729 |
-
"grad_norm": 1.426330327987671,
|
| 730 |
-
"learning_rate": 5.843120950698125e-06,
|
| 731 |
-
"loss": 0.2189,
|
| 732 |
-
"step": 51000
|
| 733 |
-
},
|
| 734 |
-
{
|
| 735 |
-
"epoch": 1.7842913072099227,
|
| 736 |
-
"grad_norm": 0.9976692199707031,
|
| 737 |
-
"learning_rate": 5.410040536326786e-06,
|
| 738 |
-
"loss": 0.2168,
|
| 739 |
-
"step": 51500
|
| 740 |
-
},
|
| 741 |
-
{
|
| 742 |
-
"epoch": 1.8016145237847763,
|
| 743 |
-
"grad_norm": 0.7628383040428162,
|
| 744 |
-
"learning_rate": 4.976960121955445e-06,
|
| 745 |
-
"loss": 0.2137,
|
| 746 |
-
"step": 52000
|
| 747 |
-
},
|
| 748 |
-
{
|
| 749 |
-
"epoch": 1.81893774035963,
|
| 750 |
-
"grad_norm": 1.1943764686584473,
|
| 751 |
-
"learning_rate": 4.543879707584105e-06,
|
| 752 |
-
"loss": 0.2108,
|
| 753 |
-
"step": 52500
|
| 754 |
-
},
|
| 755 |
-
{
|
| 756 |
-
"epoch": 1.8362609569344834,
|
| 757 |
-
"grad_norm": 1.6423786878585815,
|
| 758 |
-
"learning_rate": 4.110799293212764e-06,
|
| 759 |
-
"loss": 0.207,
|
| 760 |
-
"step": 53000
|
| 761 |
-
},
|
| 762 |
-
{
|
| 763 |
-
"epoch": 1.8535841735093372,
|
| 764 |
-
"grad_norm": 1.1439303159713745,
|
| 765 |
-
"learning_rate": 3.678585039670166e-06,
|
| 766 |
-
"loss": 0.2102,
|
| 767 |
-
"step": 53500
|
| 768 |
-
},
|
| 769 |
-
{
|
| 770 |
-
"epoch": 1.8709073900841908,
|
| 771 |
-
"grad_norm": 1.1228655576705933,
|
| 772 |
-
"learning_rate": 3.2463707861275687e-06,
|
| 773 |
-
"loss": 0.2078,
|
| 774 |
-
"step": 54000
|
| 775 |
-
},
|
| 776 |
-
{
|
| 777 |
-
"epoch": 1.8882306066590444,
|
| 778 |
-
"grad_norm": 1.6855107545852661,
|
| 779 |
-
"learning_rate": 2.8132903717562278e-06,
|
| 780 |
-
"loss": 0.2095,
|
| 781 |
-
"step": 54500
|
| 782 |
-
},
|
| 783 |
-
{
|
| 784 |
-
"epoch": 1.9055538232338982,
|
| 785 |
-
"grad_norm": 1.0492366552352905,
|
| 786 |
-
"learning_rate": 2.3802099573848872e-06,
|
| 787 |
-
"loss": 0.2059,
|
| 788 |
-
"step": 55000
|
| 789 |
-
},
|
| 790 |
-
{
|
| 791 |
-
"epoch": 1.9228770398087516,
|
| 792 |
-
"grad_norm": 1.1293083429336548,
|
| 793 |
-
"learning_rate": 1.947129543013547e-06,
|
| 794 |
-
"loss": 0.2084,
|
| 795 |
-
"step": 55500
|
| 796 |
-
},
|
| 797 |
-
{
|
| 798 |
-
"epoch": 1.9402002563836054,
|
| 799 |
-
"grad_norm": 0.6616119742393494,
|
| 800 |
-
"learning_rate": 1.5140491286422063e-06,
|
| 801 |
-
"loss": 0.2004,
|
| 802 |
-
"step": 56000
|
| 803 |
-
},
|
| 804 |
-
{
|
| 805 |
-
"epoch": 1.957523472958459,
|
| 806 |
-
"grad_norm": 1.435498595237732,
|
| 807 |
-
"learning_rate": 1.0809687142708658e-06,
|
| 808 |
-
"loss": 0.2,
|
| 809 |
-
"step": 56500
|
| 810 |
-
},
|
| 811 |
-
{
|
| 812 |
-
"epoch": 1.9748466895333125,
|
| 813 |
-
"grad_norm": 1.5348039865493774,
|
| 814 |
-
"learning_rate": 6.478882998995254e-07,
|
| 815 |
-
"loss": 0.203,
|
| 816 |
-
"step": 57000
|
| 817 |
-
},
|
| 818 |
-
{
|
| 819 |
-
"epoch": 1.9921699061081661,
|
| 820 |
-
"grad_norm": 1.3262860774993896,
|
| 821 |
-
"learning_rate": 2.1480788552818487e-07,
|
| 822 |
-
"loss": 0.2066,
|
| 823 |
-
"step": 57500
|
| 824 |
-
}
|
| 825 |
-
],
|
| 826 |
-
"logging_steps": 500,
|
| 827 |
-
"max_steps": 57726,
|
| 828 |
-
"num_input_tokens_seen": 0,
|
| 829 |
-
"num_train_epochs": 2,
|
| 830 |
-
"save_steps": 500,
|
| 831 |
-
"stateful_callbacks": {
|
| 832 |
-
"TrainerControl": {
|
| 833 |
-
"args": {
|
| 834 |
-
"should_epoch_stop": false,
|
| 835 |
-
"should_evaluate": false,
|
| 836 |
-
"should_log": false,
|
| 837 |
-
"should_save": true,
|
| 838 |
-
"should_training_stop": false
|
| 839 |
-
},
|
| 840 |
-
"attributes": {}
|
| 841 |
-
}
|
| 842 |
-
},
|
| 843 |
-
"total_flos": 1.2451391598703411e+17,
|
| 844 |
-
"train_batch_size": 32,
|
| 845 |
-
"trial_name": null,
|
| 846 |
-
"trial_params": null
|
| 847 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
checkpoints/checkpoint-57726/trainer_state.json
DELETED
|
@@ -1,847 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"best_global_step": null,
|
| 3 |
-
"best_metric": null,
|
| 4 |
-
"best_model_checkpoint": null,
|
| 5 |
-
"epoch": 2.0,
|
| 6 |
-
"eval_steps": 500,
|
| 7 |
-
"global_step": 57726,
|
| 8 |
-
"is_hyper_param_search": false,
|
| 9 |
-
"is_local_process_zero": true,
|
| 10 |
-
"is_world_process_zero": true,
|
| 11 |
-
"log_history": [
|
| 12 |
-
{
|
| 13 |
-
"epoch": 0.01732321657485362,
|
| 14 |
-
"grad_norm": 0.5299897193908691,
|
| 15 |
-
"learning_rate": 4.956778574645741e-05,
|
| 16 |
-
"loss": 2.1733,
|
| 17 |
-
"step": 500
|
| 18 |
-
},
|
| 19 |
-
{
|
| 20 |
-
"epoch": 0.03464643314970724,
|
| 21 |
-
"grad_norm": 0.4744114279747009,
|
| 22 |
-
"learning_rate": 4.913470533208606e-05,
|
| 23 |
-
"loss": 1.936,
|
| 24 |
-
"step": 1000
|
| 25 |
-
},
|
| 26 |
-
{
|
| 27 |
-
"epoch": 0.051969649724560855,
|
| 28 |
-
"grad_norm": 0.5358501672744751,
|
| 29 |
-
"learning_rate": 4.870162491771472e-05,
|
| 30 |
-
"loss": 1.8605,
|
| 31 |
-
"step": 1500
|
| 32 |
-
},
|
| 33 |
-
{
|
| 34 |
-
"epoch": 0.06929286629941447,
|
| 35 |
-
"grad_norm": 0.5026484131813049,
|
| 36 |
-
"learning_rate": 4.826854450334338e-05,
|
| 37 |
-
"loss": 1.793,
|
| 38 |
-
"step": 2000
|
| 39 |
-
},
|
| 40 |
-
{
|
| 41 |
-
"epoch": 0.0866160828742681,
|
| 42 |
-
"grad_norm": 0.5730611681938171,
|
| 43 |
-
"learning_rate": 4.7835464088972044e-05,
|
| 44 |
-
"loss": 1.7309,
|
| 45 |
-
"step": 2500
|
| 46 |
-
},
|
| 47 |
-
{
|
| 48 |
-
"epoch": 0.10393929944912171,
|
| 49 |
-
"grad_norm": 0.47230127453804016,
|
| 50 |
-
"learning_rate": 4.74023836746007e-05,
|
| 51 |
-
"loss": 1.6938,
|
| 52 |
-
"step": 3000
|
| 53 |
-
},
|
| 54 |
-
{
|
| 55 |
-
"epoch": 0.12126251602397534,
|
| 56 |
-
"grad_norm": 0.6244428157806396,
|
| 57 |
-
"learning_rate": 4.6969303260229364e-05,
|
| 58 |
-
"loss": 1.6587,
|
| 59 |
-
"step": 3500
|
| 60 |
-
},
|
| 61 |
-
{
|
| 62 |
-
"epoch": 0.13858573259882895,
|
| 63 |
-
"grad_norm": 0.5069366693496704,
|
| 64 |
-
"learning_rate": 4.653622284585802e-05,
|
| 65 |
-
"loss": 1.6241,
|
| 66 |
-
"step": 4000
|
| 67 |
-
},
|
| 68 |
-
{
|
| 69 |
-
"epoch": 0.15590894917368256,
|
| 70 |
-
"grad_norm": 0.567051887512207,
|
| 71 |
-
"learning_rate": 4.610314243148668e-05,
|
| 72 |
-
"loss": 1.5789,
|
| 73 |
-
"step": 4500
|
| 74 |
-
},
|
| 75 |
-
{
|
| 76 |
-
"epoch": 0.1732321657485362,
|
| 77 |
-
"grad_norm": 0.5564538836479187,
|
| 78 |
-
"learning_rate": 4.567006201711534e-05,
|
| 79 |
-
"loss": 1.5619,
|
| 80 |
-
"step": 5000
|
| 81 |
-
},
|
| 82 |
-
{
|
| 83 |
-
"epoch": 0.1905553823233898,
|
| 84 |
-
"grad_norm": 0.7243916988372803,
|
| 85 |
-
"learning_rate": 4.5236981602744e-05,
|
| 86 |
-
"loss": 1.5358,
|
| 87 |
-
"step": 5500
|
| 88 |
-
},
|
| 89 |
-
{
|
| 90 |
-
"epoch": 0.20787859889824342,
|
| 91 |
-
"grad_norm": 0.6585371494293213,
|
| 92 |
-
"learning_rate": 4.480390118837266e-05,
|
| 93 |
-
"loss": 1.5396,
|
| 94 |
-
"step": 6000
|
| 95 |
-
},
|
| 96 |
-
{
|
| 97 |
-
"epoch": 0.22520181547309703,
|
| 98 |
-
"grad_norm": 0.5986710786819458,
|
| 99 |
-
"learning_rate": 4.437082077400132e-05,
|
| 100 |
-
"loss": 1.4991,
|
| 101 |
-
"step": 6500
|
| 102 |
-
},
|
| 103 |
-
{
|
| 104 |
-
"epoch": 0.24252503204795067,
|
| 105 |
-
"grad_norm": 0.6701385974884033,
|
| 106 |
-
"learning_rate": 4.393774035962998e-05,
|
| 107 |
-
"loss": 1.4823,
|
| 108 |
-
"step": 7000
|
| 109 |
-
},
|
| 110 |
-
{
|
| 111 |
-
"epoch": 0.25984824862280426,
|
| 112 |
-
"grad_norm": 0.6745362877845764,
|
| 113 |
-
"learning_rate": 4.350465994525863e-05,
|
| 114 |
-
"loss": 1.4715,
|
| 115 |
-
"step": 7500
|
| 116 |
-
},
|
| 117 |
-
{
|
| 118 |
-
"epoch": 0.2771714651976579,
|
| 119 |
-
"grad_norm": 0.6930545568466187,
|
| 120 |
-
"learning_rate": 4.30715795308873e-05,
|
| 121 |
-
"loss": 1.4354,
|
| 122 |
-
"step": 8000
|
| 123 |
-
},
|
| 124 |
-
{
|
| 125 |
-
"epoch": 0.29449468177251154,
|
| 126 |
-
"grad_norm": 0.6593695878982544,
|
| 127 |
-
"learning_rate": 4.2638499116515954e-05,
|
| 128 |
-
"loss": 1.432,
|
| 129 |
-
"step": 8500
|
| 130 |
-
},
|
| 131 |
-
{
|
| 132 |
-
"epoch": 0.3118178983473651,
|
| 133 |
-
"grad_norm": 0.6918864846229553,
|
| 134 |
-
"learning_rate": 4.220541870214462e-05,
|
| 135 |
-
"loss": 1.4209,
|
| 136 |
-
"step": 9000
|
| 137 |
-
},
|
| 138 |
-
{
|
| 139 |
-
"epoch": 0.32914111492221876,
|
| 140 |
-
"grad_norm": 0.7154921889305115,
|
| 141 |
-
"learning_rate": 4.1772338287773275e-05,
|
| 142 |
-
"loss": 1.391,
|
| 143 |
-
"step": 9500
|
| 144 |
-
},
|
| 145 |
-
{
|
| 146 |
-
"epoch": 0.3464643314970724,
|
| 147 |
-
"grad_norm": 0.8634065389633179,
|
| 148 |
-
"learning_rate": 4.133925787340194e-05,
|
| 149 |
-
"loss": 1.3843,
|
| 150 |
-
"step": 10000
|
| 151 |
-
},
|
| 152 |
-
{
|
| 153 |
-
"epoch": 0.363787548071926,
|
| 154 |
-
"grad_norm": 0.8820887804031372,
|
| 155 |
-
"learning_rate": 4.0906177459030595e-05,
|
| 156 |
-
"loss": 1.3612,
|
| 157 |
-
"step": 10500
|
| 158 |
-
},
|
| 159 |
-
{
|
| 160 |
-
"epoch": 0.3811107646467796,
|
| 161 |
-
"grad_norm": 0.9643932580947876,
|
| 162 |
-
"learning_rate": 4.047309704465925e-05,
|
| 163 |
-
"loss": 1.3326,
|
| 164 |
-
"step": 11000
|
| 165 |
-
},
|
| 166 |
-
{
|
| 167 |
-
"epoch": 0.39843398122163326,
|
| 168 |
-
"grad_norm": 0.9170373678207397,
|
| 169 |
-
"learning_rate": 4.0040016630287916e-05,
|
| 170 |
-
"loss": 1.3158,
|
| 171 |
-
"step": 11500
|
| 172 |
-
},
|
| 173 |
-
{
|
| 174 |
-
"epoch": 0.41575719779648684,
|
| 175 |
-
"grad_norm": 0.8911833167076111,
|
| 176 |
-
"learning_rate": 3.960693621591657e-05,
|
| 177 |
-
"loss": 1.2982,
|
| 178 |
-
"step": 12000
|
| 179 |
-
},
|
| 180 |
-
{
|
| 181 |
-
"epoch": 0.4330804143713405,
|
| 182 |
-
"grad_norm": 1.3023812770843506,
|
| 183 |
-
"learning_rate": 3.917472196237398e-05,
|
| 184 |
-
"loss": 1.2497,
|
| 185 |
-
"step": 12500
|
| 186 |
-
},
|
| 187 |
-
{
|
| 188 |
-
"epoch": 0.45040363094619407,
|
| 189 |
-
"grad_norm": 1.3367282152175903,
|
| 190 |
-
"learning_rate": 3.874164154800263e-05,
|
| 191 |
-
"loss": 1.2436,
|
| 192 |
-
"step": 13000
|
| 193 |
-
},
|
| 194 |
-
{
|
| 195 |
-
"epoch": 0.4677268475210477,
|
| 196 |
-
"grad_norm": 1.070709228515625,
|
| 197 |
-
"learning_rate": 3.8308561133631293e-05,
|
| 198 |
-
"loss": 1.2068,
|
| 199 |
-
"step": 13500
|
| 200 |
-
},
|
| 201 |
-
{
|
| 202 |
-
"epoch": 0.48505006409590135,
|
| 203 |
-
"grad_norm": 1.2574782371520996,
|
| 204 |
-
"learning_rate": 3.787548071925995e-05,
|
| 205 |
-
"loss": 1.1867,
|
| 206 |
-
"step": 14000
|
| 207 |
-
},
|
| 208 |
-
{
|
| 209 |
-
"epoch": 0.5023732806707549,
|
| 210 |
-
"grad_norm": 1.4889111518859863,
|
| 211 |
-
"learning_rate": 3.744326646571736e-05,
|
| 212 |
-
"loss": 1.1908,
|
| 213 |
-
"step": 14500
|
| 214 |
-
},
|
| 215 |
-
{
|
| 216 |
-
"epoch": 0.5196964972456085,
|
| 217 |
-
"grad_norm": 1.2556302547454834,
|
| 218 |
-
"learning_rate": 3.7010186051346014e-05,
|
| 219 |
-
"loss": 1.1476,
|
| 220 |
-
"step": 15000
|
| 221 |
-
},
|
| 222 |
-
{
|
| 223 |
-
"epoch": 0.5370197138204622,
|
| 224 |
-
"grad_norm": 1.2582406997680664,
|
| 225 |
-
"learning_rate": 3.6577971797803415e-05,
|
| 226 |
-
"loss": 1.1268,
|
| 227 |
-
"step": 15500
|
| 228 |
-
},
|
| 229 |
-
{
|
| 230 |
-
"epoch": 0.5543429303953158,
|
| 231 |
-
"grad_norm": 1.4035013914108276,
|
| 232 |
-
"learning_rate": 3.614575754426082e-05,
|
| 233 |
-
"loss": 1.119,
|
| 234 |
-
"step": 16000
|
| 235 |
-
},
|
| 236 |
-
{
|
| 237 |
-
"epoch": 0.5716661469701694,
|
| 238 |
-
"grad_norm": 1.7117546796798706,
|
| 239 |
-
"learning_rate": 3.571267712988948e-05,
|
| 240 |
-
"loss": 1.106,
|
| 241 |
-
"step": 16500
|
| 242 |
-
},
|
| 243 |
-
{
|
| 244 |
-
"epoch": 0.5889893635450231,
|
| 245 |
-
"grad_norm": 1.503953218460083,
|
| 246 |
-
"learning_rate": 3.5279596715518135e-05,
|
| 247 |
-
"loss": 1.08,
|
| 248 |
-
"step": 17000
|
| 249 |
-
},
|
| 250 |
-
{
|
| 251 |
-
"epoch": 0.6063125801198767,
|
| 252 |
-
"grad_norm": 1.6753602027893066,
|
| 253 |
-
"learning_rate": 3.48465163011468e-05,
|
| 254 |
-
"loss": 1.055,
|
| 255 |
-
"step": 17500
|
| 256 |
-
},
|
| 257 |
-
{
|
| 258 |
-
"epoch": 0.6236357966947302,
|
| 259 |
-
"grad_norm": 1.5606533288955688,
|
| 260 |
-
"learning_rate": 3.4413435886775456e-05,
|
| 261 |
-
"loss": 1.043,
|
| 262 |
-
"step": 18000
|
| 263 |
-
},
|
| 264 |
-
{
|
| 265 |
-
"epoch": 0.6409590132695839,
|
| 266 |
-
"grad_norm": 2.1966652870178223,
|
| 267 |
-
"learning_rate": 3.398122163323286e-05,
|
| 268 |
-
"loss": 1.0294,
|
| 269 |
-
"step": 18500
|
| 270 |
-
},
|
| 271 |
-
{
|
| 272 |
-
"epoch": 0.6582822298444375,
|
| 273 |
-
"grad_norm": 2.0083296298980713,
|
| 274 |
-
"learning_rate": 3.354814121886152e-05,
|
| 275 |
-
"loss": 1.009,
|
| 276 |
-
"step": 19000
|
| 277 |
-
},
|
| 278 |
-
{
|
| 279 |
-
"epoch": 0.6756054464192911,
|
| 280 |
-
"grad_norm": 1.61182701587677,
|
| 281 |
-
"learning_rate": 3.311506080449018e-05,
|
| 282 |
-
"loss": 0.978,
|
| 283 |
-
"step": 19500
|
| 284 |
-
},
|
| 285 |
-
{
|
| 286 |
-
"epoch": 0.6929286629941448,
|
| 287 |
-
"grad_norm": 1.698030710220337,
|
| 288 |
-
"learning_rate": 3.268198039011884e-05,
|
| 289 |
-
"loss": 0.9798,
|
| 290 |
-
"step": 20000
|
| 291 |
-
},
|
| 292 |
-
{
|
| 293 |
-
"epoch": 0.7102518795689984,
|
| 294 |
-
"grad_norm": 1.810142993927002,
|
| 295 |
-
"learning_rate": 3.224976613657624e-05,
|
| 296 |
-
"loss": 0.9558,
|
| 297 |
-
"step": 20500
|
| 298 |
-
},
|
| 299 |
-
{
|
| 300 |
-
"epoch": 0.727575096143852,
|
| 301 |
-
"grad_norm": 4.462728023529053,
|
| 302 |
-
"learning_rate": 3.1816685722204904e-05,
|
| 303 |
-
"loss": 0.9337,
|
| 304 |
-
"step": 21000
|
| 305 |
-
},
|
| 306 |
-
{
|
| 307 |
-
"epoch": 0.7448983127187057,
|
| 308 |
-
"grad_norm": 3.5566582679748535,
|
| 309 |
-
"learning_rate": 3.1384471468662305e-05,
|
| 310 |
-
"loss": 0.9115,
|
| 311 |
-
"step": 21500
|
| 312 |
-
},
|
| 313 |
-
{
|
| 314 |
-
"epoch": 0.7622215292935592,
|
| 315 |
-
"grad_norm": 3.2824957370758057,
|
| 316 |
-
"learning_rate": 3.095139105429096e-05,
|
| 317 |
-
"loss": 0.8992,
|
| 318 |
-
"step": 22000
|
| 319 |
-
},
|
| 320 |
-
{
|
| 321 |
-
"epoch": 0.7795447458684128,
|
| 322 |
-
"grad_norm": 3.1051690578460693,
|
| 323 |
-
"learning_rate": 3.0518310639919625e-05,
|
| 324 |
-
"loss": 0.8945,
|
| 325 |
-
"step": 22500
|
| 326 |
-
},
|
| 327 |
-
{
|
| 328 |
-
"epoch": 0.7968679624432665,
|
| 329 |
-
"grad_norm": 5.406064510345459,
|
| 330 |
-
"learning_rate": 3.0085230225548282e-05,
|
| 331 |
-
"loss": 0.8544,
|
| 332 |
-
"step": 23000
|
| 333 |
-
},
|
| 334 |
-
{
|
| 335 |
-
"epoch": 0.8141911790181201,
|
| 336 |
-
"grad_norm": 3.3169949054718018,
|
| 337 |
-
"learning_rate": 2.9653015972005682e-05,
|
| 338 |
-
"loss": 0.8494,
|
| 339 |
-
"step": 23500
|
| 340 |
-
},
|
| 341 |
-
{
|
| 342 |
-
"epoch": 0.8315143955929737,
|
| 343 |
-
"grad_norm": 2.4768149852752686,
|
| 344 |
-
"learning_rate": 2.9219935557634343e-05,
|
| 345 |
-
"loss": 0.8145,
|
| 346 |
-
"step": 24000
|
| 347 |
-
},
|
| 348 |
-
{
|
| 349 |
-
"epoch": 0.8488376121678273,
|
| 350 |
-
"grad_norm": 2.6160104274749756,
|
| 351 |
-
"learning_rate": 2.8786855143263003e-05,
|
| 352 |
-
"loss": 0.7995,
|
| 353 |
-
"step": 24500
|
| 354 |
-
},
|
| 355 |
-
{
|
| 356 |
-
"epoch": 0.866160828742681,
|
| 357 |
-
"grad_norm": 2.7828731536865234,
|
| 358 |
-
"learning_rate": 2.8353774728891663e-05,
|
| 359 |
-
"loss": 0.7705,
|
| 360 |
-
"step": 25000
|
| 361 |
-
},
|
| 362 |
-
{
|
| 363 |
-
"epoch": 0.8834840453175345,
|
| 364 |
-
"grad_norm": 3.508859157562256,
|
| 365 |
-
"learning_rate": 2.7921560475349063e-05,
|
| 366 |
-
"loss": 0.7565,
|
| 367 |
-
"step": 25500
|
| 368 |
-
},
|
| 369 |
-
{
|
| 370 |
-
"epoch": 0.9008072618923881,
|
| 371 |
-
"grad_norm": 2.672107219696045,
|
| 372 |
-
"learning_rate": 2.7488480060977724e-05,
|
| 373 |
-
"loss": 0.7394,
|
| 374 |
-
"step": 26000
|
| 375 |
-
},
|
| 376 |
-
{
|
| 377 |
-
"epoch": 0.9181304784672418,
|
| 378 |
-
"grad_norm": 2.712963819503784,
|
| 379 |
-
"learning_rate": 2.7055399646606384e-05,
|
| 380 |
-
"loss": 0.7144,
|
| 381 |
-
"step": 26500
|
| 382 |
-
},
|
| 383 |
-
{
|
| 384 |
-
"epoch": 0.9354536950420954,
|
| 385 |
-
"grad_norm": 4.505606651306152,
|
| 386 |
-
"learning_rate": 2.6622319232235044e-05,
|
| 387 |
-
"loss": 0.6964,
|
| 388 |
-
"step": 27000
|
| 389 |
-
},
|
| 390 |
-
{
|
| 391 |
-
"epoch": 0.952776911616949,
|
| 392 |
-
"grad_norm": 1.9717707633972168,
|
| 393 |
-
"learning_rate": 2.619010497869244e-05,
|
| 394 |
-
"loss": 0.6642,
|
| 395 |
-
"step": 27500
|
| 396 |
-
},
|
| 397 |
-
{
|
| 398 |
-
"epoch": 0.9701001281918027,
|
| 399 |
-
"grad_norm": 2.501847743988037,
|
| 400 |
-
"learning_rate": 2.57570245643211e-05,
|
| 401 |
-
"loss": 0.6542,
|
| 402 |
-
"step": 28000
|
| 403 |
-
},
|
| 404 |
-
{
|
| 405 |
-
"epoch": 0.9874233447666563,
|
| 406 |
-
"grad_norm": 4.659865379333496,
|
| 407 |
-
"learning_rate": 2.532481031077851e-05,
|
| 408 |
-
"loss": 0.622,
|
| 409 |
-
"step": 28500
|
| 410 |
-
},
|
| 411 |
-
{
|
| 412 |
-
"epoch": 1.0,
|
| 413 |
-
"eval_loss": 0.3914269506931305,
|
| 414 |
-
"eval_runtime": 1756.0029,
|
| 415 |
-
"eval_samples_per_second": 350.648,
|
| 416 |
-
"eval_steps_per_second": 43.831,
|
| 417 |
-
"step": 28863
|
| 418 |
-
},
|
| 419 |
-
{
|
| 420 |
-
"epoch": 1.0047465613415099,
|
| 421 |
-
"grad_norm": 1.5512120723724365,
|
| 422 |
-
"learning_rate": 2.4891729896407165e-05,
|
| 423 |
-
"loss": 0.6,
|
| 424 |
-
"step": 29000
|
| 425 |
-
},
|
| 426 |
-
{
|
| 427 |
-
"epoch": 1.0220697779163634,
|
| 428 |
-
"grad_norm": 2.7224926948547363,
|
| 429 |
-
"learning_rate": 2.4458649482035826e-05,
|
| 430 |
-
"loss": 0.5733,
|
| 431 |
-
"step": 29500
|
| 432 |
-
},
|
| 433 |
-
{
|
| 434 |
-
"epoch": 1.039392994491217,
|
| 435 |
-
"grad_norm": 2.0893383026123047,
|
| 436 |
-
"learning_rate": 2.4025569067664486e-05,
|
| 437 |
-
"loss": 0.5517,
|
| 438 |
-
"step": 30000
|
| 439 |
-
},
|
| 440 |
-
{
|
| 441 |
-
"epoch": 1.0567162110660708,
|
| 442 |
-
"grad_norm": 2.235480785369873,
|
| 443 |
-
"learning_rate": 2.3592488653293143e-05,
|
| 444 |
-
"loss": 0.5389,
|
| 445 |
-
"step": 30500
|
| 446 |
-
},
|
| 447 |
-
{
|
| 448 |
-
"epoch": 1.0740394276409244,
|
| 449 |
-
"grad_norm": 2.4926114082336426,
|
| 450 |
-
"learning_rate": 2.3159408238921803e-05,
|
| 451 |
-
"loss": 0.5282,
|
| 452 |
-
"step": 31000
|
| 453 |
-
},
|
| 454 |
-
{
|
| 455 |
-
"epoch": 1.091362644215778,
|
| 456 |
-
"grad_norm": 1.9329782724380493,
|
| 457 |
-
"learning_rate": 2.2726327824550464e-05,
|
| 458 |
-
"loss": 0.5076,
|
| 459 |
-
"step": 31500
|
| 460 |
-
},
|
| 461 |
-
{
|
| 462 |
-
"epoch": 1.1086858607906316,
|
| 463 |
-
"grad_norm": 1.8705673217773438,
|
| 464 |
-
"learning_rate": 2.2293247410179124e-05,
|
| 465 |
-
"loss": 0.489,
|
| 466 |
-
"step": 32000
|
| 467 |
-
},
|
| 468 |
-
{
|
| 469 |
-
"epoch": 1.1260090773654852,
|
| 470 |
-
"grad_norm": 1.8372488021850586,
|
| 471 |
-
"learning_rate": 2.186016699580778e-05,
|
| 472 |
-
"loss": 0.4701,
|
| 473 |
-
"step": 32500
|
| 474 |
-
},
|
| 475 |
-
{
|
| 476 |
-
"epoch": 1.1433322939403388,
|
| 477 |
-
"grad_norm": 1.7131813764572144,
|
| 478 |
-
"learning_rate": 2.142708658143644e-05,
|
| 479 |
-
"loss": 0.4592,
|
| 480 |
-
"step": 33000
|
| 481 |
-
},
|
| 482 |
-
{
|
| 483 |
-
"epoch": 1.1606555105151926,
|
| 484 |
-
"grad_norm": 3.1449172496795654,
|
| 485 |
-
"learning_rate": 2.0994872327893845e-05,
|
| 486 |
-
"loss": 0.4408,
|
| 487 |
-
"step": 33500
|
| 488 |
-
},
|
| 489 |
-
{
|
| 490 |
-
"epoch": 1.1779787270900461,
|
| 491 |
-
"grad_norm": 2.199096918106079,
|
| 492 |
-
"learning_rate": 2.0561791913522505e-05,
|
| 493 |
-
"loss": 0.4278,
|
| 494 |
-
"step": 34000
|
| 495 |
-
},
|
| 496 |
-
{
|
| 497 |
-
"epoch": 1.1953019436648997,
|
| 498 |
-
"grad_norm": 1.6359717845916748,
|
| 499 |
-
"learning_rate": 2.0128711499151162e-05,
|
| 500 |
-
"loss": 0.4238,
|
| 501 |
-
"step": 34500
|
| 502 |
-
},
|
| 503 |
-
{
|
| 504 |
-
"epoch": 1.2126251602397533,
|
| 505 |
-
"grad_norm": 1.7747199535369873,
|
| 506 |
-
"learning_rate": 1.9695631084779822e-05,
|
| 507 |
-
"loss": 0.4024,
|
| 508 |
-
"step": 35000
|
| 509 |
-
},
|
| 510 |
-
{
|
| 511 |
-
"epoch": 1.229948376814607,
|
| 512 |
-
"grad_norm": 2.073904037475586,
|
| 513 |
-
"learning_rate": 1.9263416831237226e-05,
|
| 514 |
-
"loss": 0.39,
|
| 515 |
-
"step": 35500
|
| 516 |
-
},
|
| 517 |
-
{
|
| 518 |
-
"epoch": 1.2472715933894605,
|
| 519 |
-
"grad_norm": 1.9012521505355835,
|
| 520 |
-
"learning_rate": 1.8830336416865886e-05,
|
| 521 |
-
"loss": 0.3856,
|
| 522 |
-
"step": 36000
|
| 523 |
-
},
|
| 524 |
-
{
|
| 525 |
-
"epoch": 1.264594809964314,
|
| 526 |
-
"grad_norm": 2.2919228076934814,
|
| 527 |
-
"learning_rate": 1.8397256002494546e-05,
|
| 528 |
-
"loss": 0.3729,
|
| 529 |
-
"step": 36500
|
| 530 |
-
},
|
| 531 |
-
{
|
| 532 |
-
"epoch": 1.2819180265391679,
|
| 533 |
-
"grad_norm": 1.933583378791809,
|
| 534 |
-
"learning_rate": 1.7964175588123203e-05,
|
| 535 |
-
"loss": 0.3584,
|
| 536 |
-
"step": 37000
|
| 537 |
-
},
|
| 538 |
-
{
|
| 539 |
-
"epoch": 1.2992412431140214,
|
| 540 |
-
"grad_norm": 2.1168253421783447,
|
| 541 |
-
"learning_rate": 1.7531095173751864e-05,
|
| 542 |
-
"loss": 0.356,
|
| 543 |
-
"step": 37500
|
| 544 |
-
},
|
| 545 |
-
{
|
| 546 |
-
"epoch": 1.316564459688875,
|
| 547 |
-
"grad_norm": 1.7472686767578125,
|
| 548 |
-
"learning_rate": 1.7098014759380524e-05,
|
| 549 |
-
"loss": 0.339,
|
| 550 |
-
"step": 38000
|
| 551 |
-
},
|
| 552 |
-
{
|
| 553 |
-
"epoch": 1.3338876762637286,
|
| 554 |
-
"grad_norm": 1.9178825616836548,
|
| 555 |
-
"learning_rate": 1.666493434500918e-05,
|
| 556 |
-
"loss": 0.3294,
|
| 557 |
-
"step": 38500
|
| 558 |
-
},
|
| 559 |
-
{
|
| 560 |
-
"epoch": 1.3512108928385822,
|
| 561 |
-
"grad_norm": 1.4763795137405396,
|
| 562 |
-
"learning_rate": 1.623185393063784e-05,
|
| 563 |
-
"loss": 0.3299,
|
| 564 |
-
"step": 39000
|
| 565 |
-
},
|
| 566 |
-
{
|
| 567 |
-
"epoch": 1.368534109413436,
|
| 568 |
-
"grad_norm": 1.974442481994629,
|
| 569 |
-
"learning_rate": 1.57987735162665e-05,
|
| 570 |
-
"loss": 0.3206,
|
| 571 |
-
"step": 39500
|
| 572 |
-
},
|
| 573 |
-
{
|
| 574 |
-
"epoch": 1.3858573259882894,
|
| 575 |
-
"grad_norm": 1.315453052520752,
|
| 576 |
-
"learning_rate": 1.5366559262723905e-05,
|
| 577 |
-
"loss": 0.313,
|
| 578 |
-
"step": 40000
|
| 579 |
-
},
|
| 580 |
-
{
|
| 581 |
-
"epoch": 1.4031805425631432,
|
| 582 |
-
"grad_norm": 1.605042815208435,
|
| 583 |
-
"learning_rate": 1.4934345009181305e-05,
|
| 584 |
-
"loss": 0.2998,
|
| 585 |
-
"step": 40500
|
| 586 |
-
},
|
| 587 |
-
{
|
| 588 |
-
"epoch": 1.4205037591379968,
|
| 589 |
-
"grad_norm": 1.5163339376449585,
|
| 590 |
-
"learning_rate": 1.4501264594809966e-05,
|
| 591 |
-
"loss": 0.3031,
|
| 592 |
-
"step": 41000
|
| 593 |
-
},
|
| 594 |
-
{
|
| 595 |
-
"epoch": 1.4378269757128503,
|
| 596 |
-
"grad_norm": 1.9643980264663696,
|
| 597 |
-
"learning_rate": 1.4068184180438623e-05,
|
| 598 |
-
"loss": 0.2922,
|
| 599 |
-
"step": 41500
|
| 600 |
-
},
|
| 601 |
-
{
|
| 602 |
-
"epoch": 1.455150192287704,
|
| 603 |
-
"grad_norm": 1.4320045709609985,
|
| 604 |
-
"learning_rate": 1.3635103766067284e-05,
|
| 605 |
-
"loss": 0.2858,
|
| 606 |
-
"step": 42000
|
| 607 |
-
},
|
| 608 |
-
{
|
| 609 |
-
"epoch": 1.4724734088625575,
|
| 610 |
-
"grad_norm": 1.4415267705917358,
|
| 611 |
-
"learning_rate": 1.3202889512524685e-05,
|
| 612 |
-
"loss": 0.2807,
|
| 613 |
-
"step": 42500
|
| 614 |
-
},
|
| 615 |
-
{
|
| 616 |
-
"epoch": 1.4897966254374113,
|
| 617 |
-
"grad_norm": 1.1676790714263916,
|
| 618 |
-
"learning_rate": 1.2769809098153345e-05,
|
| 619 |
-
"loss": 0.279,
|
| 620 |
-
"step": 43000
|
| 621 |
-
},
|
| 622 |
-
{
|
| 623 |
-
"epoch": 1.5071198420122647,
|
| 624 |
-
"grad_norm": 1.2914575338363647,
|
| 625 |
-
"learning_rate": 1.2336728683782005e-05,
|
| 626 |
-
"loss": 0.2658,
|
| 627 |
-
"step": 43500
|
| 628 |
-
},
|
| 629 |
-
{
|
| 630 |
-
"epoch": 1.5244430585871185,
|
| 631 |
-
"grad_norm": 1.8342537879943848,
|
| 632 |
-
"learning_rate": 1.1903648269410666e-05,
|
| 633 |
-
"loss": 0.2669,
|
| 634 |
-
"step": 44000
|
| 635 |
-
},
|
| 636 |
-
{
|
| 637 |
-
"epoch": 1.541766275161972,
|
| 638 |
-
"grad_norm": 1.6250979900360107,
|
| 639 |
-
"learning_rate": 1.1471434015868068e-05,
|
| 640 |
-
"loss": 0.2615,
|
| 641 |
-
"step": 44500
|
| 642 |
-
},
|
| 643 |
-
{
|
| 644 |
-
"epoch": 1.5590894917368257,
|
| 645 |
-
"grad_norm": 1.8935564756393433,
|
| 646 |
-
"learning_rate": 1.1038353601496726e-05,
|
| 647 |
-
"loss": 0.2513,
|
| 648 |
-
"step": 45000
|
| 649 |
-
},
|
| 650 |
-
{
|
| 651 |
-
"epoch": 1.5764127083116795,
|
| 652 |
-
"grad_norm": 1.7190736532211304,
|
| 653 |
-
"learning_rate": 1.0605273187125386e-05,
|
| 654 |
-
"loss": 0.2488,
|
| 655 |
-
"step": 45500
|
| 656 |
-
},
|
| 657 |
-
{
|
| 658 |
-
"epoch": 1.5937359248865328,
|
| 659 |
-
"grad_norm": 1.5352853536605835,
|
| 660 |
-
"learning_rate": 1.0172192772754045e-05,
|
| 661 |
-
"loss": 0.2502,
|
| 662 |
-
"step": 46000
|
| 663 |
-
},
|
| 664 |
-
{
|
| 665 |
-
"epoch": 1.6110591414613866,
|
| 666 |
-
"grad_norm": 1.452185869216919,
|
| 667 |
-
"learning_rate": 9.739112358382704e-06,
|
| 668 |
-
"loss": 0.2441,
|
| 669 |
-
"step": 46500
|
| 670 |
-
},
|
| 671 |
-
{
|
| 672 |
-
"epoch": 1.6283823580362402,
|
| 673 |
-
"grad_norm": 1.2263524532318115,
|
| 674 |
-
"learning_rate": 9.306031944011364e-06,
|
| 675 |
-
"loss": 0.2366,
|
| 676 |
-
"step": 47000
|
| 677 |
-
},
|
| 678 |
-
{
|
| 679 |
-
"epoch": 1.6457055746110938,
|
| 680 |
-
"grad_norm": 1.7002058029174805,
|
| 681 |
-
"learning_rate": 8.872951529640024e-06,
|
| 682 |
-
"loss": 0.2361,
|
| 683 |
-
"step": 47500
|
| 684 |
-
},
|
| 685 |
-
{
|
| 686 |
-
"epoch": 1.6630287911859474,
|
| 687 |
-
"grad_norm": 0.8775345087051392,
|
| 688 |
-
"learning_rate": 8.439871115268685e-06,
|
| 689 |
-
"loss": 0.2314,
|
| 690 |
-
"step": 48000
|
| 691 |
-
},
|
| 692 |
-
{
|
| 693 |
-
"epoch": 1.680352007760801,
|
| 694 |
-
"grad_norm": 1.68264639377594,
|
| 695 |
-
"learning_rate": 8.007656861726086e-06,
|
| 696 |
-
"loss": 0.2341,
|
| 697 |
-
"step": 48500
|
| 698 |
-
},
|
| 699 |
-
{
|
| 700 |
-
"epoch": 1.6976752243356548,
|
| 701 |
-
"grad_norm": 1.4811515808105469,
|
| 702 |
-
"learning_rate": 7.574576447354745e-06,
|
| 703 |
-
"loss": 0.2238,
|
| 704 |
-
"step": 49000
|
| 705 |
-
},
|
| 706 |
-
{
|
| 707 |
-
"epoch": 1.7149984409105081,
|
| 708 |
-
"grad_norm": 1.5231512784957886,
|
| 709 |
-
"learning_rate": 7.142362193812148e-06,
|
| 710 |
-
"loss": 0.2229,
|
| 711 |
-
"step": 49500
|
| 712 |
-
},
|
| 713 |
-
{
|
| 714 |
-
"epoch": 1.732321657485362,
|
| 715 |
-
"grad_norm": 1.870348334312439,
|
| 716 |
-
"learning_rate": 6.709281779440807e-06,
|
| 717 |
-
"loss": 0.2211,
|
| 718 |
-
"step": 50000
|
| 719 |
-
},
|
| 720 |
-
{
|
| 721 |
-
"epoch": 1.7496448740602155,
|
| 722 |
-
"grad_norm": 1.2438400983810425,
|
| 723 |
-
"learning_rate": 6.276201365069466e-06,
|
| 724 |
-
"loss": 0.2205,
|
| 725 |
-
"step": 50500
|
| 726 |
-
},
|
| 727 |
-
{
|
| 728 |
-
"epoch": 1.766968090635069,
|
| 729 |
-
"grad_norm": 1.426330327987671,
|
| 730 |
-
"learning_rate": 5.843120950698125e-06,
|
| 731 |
-
"loss": 0.2189,
|
| 732 |
-
"step": 51000
|
| 733 |
-
},
|
| 734 |
-
{
|
| 735 |
-
"epoch": 1.7842913072099227,
|
| 736 |
-
"grad_norm": 0.9976692199707031,
|
| 737 |
-
"learning_rate": 5.410040536326786e-06,
|
| 738 |
-
"loss": 0.2168,
|
| 739 |
-
"step": 51500
|
| 740 |
-
},
|
| 741 |
-
{
|
| 742 |
-
"epoch": 1.8016145237847763,
|
| 743 |
-
"grad_norm": 0.7628383040428162,
|
| 744 |
-
"learning_rate": 4.976960121955445e-06,
|
| 745 |
-
"loss": 0.2137,
|
| 746 |
-
"step": 52000
|
| 747 |
-
},
|
| 748 |
-
{
|
| 749 |
-
"epoch": 1.81893774035963,
|
| 750 |
-
"grad_norm": 1.1943764686584473,
|
| 751 |
-
"learning_rate": 4.543879707584105e-06,
|
| 752 |
-
"loss": 0.2108,
|
| 753 |
-
"step": 52500
|
| 754 |
-
},
|
| 755 |
-
{
|
| 756 |
-
"epoch": 1.8362609569344834,
|
| 757 |
-
"grad_norm": 1.6423786878585815,
|
| 758 |
-
"learning_rate": 4.110799293212764e-06,
|
| 759 |
-
"loss": 0.207,
|
| 760 |
-
"step": 53000
|
| 761 |
-
},
|
| 762 |
-
{
|
| 763 |
-
"epoch": 1.8535841735093372,
|
| 764 |
-
"grad_norm": 1.1439303159713745,
|
| 765 |
-
"learning_rate": 3.678585039670166e-06,
|
| 766 |
-
"loss": 0.2102,
|
| 767 |
-
"step": 53500
|
| 768 |
-
},
|
| 769 |
-
{
|
| 770 |
-
"epoch": 1.8709073900841908,
|
| 771 |
-
"grad_norm": 1.1228655576705933,
|
| 772 |
-
"learning_rate": 3.2463707861275687e-06,
|
| 773 |
-
"loss": 0.2078,
|
| 774 |
-
"step": 54000
|
| 775 |
-
},
|
| 776 |
-
{
|
| 777 |
-
"epoch": 1.8882306066590444,
|
| 778 |
-
"grad_norm": 1.6855107545852661,
|
| 779 |
-
"learning_rate": 2.8132903717562278e-06,
|
| 780 |
-
"loss": 0.2095,
|
| 781 |
-
"step": 54500
|
| 782 |
-
},
|
| 783 |
-
{
|
| 784 |
-
"epoch": 1.9055538232338982,
|
| 785 |
-
"grad_norm": 1.0492366552352905,
|
| 786 |
-
"learning_rate": 2.3802099573848872e-06,
|
| 787 |
-
"loss": 0.2059,
|
| 788 |
-
"step": 55000
|
| 789 |
-
},
|
| 790 |
-
{
|
| 791 |
-
"epoch": 1.9228770398087516,
|
| 792 |
-
"grad_norm": 1.1293083429336548,
|
| 793 |
-
"learning_rate": 1.947129543013547e-06,
|
| 794 |
-
"loss": 0.2084,
|
| 795 |
-
"step": 55500
|
| 796 |
-
},
|
| 797 |
-
{
|
| 798 |
-
"epoch": 1.9402002563836054,
|
| 799 |
-
"grad_norm": 0.6616119742393494,
|
| 800 |
-
"learning_rate": 1.5140491286422063e-06,
|
| 801 |
-
"loss": 0.2004,
|
| 802 |
-
"step": 56000
|
| 803 |
-
},
|
| 804 |
-
{
|
| 805 |
-
"epoch": 1.957523472958459,
|
| 806 |
-
"grad_norm": 1.435498595237732,
|
| 807 |
-
"learning_rate": 1.0809687142708658e-06,
|
| 808 |
-
"loss": 0.2,
|
| 809 |
-
"step": 56500
|
| 810 |
-
},
|
| 811 |
-
{
|
| 812 |
-
"epoch": 1.9748466895333125,
|
| 813 |
-
"grad_norm": 1.5348039865493774,
|
| 814 |
-
"learning_rate": 6.478882998995254e-07,
|
| 815 |
-
"loss": 0.203,
|
| 816 |
-
"step": 57000
|
| 817 |
-
},
|
| 818 |
-
{
|
| 819 |
-
"epoch": 1.9921699061081661,
|
| 820 |
-
"grad_norm": 1.3262860774993896,
|
| 821 |
-
"learning_rate": 2.1480788552818487e-07,
|
| 822 |
-
"loss": 0.2066,
|
| 823 |
-
"step": 57500
|
| 824 |
-
}
|
| 825 |
-
],
|
| 826 |
-
"logging_steps": 500,
|
| 827 |
-
"max_steps": 57726,
|
| 828 |
-
"num_input_tokens_seen": 0,
|
| 829 |
-
"num_train_epochs": 2,
|
| 830 |
-
"save_steps": 500,
|
| 831 |
-
"stateful_callbacks": {
|
| 832 |
-
"TrainerControl": {
|
| 833 |
-
"args": {
|
| 834 |
-
"should_epoch_stop": false,
|
| 835 |
-
"should_evaluate": false,
|
| 836 |
-
"should_log": false,
|
| 837 |
-
"should_save": true,
|
| 838 |
-
"should_training_stop": true
|
| 839 |
-
},
|
| 840 |
-
"attributes": {}
|
| 841 |
-
}
|
| 842 |
-
},
|
| 843 |
-
"total_flos": 1.2500277057395098e+17,
|
| 844 |
-
"train_batch_size": 32,
|
| 845 |
-
"trial_name": null,
|
| 846 |
-
"trial_params": null
|
| 847 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
checkpoints/{checkpoint-57000 → checkpoint-61500}/config.json
RENAMED
|
File without changes
|
checkpoints/{checkpoint-57000 → checkpoint-61500}/generation_config.json
RENAMED
|
File without changes
|
checkpoints/{checkpoint-57000 → checkpoint-61500}/model.safetensors
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 258368552
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:83c04cb0e96a506016467863a6b4674ff881a6422d2034e2295fbf658914a398
|
| 3 |
size 258368552
|
checkpoints/{checkpoint-57500 → checkpoint-61500}/optimizer.pt
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 516816826
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bec2188f9ea536c328c3af0b020aed0f233ddedcafae07ec2d0d3e2ee0120d24
|
| 3 |
size 516816826
|
checkpoints/{checkpoint-57726 → checkpoint-61500}/rng_state.pth
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a47ee310263d5d6300401264828ccfb4fec9c1c4181961358106a0f617b86d6f
|
| 3 |
size 14244
|
checkpoints/{checkpoint-57726 → checkpoint-61500}/scaler.pt
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 988
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c521173cdf16365f5a532f1fd1c976b664d848a2f833b7b603db15856db89a60
|
| 3 |
size 988
|
checkpoints/{checkpoint-57000 → checkpoint-61500}/scheduler.pt
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5f1363f9495d21091d229315dab764d88aa8a5a850ff84e8ec9dd37b5a8c9d6f
|
| 3 |
size 1064
|
checkpoints/{checkpoint-57000 → checkpoint-61500}/special_tokens_map.json
RENAMED
|
File without changes
|
checkpoints/{checkpoint-57500 → checkpoint-61500}/spiece.model
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:610d79d092a886e9af7d31b94512c46fd20a5f62557bd5b8d8a5b23f1f78650a
|
| 3 |
+
size 1042971
|
checkpoints/{checkpoint-57000 → checkpoint-61500}/tokenizer.json
RENAMED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoints/{checkpoint-57000 → checkpoint-61500}/tokenizer_config.json
RENAMED
|
File without changes
|
checkpoints/checkpoint-61500/trainer_state.json
ADDED
|
@@ -0,0 +1,903 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 1.9785892383173813,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 61500,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.016086222150727903,
|
| 14 |
+
"grad_norm": 0.5078127384185791,
|
| 15 |
+
"learning_rate": 4.959864230101023e-05,
|
| 16 |
+
"loss": 2.1432,
|
| 17 |
+
"step": 500
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.032172444301455806,
|
| 21 |
+
"grad_norm": 0.4508506655693054,
|
| 22 |
+
"learning_rate": 4.9196480277974395e-05,
|
| 23 |
+
"loss": 1.9093,
|
| 24 |
+
"step": 1000
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.048258666452183706,
|
| 28 |
+
"grad_norm": 0.4430558979511261,
|
| 29 |
+
"learning_rate": 4.879431825493855e-05,
|
| 30 |
+
"loss": 1.8418,
|
| 31 |
+
"step": 1500
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.06434488860291161,
|
| 35 |
+
"grad_norm": 0.4775325059890747,
|
| 36 |
+
"learning_rate": 4.8392156231902713e-05,
|
| 37 |
+
"loss": 1.7771,
|
| 38 |
+
"step": 2000
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.08043111075363951,
|
| 42 |
+
"grad_norm": 0.49685001373291016,
|
| 43 |
+
"learning_rate": 4.7989994208866876e-05,
|
| 44 |
+
"loss": 1.7226,
|
| 45 |
+
"step": 2500
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.09651733290436741,
|
| 49 |
+
"grad_norm": 0.5552434325218201,
|
| 50 |
+
"learning_rate": 4.7587832185831025e-05,
|
| 51 |
+
"loss": 1.6767,
|
| 52 |
+
"step": 3000
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.11260355505509531,
|
| 56 |
+
"grad_norm": 0.6779139637947083,
|
| 57 |
+
"learning_rate": 4.718567016279519e-05,
|
| 58 |
+
"loss": 1.6588,
|
| 59 |
+
"step": 3500
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.12868977720582322,
|
| 63 |
+
"grad_norm": 0.5552022457122803,
|
| 64 |
+
"learning_rate": 4.6783508139759344e-05,
|
| 65 |
+
"loss": 1.603,
|
| 66 |
+
"step": 4000
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.1447759993565511,
|
| 70 |
+
"grad_norm": 0.5302042365074158,
|
| 71 |
+
"learning_rate": 4.638134611672351e-05,
|
| 72 |
+
"loss": 1.5776,
|
| 73 |
+
"step": 4500
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.16086222150727902,
|
| 77 |
+
"grad_norm": 0.5810815691947937,
|
| 78 |
+
"learning_rate": 4.597918409368766e-05,
|
| 79 |
+
"loss": 1.5333,
|
| 80 |
+
"step": 5000
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.1769484436580069,
|
| 84 |
+
"grad_norm": 0.5819700956344604,
|
| 85 |
+
"learning_rate": 4.5577022070651826e-05,
|
| 86 |
+
"loss": 1.5168,
|
| 87 |
+
"step": 5500
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.19303466580873482,
|
| 91 |
+
"grad_norm": 0.6134072542190552,
|
| 92 |
+
"learning_rate": 4.517486004761599e-05,
|
| 93 |
+
"loss": 1.4748,
|
| 94 |
+
"step": 6000
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.2091208879594627,
|
| 98 |
+
"grad_norm": 0.5746152400970459,
|
| 99 |
+
"learning_rate": 4.4772698024580144e-05,
|
| 100 |
+
"loss": 1.4622,
|
| 101 |
+
"step": 6500
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.22520711011019062,
|
| 105 |
+
"grad_norm": 0.7663710713386536,
|
| 106 |
+
"learning_rate": 4.437053600154431e-05,
|
| 107 |
+
"loss": 1.4767,
|
| 108 |
+
"step": 7000
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.24129333226091854,
|
| 112 |
+
"grad_norm": 0.7993176579475403,
|
| 113 |
+
"learning_rate": 4.396837397850846e-05,
|
| 114 |
+
"loss": 1.4527,
|
| 115 |
+
"step": 7500
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.25737955441164645,
|
| 119 |
+
"grad_norm": 0.6892676949501038,
|
| 120 |
+
"learning_rate": 4.3566211955472626e-05,
|
| 121 |
+
"loss": 1.4325,
|
| 122 |
+
"step": 8000
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.2734657765623743,
|
| 126 |
+
"grad_norm": 0.6928556561470032,
|
| 127 |
+
"learning_rate": 4.316404993243678e-05,
|
| 128 |
+
"loss": 1.4038,
|
| 129 |
+
"step": 8500
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.2895519987131022,
|
| 133 |
+
"grad_norm": 0.7578593492507935,
|
| 134 |
+
"learning_rate": 4.2761887909400944e-05,
|
| 135 |
+
"loss": 1.3945,
|
| 136 |
+
"step": 9000
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.30563822086383013,
|
| 140 |
+
"grad_norm": 0.7504703402519226,
|
| 141 |
+
"learning_rate": 4.23597258863651e-05,
|
| 142 |
+
"loss": 1.3644,
|
| 143 |
+
"step": 9500
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.32172444301455805,
|
| 147 |
+
"grad_norm": 0.8370710611343384,
|
| 148 |
+
"learning_rate": 4.1957563863329256e-05,
|
| 149 |
+
"loss": 1.3619,
|
| 150 |
+
"step": 10000
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.3378106651652859,
|
| 154 |
+
"grad_norm": 0.8501142263412476,
|
| 155 |
+
"learning_rate": 4.155540184029342e-05,
|
| 156 |
+
"loss": 1.3448,
|
| 157 |
+
"step": 10500
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.3538968873160138,
|
| 161 |
+
"grad_norm": 0.9001900553703308,
|
| 162 |
+
"learning_rate": 4.1153239817257575e-05,
|
| 163 |
+
"loss": 1.3004,
|
| 164 |
+
"step": 11000
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.36998310946674173,
|
| 168 |
+
"grad_norm": 1.0658681392669678,
|
| 169 |
+
"learning_rate": 4.075107779422174e-05,
|
| 170 |
+
"loss": 1.2789,
|
| 171 |
+
"step": 11500
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.38606933161746965,
|
| 175 |
+
"grad_norm": 1.1038371324539185,
|
| 176 |
+
"learning_rate": 4.0348915771185894e-05,
|
| 177 |
+
"loss": 1.2651,
|
| 178 |
+
"step": 12000
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.40215555376819756,
|
| 182 |
+
"grad_norm": 1.2004213333129883,
|
| 183 |
+
"learning_rate": 3.994755807219613e-05,
|
| 184 |
+
"loss": 1.2216,
|
| 185 |
+
"step": 12500
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.4182417759189254,
|
| 189 |
+
"grad_norm": 1.235543966293335,
|
| 190 |
+
"learning_rate": 3.9545396049160286e-05,
|
| 191 |
+
"loss": 1.1955,
|
| 192 |
+
"step": 13000
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.43432799806965333,
|
| 196 |
+
"grad_norm": 1.5088828802108765,
|
| 197 |
+
"learning_rate": 3.914323402612445e-05,
|
| 198 |
+
"loss": 1.1836,
|
| 199 |
+
"step": 13500
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.45041422022038125,
|
| 203 |
+
"grad_norm": 1.264153242111206,
|
| 204 |
+
"learning_rate": 3.8741072003088605e-05,
|
| 205 |
+
"loss": 1.1658,
|
| 206 |
+
"step": 14000
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.46650044237110916,
|
| 210 |
+
"grad_norm": 1.3023343086242676,
|
| 211 |
+
"learning_rate": 3.833971430409884e-05,
|
| 212 |
+
"loss": 1.1481,
|
| 213 |
+
"step": 14500
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.48258666452183707,
|
| 217 |
+
"grad_norm": 1.3824670314788818,
|
| 218 |
+
"learning_rate": 3.7938356605109064e-05,
|
| 219 |
+
"loss": 1.1221,
|
| 220 |
+
"step": 15000
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.49867288667256493,
|
| 224 |
+
"grad_norm": 1.4364969730377197,
|
| 225 |
+
"learning_rate": 3.75369989061193e-05,
|
| 226 |
+
"loss": 1.1057,
|
| 227 |
+
"step": 15500
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.5147591088232929,
|
| 231 |
+
"grad_norm": 2.051701545715332,
|
| 232 |
+
"learning_rate": 3.7134836883083456e-05,
|
| 233 |
+
"loss": 1.0873,
|
| 234 |
+
"step": 16000
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.5308453309740208,
|
| 238 |
+
"grad_norm": 1.4329720735549927,
|
| 239 |
+
"learning_rate": 3.673267486004762e-05,
|
| 240 |
+
"loss": 1.0607,
|
| 241 |
+
"step": 16500
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.5469315531247486,
|
| 245 |
+
"grad_norm": 1.4981014728546143,
|
| 246 |
+
"learning_rate": 3.6330512837011775e-05,
|
| 247 |
+
"loss": 1.0516,
|
| 248 |
+
"step": 17000
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.5630177752754766,
|
| 252 |
+
"grad_norm": 1.3012079000473022,
|
| 253 |
+
"learning_rate": 3.592835081397594e-05,
|
| 254 |
+
"loss": 1.0317,
|
| 255 |
+
"step": 17500
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.5791039974262044,
|
| 259 |
+
"grad_norm": 1.401825189590454,
|
| 260 |
+
"learning_rate": 3.552699311498617e-05,
|
| 261 |
+
"loss": 1.0183,
|
| 262 |
+
"step": 18000
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.5951902195769324,
|
| 266 |
+
"grad_norm": 2.0783369541168213,
|
| 267 |
+
"learning_rate": 3.512483109195033e-05,
|
| 268 |
+
"loss": 0.9985,
|
| 269 |
+
"step": 18500
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.6112764417276603,
|
| 273 |
+
"grad_norm": 2.3940794467926025,
|
| 274 |
+
"learning_rate": 3.4722669068914486e-05,
|
| 275 |
+
"loss": 0.9698,
|
| 276 |
+
"step": 19000
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.6273626638783881,
|
| 280 |
+
"grad_norm": 1.4747998714447021,
|
| 281 |
+
"learning_rate": 3.432050704587865e-05,
|
| 282 |
+
"loss": 0.9657,
|
| 283 |
+
"step": 19500
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.6434488860291161,
|
| 287 |
+
"grad_norm": 3.0782012939453125,
|
| 288 |
+
"learning_rate": 3.391914934688888e-05,
|
| 289 |
+
"loss": 0.9379,
|
| 290 |
+
"step": 20000
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.659535108179844,
|
| 294 |
+
"grad_norm": 2.4914307594299316,
|
| 295 |
+
"learning_rate": 3.3516987323853034e-05,
|
| 296 |
+
"loss": 0.915,
|
| 297 |
+
"step": 20500
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.6756213303305718,
|
| 301 |
+
"grad_norm": 2.772120237350464,
|
| 302 |
+
"learning_rate": 3.3115629624863264e-05,
|
| 303 |
+
"loss": 0.9047,
|
| 304 |
+
"step": 21000
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 0.6917075524812998,
|
| 308 |
+
"grad_norm": 2.519575595855713,
|
| 309 |
+
"learning_rate": 3.271346760182743e-05,
|
| 310 |
+
"loss": 0.8688,
|
| 311 |
+
"step": 21500
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.7077937746320276,
|
| 315 |
+
"grad_norm": 4.085098743438721,
|
| 316 |
+
"learning_rate": 3.231130557879158e-05,
|
| 317 |
+
"loss": 0.8581,
|
| 318 |
+
"step": 22000
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.7238799967827556,
|
| 322 |
+
"grad_norm": 1.4670002460479736,
|
| 323 |
+
"learning_rate": 3.1909143555755745e-05,
|
| 324 |
+
"loss": 0.8354,
|
| 325 |
+
"step": 22500
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.7399662189334835,
|
| 329 |
+
"grad_norm": 2.4749488830566406,
|
| 330 |
+
"learning_rate": 3.1507785856765975e-05,
|
| 331 |
+
"loss": 0.8108,
|
| 332 |
+
"step": 23000
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.7560524410842113,
|
| 336 |
+
"grad_norm": 1.8635029792785645,
|
| 337 |
+
"learning_rate": 3.110562383373014e-05,
|
| 338 |
+
"loss": 0.7773,
|
| 339 |
+
"step": 23500
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.7721386632349393,
|
| 343 |
+
"grad_norm": 3.5713748931884766,
|
| 344 |
+
"learning_rate": 3.0703461810694294e-05,
|
| 345 |
+
"loss": 0.756,
|
| 346 |
+
"step": 24000
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.7882248853856672,
|
| 350 |
+
"grad_norm": 1.8903526067733765,
|
| 351 |
+
"learning_rate": 3.0301299787658456e-05,
|
| 352 |
+
"loss": 0.7326,
|
| 353 |
+
"step": 24500
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.8043111075363951,
|
| 357 |
+
"grad_norm": 8.286703109741211,
|
| 358 |
+
"learning_rate": 2.9899942088668686e-05,
|
| 359 |
+
"loss": 0.6948,
|
| 360 |
+
"step": 25000
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.820397329687123,
|
| 364 |
+
"grad_norm": 2.2209272384643555,
|
| 365 |
+
"learning_rate": 2.9497780065632845e-05,
|
| 366 |
+
"loss": 0.6914,
|
| 367 |
+
"step": 25500
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.8364835518378508,
|
| 371 |
+
"grad_norm": 2.2284536361694336,
|
| 372 |
+
"learning_rate": 2.9095618042597e-05,
|
| 373 |
+
"loss": 0.6585,
|
| 374 |
+
"step": 26000
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.8525697739885788,
|
| 378 |
+
"grad_norm": 3.4615938663482666,
|
| 379 |
+
"learning_rate": 2.869345601956116e-05,
|
| 380 |
+
"loss": 0.633,
|
| 381 |
+
"step": 26500
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.8686559961393067,
|
| 385 |
+
"grad_norm": 3.1158838272094727,
|
| 386 |
+
"learning_rate": 2.829209832057139e-05,
|
| 387 |
+
"loss": 0.6181,
|
| 388 |
+
"step": 27000
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.8847422182900346,
|
| 392 |
+
"grad_norm": 2.3320417404174805,
|
| 393 |
+
"learning_rate": 2.7889936297535553e-05,
|
| 394 |
+
"loss": 0.5993,
|
| 395 |
+
"step": 27500
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.9008284404407625,
|
| 399 |
+
"grad_norm": 1.8331427574157715,
|
| 400 |
+
"learning_rate": 2.7487774274499712e-05,
|
| 401 |
+
"loss": 0.5839,
|
| 402 |
+
"step": 28000
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.9169146625914903,
|
| 406 |
+
"grad_norm": 3.2398369312286377,
|
| 407 |
+
"learning_rate": 2.708561225146387e-05,
|
| 408 |
+
"loss": 0.562,
|
| 409 |
+
"step": 28500
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.9330008847422183,
|
| 413 |
+
"grad_norm": 1.6575061082839966,
|
| 414 |
+
"learning_rate": 2.66842545524741e-05,
|
| 415 |
+
"loss": 0.5313,
|
| 416 |
+
"step": 29000
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.9490871068929462,
|
| 420 |
+
"grad_norm": 2.1604230403900146,
|
| 421 |
+
"learning_rate": 2.6282092529438264e-05,
|
| 422 |
+
"loss": 0.5203,
|
| 423 |
+
"step": 29500
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.9651733290436741,
|
| 427 |
+
"grad_norm": 3.3743808269500732,
|
| 428 |
+
"learning_rate": 2.5879930506402423e-05,
|
| 429 |
+
"loss": 0.4938,
|
| 430 |
+
"step": 30000
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 0.981259551194402,
|
| 434 |
+
"grad_norm": 3.766514301300049,
|
| 435 |
+
"learning_rate": 2.5477768483366583e-05,
|
| 436 |
+
"loss": 0.4724,
|
| 437 |
+
"step": 30500
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 0.9973457733451299,
|
| 441 |
+
"grad_norm": 2.26712703704834,
|
| 442 |
+
"learning_rate": 2.5075606460330742e-05,
|
| 443 |
+
"loss": 0.4656,
|
| 444 |
+
"step": 31000
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 1.0,
|
| 448 |
+
"eval_loss": 0.26554691791534424,
|
| 449 |
+
"eval_runtime": 1917.4803,
|
| 450 |
+
"eval_samples_per_second": 345.81,
|
| 451 |
+
"eval_steps_per_second": 43.227,
|
| 452 |
+
"step": 31083
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 1.0134159092737072,
|
| 456 |
+
"grad_norm": 2.1041958332061768,
|
| 457 |
+
"learning_rate": 2.467424876134097e-05,
|
| 458 |
+
"loss": 0.4381,
|
| 459 |
+
"step": 31500
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 1.029502131424435,
|
| 463 |
+
"grad_norm": 1.7629106044769287,
|
| 464 |
+
"learning_rate": 2.427208673830513e-05,
|
| 465 |
+
"loss": 0.4298,
|
| 466 |
+
"step": 32000
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 1.0455883535751629,
|
| 470 |
+
"grad_norm": 2.5032904148101807,
|
| 471 |
+
"learning_rate": 2.386992471526929e-05,
|
| 472 |
+
"loss": 0.4188,
|
| 473 |
+
"step": 32500
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 1.0616745757258907,
|
| 477 |
+
"grad_norm": 1.6467881202697754,
|
| 478 |
+
"learning_rate": 2.3467762692233446e-05,
|
| 479 |
+
"loss": 0.3986,
|
| 480 |
+
"step": 33000
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"epoch": 1.0777607978766186,
|
| 484 |
+
"grad_norm": 1.957220435142517,
|
| 485 |
+
"learning_rate": 2.3065600669197606e-05,
|
| 486 |
+
"loss": 0.382,
|
| 487 |
+
"step": 33500
|
| 488 |
+
},
|
| 489 |
+
{
|
| 490 |
+
"epoch": 1.0938470200273467,
|
| 491 |
+
"grad_norm": 1.6566946506500244,
|
| 492 |
+
"learning_rate": 2.2663438646161765e-05,
|
| 493 |
+
"loss": 0.3689,
|
| 494 |
+
"step": 34000
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 1.1099332421780745,
|
| 498 |
+
"grad_norm": 2.081613540649414,
|
| 499 |
+
"learning_rate": 2.2261276623125928e-05,
|
| 500 |
+
"loss": 0.3603,
|
| 501 |
+
"step": 34500
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 1.1260194643288024,
|
| 505 |
+
"grad_norm": 2.155226945877075,
|
| 506 |
+
"learning_rate": 2.1859918924136157e-05,
|
| 507 |
+
"loss": 0.3478,
|
| 508 |
+
"step": 35000
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 1.1421056864795303,
|
| 512 |
+
"grad_norm": 1.9459590911865234,
|
| 513 |
+
"learning_rate": 2.1457756901100317e-05,
|
| 514 |
+
"loss": 0.3315,
|
| 515 |
+
"step": 35500
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 1.1581919086302581,
|
| 519 |
+
"grad_norm": 2.3381567001342773,
|
| 520 |
+
"learning_rate": 2.1055594878064476e-05,
|
| 521 |
+
"loss": 0.3259,
|
| 522 |
+
"step": 36000
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 1.1742781307809862,
|
| 526 |
+
"grad_norm": 1.4302254915237427,
|
| 527 |
+
"learning_rate": 2.0653432855028635e-05,
|
| 528 |
+
"loss": 0.3168,
|
| 529 |
+
"step": 36500
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 1.190364352931714,
|
| 533 |
+
"grad_norm": 1.1770597696304321,
|
| 534 |
+
"learning_rate": 2.0251270831992795e-05,
|
| 535 |
+
"loss": 0.3082,
|
| 536 |
+
"step": 37000
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 1.206450575082442,
|
| 540 |
+
"grad_norm": 1.7475298643112183,
|
| 541 |
+
"learning_rate": 1.9849913133003024e-05,
|
| 542 |
+
"loss": 0.3014,
|
| 543 |
+
"step": 37500
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 1.2225367972331698,
|
| 547 |
+
"grad_norm": 1.2397468090057373,
|
| 548 |
+
"learning_rate": 1.9447751109967187e-05,
|
| 549 |
+
"loss": 0.288,
|
| 550 |
+
"step": 38000
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 1.2386230193838976,
|
| 554 |
+
"grad_norm": 1.6603740453720093,
|
| 555 |
+
"learning_rate": 1.9045589086931343e-05,
|
| 556 |
+
"loss": 0.2797,
|
| 557 |
+
"step": 38500
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 1.2547092415346257,
|
| 561 |
+
"grad_norm": 1.7009538412094116,
|
| 562 |
+
"learning_rate": 1.8643427063895502e-05,
|
| 563 |
+
"loss": 0.275,
|
| 564 |
+
"step": 39000
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 1.2707954636853536,
|
| 568 |
+
"grad_norm": 1.4941717386245728,
|
| 569 |
+
"learning_rate": 1.8241265040859662e-05,
|
| 570 |
+
"loss": 0.2623,
|
| 571 |
+
"step": 39500
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 1.2868816858360814,
|
| 575 |
+
"grad_norm": 1.941115140914917,
|
| 576 |
+
"learning_rate": 1.7839907341869895e-05,
|
| 577 |
+
"loss": 0.2572,
|
| 578 |
+
"step": 40000
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 1.3029679079868093,
|
| 582 |
+
"grad_norm": 1.487726092338562,
|
| 583 |
+
"learning_rate": 1.7437745318834054e-05,
|
| 584 |
+
"loss": 0.2502,
|
| 585 |
+
"step": 40500
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 1.3190541301375371,
|
| 589 |
+
"grad_norm": 1.4628674983978271,
|
| 590 |
+
"learning_rate": 1.7035583295798213e-05,
|
| 591 |
+
"loss": 0.2437,
|
| 592 |
+
"step": 41000
|
| 593 |
+
},
|
| 594 |
+
{
|
| 595 |
+
"epoch": 1.3351403522882652,
|
| 596 |
+
"grad_norm": 1.401607632637024,
|
| 597 |
+
"learning_rate": 1.663342127276237e-05,
|
| 598 |
+
"loss": 0.2421,
|
| 599 |
+
"step": 41500
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 1.351226574438993,
|
| 603 |
+
"grad_norm": 1.1497563123703003,
|
| 604 |
+
"learning_rate": 1.623125924972653e-05,
|
| 605 |
+
"loss": 0.231,
|
| 606 |
+
"step": 42000
|
| 607 |
+
},
|
| 608 |
+
{
|
| 609 |
+
"epoch": 1.367312796589721,
|
| 610 |
+
"grad_norm": 1.322836995124817,
|
| 611 |
+
"learning_rate": 1.5829097226690688e-05,
|
| 612 |
+
"loss": 0.2261,
|
| 613 |
+
"step": 42500
|
| 614 |
+
},
|
| 615 |
+
{
|
| 616 |
+
"epoch": 1.3833990187404488,
|
| 617 |
+
"grad_norm": 1.5328525304794312,
|
| 618 |
+
"learning_rate": 1.542773952770092e-05,
|
| 619 |
+
"loss": 0.2177,
|
| 620 |
+
"step": 43000
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 1.3994852408911767,
|
| 624 |
+
"grad_norm": 1.7748241424560547,
|
| 625 |
+
"learning_rate": 1.502557750466508e-05,
|
| 626 |
+
"loss": 0.2186,
|
| 627 |
+
"step": 43500
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 1.4155714630419047,
|
| 631 |
+
"grad_norm": 1.6542141437530518,
|
| 632 |
+
"learning_rate": 1.4623415481629241e-05,
|
| 633 |
+
"loss": 0.2138,
|
| 634 |
+
"step": 44000
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 1.4316576851926326,
|
| 638 |
+
"grad_norm": 1.3098843097686768,
|
| 639 |
+
"learning_rate": 1.4221253458593397e-05,
|
| 640 |
+
"loss": 0.211,
|
| 641 |
+
"step": 44500
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 1.4477439073433604,
|
| 645 |
+
"grad_norm": 1.345651626586914,
|
| 646 |
+
"learning_rate": 1.3819091435557557e-05,
|
| 647 |
+
"loss": 0.2027,
|
| 648 |
+
"step": 45000
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 1.4638301294940883,
|
| 652 |
+
"grad_norm": 1.4520297050476074,
|
| 653 |
+
"learning_rate": 1.3416929412521718e-05,
|
| 654 |
+
"loss": 0.2039,
|
| 655 |
+
"step": 45500
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 1.4799163516448162,
|
| 659 |
+
"grad_norm": 1.5913499593734741,
|
| 660 |
+
"learning_rate": 1.3014767389485877e-05,
|
| 661 |
+
"loss": 0.1939,
|
| 662 |
+
"step": 46000
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 1.4960025737955442,
|
| 666 |
+
"grad_norm": 1.1803226470947266,
|
| 667 |
+
"learning_rate": 1.2612605366450037e-05,
|
| 668 |
+
"loss": 0.1887,
|
| 669 |
+
"step": 46500
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 1.5120887959462719,
|
| 673 |
+
"grad_norm": 1.1462236642837524,
|
| 674 |
+
"learning_rate": 1.2210443343414194e-05,
|
| 675 |
+
"loss": 0.1883,
|
| 676 |
+
"step": 47000
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"epoch": 1.528175018097,
|
| 680 |
+
"grad_norm": 0.8483968377113342,
|
| 681 |
+
"learning_rate": 1.1808281320378355e-05,
|
| 682 |
+
"loss": 0.1809,
|
| 683 |
+
"step": 47500
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 1.5442612402477278,
|
| 687 |
+
"grad_norm": 1.1205823421478271,
|
| 688 |
+
"learning_rate": 1.1406119297342515e-05,
|
| 689 |
+
"loss": 0.1813,
|
| 690 |
+
"step": 48000
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"epoch": 1.5603474623984557,
|
| 694 |
+
"grad_norm": 1.417622447013855,
|
| 695 |
+
"learning_rate": 1.1003957274306672e-05,
|
| 696 |
+
"loss": 0.1788,
|
| 697 |
+
"step": 48500
|
| 698 |
+
},
|
| 699 |
+
{
|
| 700 |
+
"epoch": 1.5764336845491838,
|
| 701 |
+
"grad_norm": 1.179103970527649,
|
| 702 |
+
"learning_rate": 1.0602599575316904e-05,
|
| 703 |
+
"loss": 0.1809,
|
| 704 |
+
"step": 49000
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 1.5925199066999114,
|
| 708 |
+
"grad_norm": 1.1092889308929443,
|
| 709 |
+
"learning_rate": 1.0200437552281065e-05,
|
| 710 |
+
"loss": 0.1734,
|
| 711 |
+
"step": 49500
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 1.6086061288506395,
|
| 715 |
+
"grad_norm": 1.0196574926376343,
|
| 716 |
+
"learning_rate": 9.798275529245222e-06,
|
| 717 |
+
"loss": 0.1688,
|
| 718 |
+
"step": 50000
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 1.6246923510013673,
|
| 722 |
+
"grad_norm": 1.1376862525939941,
|
| 723 |
+
"learning_rate": 9.396113506209382e-06,
|
| 724 |
+
"loss": 0.1703,
|
| 725 |
+
"step": 50500
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 1.6407785731520952,
|
| 729 |
+
"grad_norm": 0.8885149955749512,
|
| 730 |
+
"learning_rate": 8.995560131265685e-06,
|
| 731 |
+
"loss": 0.1691,
|
| 732 |
+
"step": 51000
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 1.6568647953028233,
|
| 736 |
+
"grad_norm": 1.2574944496154785,
|
| 737 |
+
"learning_rate": 8.593398108229844e-06,
|
| 738 |
+
"loss": 0.1615,
|
| 739 |
+
"step": 51500
|
| 740 |
+
},
|
| 741 |
+
{
|
| 742 |
+
"epoch": 1.672951017453551,
|
| 743 |
+
"grad_norm": 1.2620723247528076,
|
| 744 |
+
"learning_rate": 8.191236085194004e-06,
|
| 745 |
+
"loss": 0.1593,
|
| 746 |
+
"step": 52000
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 1.689037239604279,
|
| 750 |
+
"grad_norm": 1.551480770111084,
|
| 751 |
+
"learning_rate": 7.789074062158163e-06,
|
| 752 |
+
"loss": 0.1639,
|
| 753 |
+
"step": 52500
|
| 754 |
+
},
|
| 755 |
+
{
|
| 756 |
+
"epoch": 1.7051234617550068,
|
| 757 |
+
"grad_norm": 1.5938962697982788,
|
| 758 |
+
"learning_rate": 7.386912039122322e-06,
|
| 759 |
+
"loss": 0.1587,
|
| 760 |
+
"step": 53000
|
| 761 |
+
},
|
| 762 |
+
{
|
| 763 |
+
"epoch": 1.7212096839057347,
|
| 764 |
+
"grad_norm": 1.0503953695297241,
|
| 765 |
+
"learning_rate": 6.984750016086482e-06,
|
| 766 |
+
"loss": 0.1599,
|
| 767 |
+
"step": 53500
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 1.7372959060564628,
|
| 771 |
+
"grad_norm": 1.1205036640167236,
|
| 772 |
+
"learning_rate": 6.583392317096712e-06,
|
| 773 |
+
"loss": 0.1541,
|
| 774 |
+
"step": 54000
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 1.7533821282071904,
|
| 778 |
+
"grad_norm": 0.7524433732032776,
|
| 779 |
+
"learning_rate": 6.181230294060872e-06,
|
| 780 |
+
"loss": 0.1521,
|
| 781 |
+
"step": 54500
|
| 782 |
+
},
|
| 783 |
+
{
|
| 784 |
+
"epoch": 1.7694683503579185,
|
| 785 |
+
"grad_norm": 0.9619775414466858,
|
| 786 |
+
"learning_rate": 5.779068271025031e-06,
|
| 787 |
+
"loss": 0.1521,
|
| 788 |
+
"step": 55000
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 1.7855545725086464,
|
| 792 |
+
"grad_norm": 0.9406844973564148,
|
| 793 |
+
"learning_rate": 5.37690624798919e-06,
|
| 794 |
+
"loss": 0.1509,
|
| 795 |
+
"step": 55500
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 1.8016407946593742,
|
| 799 |
+
"grad_norm": 0.9363726377487183,
|
| 800 |
+
"learning_rate": 4.975548548999421e-06,
|
| 801 |
+
"loss": 0.1513,
|
| 802 |
+
"step": 56000
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 1.8177270168101023,
|
| 806 |
+
"grad_norm": 0.9941402673721313,
|
| 807 |
+
"learning_rate": 4.573386525963581e-06,
|
| 808 |
+
"loss": 0.1484,
|
| 809 |
+
"step": 56500
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 1.83381323896083,
|
| 813 |
+
"grad_norm": 1.3756345510482788,
|
| 814 |
+
"learning_rate": 4.17122450292774e-06,
|
| 815 |
+
"loss": 0.1509,
|
| 816 |
+
"step": 57000
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 1.849899461111558,
|
| 820 |
+
"grad_norm": 1.0644595623016357,
|
| 821 |
+
"learning_rate": 3.7690624798918986e-06,
|
| 822 |
+
"loss": 0.1486,
|
| 823 |
+
"step": 57500
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"epoch": 1.8659856832622859,
|
| 827 |
+
"grad_norm": 1.070890188217163,
|
| 828 |
+
"learning_rate": 3.3669004568560584e-06,
|
| 829 |
+
"loss": 0.1462,
|
| 830 |
+
"step": 58000
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 1.8820719054130137,
|
| 834 |
+
"grad_norm": 1.3034768104553223,
|
| 835 |
+
"learning_rate": 2.9647384338202173e-06,
|
| 836 |
+
"loss": 0.1481,
|
| 837 |
+
"step": 58500
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 1.8981581275637418,
|
| 841 |
+
"grad_norm": 1.127517580986023,
|
| 842 |
+
"learning_rate": 2.5625764107843767e-06,
|
| 843 |
+
"loss": 0.1451,
|
| 844 |
+
"step": 59000
|
| 845 |
+
},
|
| 846 |
+
{
|
| 847 |
+
"epoch": 1.9142443497144694,
|
| 848 |
+
"grad_norm": 0.9431403279304504,
|
| 849 |
+
"learning_rate": 2.1604143877485364e-06,
|
| 850 |
+
"loss": 0.1458,
|
| 851 |
+
"step": 59500
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 1.9303305718651975,
|
| 855 |
+
"grad_norm": 1.271483302116394,
|
| 856 |
+
"learning_rate": 1.7590566887587673e-06,
|
| 857 |
+
"loss": 0.1463,
|
| 858 |
+
"step": 60000
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 1.9464167940159254,
|
| 862 |
+
"grad_norm": 0.7327952980995178,
|
| 863 |
+
"learning_rate": 1.3568946657229264e-06,
|
| 864 |
+
"loss": 0.1434,
|
| 865 |
+
"step": 60500
|
| 866 |
+
},
|
| 867 |
+
{
|
| 868 |
+
"epoch": 1.9625030161666532,
|
| 869 |
+
"grad_norm": 1.0670543909072876,
|
| 870 |
+
"learning_rate": 9.547326426870858e-07,
|
| 871 |
+
"loss": 0.1424,
|
| 872 |
+
"step": 61000
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 1.9785892383173813,
|
| 876 |
+
"grad_norm": 1.2705425024032593,
|
| 877 |
+
"learning_rate": 5.525706196512451e-07,
|
| 878 |
+
"loss": 0.1431,
|
| 879 |
+
"step": 61500
|
| 880 |
+
}
|
| 881 |
+
],
|
| 882 |
+
"logging_steps": 500,
|
| 883 |
+
"max_steps": 62164,
|
| 884 |
+
"num_input_tokens_seen": 0,
|
| 885 |
+
"num_train_epochs": 2,
|
| 886 |
+
"save_steps": 500,
|
| 887 |
+
"stateful_callbacks": {
|
| 888 |
+
"TrainerControl": {
|
| 889 |
+
"args": {
|
| 890 |
+
"should_epoch_stop": false,
|
| 891 |
+
"should_evaluate": false,
|
| 892 |
+
"should_log": false,
|
| 893 |
+
"should_save": true,
|
| 894 |
+
"should_training_stop": false
|
| 895 |
+
},
|
| 896 |
+
"attributes": {}
|
| 897 |
+
}
|
| 898 |
+
},
|
| 899 |
+
"total_flos": 1.3317420101507482e+17,
|
| 900 |
+
"train_batch_size": 32,
|
| 901 |
+
"trial_name": null,
|
| 902 |
+
"trial_params": null
|
| 903 |
+
}
|
checkpoints/{checkpoint-57000 → checkpoint-61500}/training_args.bin
RENAMED
|
File without changes
|
checkpoints/{checkpoint-57500 → checkpoint-62000}/config.json
RENAMED
|
File without changes
|
checkpoints/{checkpoint-57500 → checkpoint-62000}/generation_config.json
RENAMED
|
File without changes
|
checkpoints/{checkpoint-57500 → checkpoint-62000}/model.safetensors
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 258368552
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2d634b9fb199c6f1740da63ed5e87bdff20b813c9bbfbdb4c438dd27d4ab4a43
|
| 3 |
size 258368552
|
checkpoints/{checkpoint-57000 → checkpoint-62000}/optimizer.pt
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 516816826
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:92792cccad0cffd7b1d855e6c4fcd33ba25e1aa7b12ae985fc20d5c190b15298
|
| 3 |
size 516816826
|
checkpoints/{checkpoint-57500 → checkpoint-62000}/rng_state.pth
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:82cb63ee008746fe97e2f989e0faf59793321720b17b5d1bb583c8252b44d2dd
|
| 3 |
size 14244
|
checkpoints/{checkpoint-57500 → checkpoint-62000}/scaler.pt
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 988
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b157befbc8b431a674c196f6e4cac76d2908e26ebf928dd38c5fd8e80ccd7dfb
|
| 3 |
size 988
|
checkpoints/{checkpoint-57726 → checkpoint-62000}/scheduler.pt
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:82a779fd4c5c40ea80ef0219096be07750b466b66cb885a86a58a8d3e8a2829d
|
| 3 |
size 1064
|
checkpoints/{checkpoint-57500 → checkpoint-62000}/special_tokens_map.json
RENAMED
|
File without changes
|
checkpoints/{checkpoint-57000 → checkpoint-62000}/spiece.model
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:610d79d092a886e9af7d31b94512c46fd20a5f62557bd5b8d8a5b23f1f78650a
|
| 3 |
+
size 1042971
|
checkpoints/{checkpoint-57500 → checkpoint-62000}/tokenizer.json
RENAMED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoints/{checkpoint-57500 → checkpoint-62000}/tokenizer_config.json
RENAMED
|
File without changes
|
checkpoints/checkpoint-62000/trainer_state.json
ADDED
|
@@ -0,0 +1,910 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 1.994675460468109,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 62000,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.016086222150727903,
|
| 14 |
+
"grad_norm": 0.5078127384185791,
|
| 15 |
+
"learning_rate": 4.959864230101023e-05,
|
| 16 |
+
"loss": 2.1432,
|
| 17 |
+
"step": 500
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.032172444301455806,
|
| 21 |
+
"grad_norm": 0.4508506655693054,
|
| 22 |
+
"learning_rate": 4.9196480277974395e-05,
|
| 23 |
+
"loss": 1.9093,
|
| 24 |
+
"step": 1000
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.048258666452183706,
|
| 28 |
+
"grad_norm": 0.4430558979511261,
|
| 29 |
+
"learning_rate": 4.879431825493855e-05,
|
| 30 |
+
"loss": 1.8418,
|
| 31 |
+
"step": 1500
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.06434488860291161,
|
| 35 |
+
"grad_norm": 0.4775325059890747,
|
| 36 |
+
"learning_rate": 4.8392156231902713e-05,
|
| 37 |
+
"loss": 1.7771,
|
| 38 |
+
"step": 2000
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.08043111075363951,
|
| 42 |
+
"grad_norm": 0.49685001373291016,
|
| 43 |
+
"learning_rate": 4.7989994208866876e-05,
|
| 44 |
+
"loss": 1.7226,
|
| 45 |
+
"step": 2500
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.09651733290436741,
|
| 49 |
+
"grad_norm": 0.5552434325218201,
|
| 50 |
+
"learning_rate": 4.7587832185831025e-05,
|
| 51 |
+
"loss": 1.6767,
|
| 52 |
+
"step": 3000
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.11260355505509531,
|
| 56 |
+
"grad_norm": 0.6779139637947083,
|
| 57 |
+
"learning_rate": 4.718567016279519e-05,
|
| 58 |
+
"loss": 1.6588,
|
| 59 |
+
"step": 3500
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.12868977720582322,
|
| 63 |
+
"grad_norm": 0.5552022457122803,
|
| 64 |
+
"learning_rate": 4.6783508139759344e-05,
|
| 65 |
+
"loss": 1.603,
|
| 66 |
+
"step": 4000
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.1447759993565511,
|
| 70 |
+
"grad_norm": 0.5302042365074158,
|
| 71 |
+
"learning_rate": 4.638134611672351e-05,
|
| 72 |
+
"loss": 1.5776,
|
| 73 |
+
"step": 4500
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.16086222150727902,
|
| 77 |
+
"grad_norm": 0.5810815691947937,
|
| 78 |
+
"learning_rate": 4.597918409368766e-05,
|
| 79 |
+
"loss": 1.5333,
|
| 80 |
+
"step": 5000
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.1769484436580069,
|
| 84 |
+
"grad_norm": 0.5819700956344604,
|
| 85 |
+
"learning_rate": 4.5577022070651826e-05,
|
| 86 |
+
"loss": 1.5168,
|
| 87 |
+
"step": 5500
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.19303466580873482,
|
| 91 |
+
"grad_norm": 0.6134072542190552,
|
| 92 |
+
"learning_rate": 4.517486004761599e-05,
|
| 93 |
+
"loss": 1.4748,
|
| 94 |
+
"step": 6000
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.2091208879594627,
|
| 98 |
+
"grad_norm": 0.5746152400970459,
|
| 99 |
+
"learning_rate": 4.4772698024580144e-05,
|
| 100 |
+
"loss": 1.4622,
|
| 101 |
+
"step": 6500
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.22520711011019062,
|
| 105 |
+
"grad_norm": 0.7663710713386536,
|
| 106 |
+
"learning_rate": 4.437053600154431e-05,
|
| 107 |
+
"loss": 1.4767,
|
| 108 |
+
"step": 7000
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.24129333226091854,
|
| 112 |
+
"grad_norm": 0.7993176579475403,
|
| 113 |
+
"learning_rate": 4.396837397850846e-05,
|
| 114 |
+
"loss": 1.4527,
|
| 115 |
+
"step": 7500
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.25737955441164645,
|
| 119 |
+
"grad_norm": 0.6892676949501038,
|
| 120 |
+
"learning_rate": 4.3566211955472626e-05,
|
| 121 |
+
"loss": 1.4325,
|
| 122 |
+
"step": 8000
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.2734657765623743,
|
| 126 |
+
"grad_norm": 0.6928556561470032,
|
| 127 |
+
"learning_rate": 4.316404993243678e-05,
|
| 128 |
+
"loss": 1.4038,
|
| 129 |
+
"step": 8500
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.2895519987131022,
|
| 133 |
+
"grad_norm": 0.7578593492507935,
|
| 134 |
+
"learning_rate": 4.2761887909400944e-05,
|
| 135 |
+
"loss": 1.3945,
|
| 136 |
+
"step": 9000
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.30563822086383013,
|
| 140 |
+
"grad_norm": 0.7504703402519226,
|
| 141 |
+
"learning_rate": 4.23597258863651e-05,
|
| 142 |
+
"loss": 1.3644,
|
| 143 |
+
"step": 9500
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.32172444301455805,
|
| 147 |
+
"grad_norm": 0.8370710611343384,
|
| 148 |
+
"learning_rate": 4.1957563863329256e-05,
|
| 149 |
+
"loss": 1.3619,
|
| 150 |
+
"step": 10000
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.3378106651652859,
|
| 154 |
+
"grad_norm": 0.8501142263412476,
|
| 155 |
+
"learning_rate": 4.155540184029342e-05,
|
| 156 |
+
"loss": 1.3448,
|
| 157 |
+
"step": 10500
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.3538968873160138,
|
| 161 |
+
"grad_norm": 0.9001900553703308,
|
| 162 |
+
"learning_rate": 4.1153239817257575e-05,
|
| 163 |
+
"loss": 1.3004,
|
| 164 |
+
"step": 11000
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.36998310946674173,
|
| 168 |
+
"grad_norm": 1.0658681392669678,
|
| 169 |
+
"learning_rate": 4.075107779422174e-05,
|
| 170 |
+
"loss": 1.2789,
|
| 171 |
+
"step": 11500
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.38606933161746965,
|
| 175 |
+
"grad_norm": 1.1038371324539185,
|
| 176 |
+
"learning_rate": 4.0348915771185894e-05,
|
| 177 |
+
"loss": 1.2651,
|
| 178 |
+
"step": 12000
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.40215555376819756,
|
| 182 |
+
"grad_norm": 1.2004213333129883,
|
| 183 |
+
"learning_rate": 3.994755807219613e-05,
|
| 184 |
+
"loss": 1.2216,
|
| 185 |
+
"step": 12500
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.4182417759189254,
|
| 189 |
+
"grad_norm": 1.235543966293335,
|
| 190 |
+
"learning_rate": 3.9545396049160286e-05,
|
| 191 |
+
"loss": 1.1955,
|
| 192 |
+
"step": 13000
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.43432799806965333,
|
| 196 |
+
"grad_norm": 1.5088828802108765,
|
| 197 |
+
"learning_rate": 3.914323402612445e-05,
|
| 198 |
+
"loss": 1.1836,
|
| 199 |
+
"step": 13500
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.45041422022038125,
|
| 203 |
+
"grad_norm": 1.264153242111206,
|
| 204 |
+
"learning_rate": 3.8741072003088605e-05,
|
| 205 |
+
"loss": 1.1658,
|
| 206 |
+
"step": 14000
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.46650044237110916,
|
| 210 |
+
"grad_norm": 1.3023343086242676,
|
| 211 |
+
"learning_rate": 3.833971430409884e-05,
|
| 212 |
+
"loss": 1.1481,
|
| 213 |
+
"step": 14500
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.48258666452183707,
|
| 217 |
+
"grad_norm": 1.3824670314788818,
|
| 218 |
+
"learning_rate": 3.7938356605109064e-05,
|
| 219 |
+
"loss": 1.1221,
|
| 220 |
+
"step": 15000
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.49867288667256493,
|
| 224 |
+
"grad_norm": 1.4364969730377197,
|
| 225 |
+
"learning_rate": 3.75369989061193e-05,
|
| 226 |
+
"loss": 1.1057,
|
| 227 |
+
"step": 15500
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.5147591088232929,
|
| 231 |
+
"grad_norm": 2.051701545715332,
|
| 232 |
+
"learning_rate": 3.7134836883083456e-05,
|
| 233 |
+
"loss": 1.0873,
|
| 234 |
+
"step": 16000
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.5308453309740208,
|
| 238 |
+
"grad_norm": 1.4329720735549927,
|
| 239 |
+
"learning_rate": 3.673267486004762e-05,
|
| 240 |
+
"loss": 1.0607,
|
| 241 |
+
"step": 16500
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.5469315531247486,
|
| 245 |
+
"grad_norm": 1.4981014728546143,
|
| 246 |
+
"learning_rate": 3.6330512837011775e-05,
|
| 247 |
+
"loss": 1.0516,
|
| 248 |
+
"step": 17000
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.5630177752754766,
|
| 252 |
+
"grad_norm": 1.3012079000473022,
|
| 253 |
+
"learning_rate": 3.592835081397594e-05,
|
| 254 |
+
"loss": 1.0317,
|
| 255 |
+
"step": 17500
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.5791039974262044,
|
| 259 |
+
"grad_norm": 1.401825189590454,
|
| 260 |
+
"learning_rate": 3.552699311498617e-05,
|
| 261 |
+
"loss": 1.0183,
|
| 262 |
+
"step": 18000
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.5951902195769324,
|
| 266 |
+
"grad_norm": 2.0783369541168213,
|
| 267 |
+
"learning_rate": 3.512483109195033e-05,
|
| 268 |
+
"loss": 0.9985,
|
| 269 |
+
"step": 18500
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.6112764417276603,
|
| 273 |
+
"grad_norm": 2.3940794467926025,
|
| 274 |
+
"learning_rate": 3.4722669068914486e-05,
|
| 275 |
+
"loss": 0.9698,
|
| 276 |
+
"step": 19000
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.6273626638783881,
|
| 280 |
+
"grad_norm": 1.4747998714447021,
|
| 281 |
+
"learning_rate": 3.432050704587865e-05,
|
| 282 |
+
"loss": 0.9657,
|
| 283 |
+
"step": 19500
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.6434488860291161,
|
| 287 |
+
"grad_norm": 3.0782012939453125,
|
| 288 |
+
"learning_rate": 3.391914934688888e-05,
|
| 289 |
+
"loss": 0.9379,
|
| 290 |
+
"step": 20000
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.659535108179844,
|
| 294 |
+
"grad_norm": 2.4914307594299316,
|
| 295 |
+
"learning_rate": 3.3516987323853034e-05,
|
| 296 |
+
"loss": 0.915,
|
| 297 |
+
"step": 20500
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.6756213303305718,
|
| 301 |
+
"grad_norm": 2.772120237350464,
|
| 302 |
+
"learning_rate": 3.3115629624863264e-05,
|
| 303 |
+
"loss": 0.9047,
|
| 304 |
+
"step": 21000
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 0.6917075524812998,
|
| 308 |
+
"grad_norm": 2.519575595855713,
|
| 309 |
+
"learning_rate": 3.271346760182743e-05,
|
| 310 |
+
"loss": 0.8688,
|
| 311 |
+
"step": 21500
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.7077937746320276,
|
| 315 |
+
"grad_norm": 4.085098743438721,
|
| 316 |
+
"learning_rate": 3.231130557879158e-05,
|
| 317 |
+
"loss": 0.8581,
|
| 318 |
+
"step": 22000
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.7238799967827556,
|
| 322 |
+
"grad_norm": 1.4670002460479736,
|
| 323 |
+
"learning_rate": 3.1909143555755745e-05,
|
| 324 |
+
"loss": 0.8354,
|
| 325 |
+
"step": 22500
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.7399662189334835,
|
| 329 |
+
"grad_norm": 2.4749488830566406,
|
| 330 |
+
"learning_rate": 3.1507785856765975e-05,
|
| 331 |
+
"loss": 0.8108,
|
| 332 |
+
"step": 23000
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.7560524410842113,
|
| 336 |
+
"grad_norm": 1.8635029792785645,
|
| 337 |
+
"learning_rate": 3.110562383373014e-05,
|
| 338 |
+
"loss": 0.7773,
|
| 339 |
+
"step": 23500
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.7721386632349393,
|
| 343 |
+
"grad_norm": 3.5713748931884766,
|
| 344 |
+
"learning_rate": 3.0703461810694294e-05,
|
| 345 |
+
"loss": 0.756,
|
| 346 |
+
"step": 24000
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.7882248853856672,
|
| 350 |
+
"grad_norm": 1.8903526067733765,
|
| 351 |
+
"learning_rate": 3.0301299787658456e-05,
|
| 352 |
+
"loss": 0.7326,
|
| 353 |
+
"step": 24500
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.8043111075363951,
|
| 357 |
+
"grad_norm": 8.286703109741211,
|
| 358 |
+
"learning_rate": 2.9899942088668686e-05,
|
| 359 |
+
"loss": 0.6948,
|
| 360 |
+
"step": 25000
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.820397329687123,
|
| 364 |
+
"grad_norm": 2.2209272384643555,
|
| 365 |
+
"learning_rate": 2.9497780065632845e-05,
|
| 366 |
+
"loss": 0.6914,
|
| 367 |
+
"step": 25500
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.8364835518378508,
|
| 371 |
+
"grad_norm": 2.2284536361694336,
|
| 372 |
+
"learning_rate": 2.9095618042597e-05,
|
| 373 |
+
"loss": 0.6585,
|
| 374 |
+
"step": 26000
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.8525697739885788,
|
| 378 |
+
"grad_norm": 3.4615938663482666,
|
| 379 |
+
"learning_rate": 2.869345601956116e-05,
|
| 380 |
+
"loss": 0.633,
|
| 381 |
+
"step": 26500
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.8686559961393067,
|
| 385 |
+
"grad_norm": 3.1158838272094727,
|
| 386 |
+
"learning_rate": 2.829209832057139e-05,
|
| 387 |
+
"loss": 0.6181,
|
| 388 |
+
"step": 27000
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.8847422182900346,
|
| 392 |
+
"grad_norm": 2.3320417404174805,
|
| 393 |
+
"learning_rate": 2.7889936297535553e-05,
|
| 394 |
+
"loss": 0.5993,
|
| 395 |
+
"step": 27500
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.9008284404407625,
|
| 399 |
+
"grad_norm": 1.8331427574157715,
|
| 400 |
+
"learning_rate": 2.7487774274499712e-05,
|
| 401 |
+
"loss": 0.5839,
|
| 402 |
+
"step": 28000
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.9169146625914903,
|
| 406 |
+
"grad_norm": 3.2398369312286377,
|
| 407 |
+
"learning_rate": 2.708561225146387e-05,
|
| 408 |
+
"loss": 0.562,
|
| 409 |
+
"step": 28500
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.9330008847422183,
|
| 413 |
+
"grad_norm": 1.6575061082839966,
|
| 414 |
+
"learning_rate": 2.66842545524741e-05,
|
| 415 |
+
"loss": 0.5313,
|
| 416 |
+
"step": 29000
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.9490871068929462,
|
| 420 |
+
"grad_norm": 2.1604230403900146,
|
| 421 |
+
"learning_rate": 2.6282092529438264e-05,
|
| 422 |
+
"loss": 0.5203,
|
| 423 |
+
"step": 29500
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.9651733290436741,
|
| 427 |
+
"grad_norm": 3.3743808269500732,
|
| 428 |
+
"learning_rate": 2.5879930506402423e-05,
|
| 429 |
+
"loss": 0.4938,
|
| 430 |
+
"step": 30000
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 0.981259551194402,
|
| 434 |
+
"grad_norm": 3.766514301300049,
|
| 435 |
+
"learning_rate": 2.5477768483366583e-05,
|
| 436 |
+
"loss": 0.4724,
|
| 437 |
+
"step": 30500
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 0.9973457733451299,
|
| 441 |
+
"grad_norm": 2.26712703704834,
|
| 442 |
+
"learning_rate": 2.5075606460330742e-05,
|
| 443 |
+
"loss": 0.4656,
|
| 444 |
+
"step": 31000
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 1.0,
|
| 448 |
+
"eval_loss": 0.26554691791534424,
|
| 449 |
+
"eval_runtime": 1917.4803,
|
| 450 |
+
"eval_samples_per_second": 345.81,
|
| 451 |
+
"eval_steps_per_second": 43.227,
|
| 452 |
+
"step": 31083
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 1.0134159092737072,
|
| 456 |
+
"grad_norm": 2.1041958332061768,
|
| 457 |
+
"learning_rate": 2.467424876134097e-05,
|
| 458 |
+
"loss": 0.4381,
|
| 459 |
+
"step": 31500
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 1.029502131424435,
|
| 463 |
+
"grad_norm": 1.7629106044769287,
|
| 464 |
+
"learning_rate": 2.427208673830513e-05,
|
| 465 |
+
"loss": 0.4298,
|
| 466 |
+
"step": 32000
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 1.0455883535751629,
|
| 470 |
+
"grad_norm": 2.5032904148101807,
|
| 471 |
+
"learning_rate": 2.386992471526929e-05,
|
| 472 |
+
"loss": 0.4188,
|
| 473 |
+
"step": 32500
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 1.0616745757258907,
|
| 477 |
+
"grad_norm": 1.6467881202697754,
|
| 478 |
+
"learning_rate": 2.3467762692233446e-05,
|
| 479 |
+
"loss": 0.3986,
|
| 480 |
+
"step": 33000
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"epoch": 1.0777607978766186,
|
| 484 |
+
"grad_norm": 1.957220435142517,
|
| 485 |
+
"learning_rate": 2.3065600669197606e-05,
|
| 486 |
+
"loss": 0.382,
|
| 487 |
+
"step": 33500
|
| 488 |
+
},
|
| 489 |
+
{
|
| 490 |
+
"epoch": 1.0938470200273467,
|
| 491 |
+
"grad_norm": 1.6566946506500244,
|
| 492 |
+
"learning_rate": 2.2663438646161765e-05,
|
| 493 |
+
"loss": 0.3689,
|
| 494 |
+
"step": 34000
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 1.1099332421780745,
|
| 498 |
+
"grad_norm": 2.081613540649414,
|
| 499 |
+
"learning_rate": 2.2261276623125928e-05,
|
| 500 |
+
"loss": 0.3603,
|
| 501 |
+
"step": 34500
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 1.1260194643288024,
|
| 505 |
+
"grad_norm": 2.155226945877075,
|
| 506 |
+
"learning_rate": 2.1859918924136157e-05,
|
| 507 |
+
"loss": 0.3478,
|
| 508 |
+
"step": 35000
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 1.1421056864795303,
|
| 512 |
+
"grad_norm": 1.9459590911865234,
|
| 513 |
+
"learning_rate": 2.1457756901100317e-05,
|
| 514 |
+
"loss": 0.3315,
|
| 515 |
+
"step": 35500
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 1.1581919086302581,
|
| 519 |
+
"grad_norm": 2.3381567001342773,
|
| 520 |
+
"learning_rate": 2.1055594878064476e-05,
|
| 521 |
+
"loss": 0.3259,
|
| 522 |
+
"step": 36000
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 1.1742781307809862,
|
| 526 |
+
"grad_norm": 1.4302254915237427,
|
| 527 |
+
"learning_rate": 2.0653432855028635e-05,
|
| 528 |
+
"loss": 0.3168,
|
| 529 |
+
"step": 36500
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 1.190364352931714,
|
| 533 |
+
"grad_norm": 1.1770597696304321,
|
| 534 |
+
"learning_rate": 2.0251270831992795e-05,
|
| 535 |
+
"loss": 0.3082,
|
| 536 |
+
"step": 37000
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 1.206450575082442,
|
| 540 |
+
"grad_norm": 1.7475298643112183,
|
| 541 |
+
"learning_rate": 1.9849913133003024e-05,
|
| 542 |
+
"loss": 0.3014,
|
| 543 |
+
"step": 37500
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 1.2225367972331698,
|
| 547 |
+
"grad_norm": 1.2397468090057373,
|
| 548 |
+
"learning_rate": 1.9447751109967187e-05,
|
| 549 |
+
"loss": 0.288,
|
| 550 |
+
"step": 38000
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 1.2386230193838976,
|
| 554 |
+
"grad_norm": 1.6603740453720093,
|
| 555 |
+
"learning_rate": 1.9045589086931343e-05,
|
| 556 |
+
"loss": 0.2797,
|
| 557 |
+
"step": 38500
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 1.2547092415346257,
|
| 561 |
+
"grad_norm": 1.7009538412094116,
|
| 562 |
+
"learning_rate": 1.8643427063895502e-05,
|
| 563 |
+
"loss": 0.275,
|
| 564 |
+
"step": 39000
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 1.2707954636853536,
|
| 568 |
+
"grad_norm": 1.4941717386245728,
|
| 569 |
+
"learning_rate": 1.8241265040859662e-05,
|
| 570 |
+
"loss": 0.2623,
|
| 571 |
+
"step": 39500
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 1.2868816858360814,
|
| 575 |
+
"grad_norm": 1.941115140914917,
|
| 576 |
+
"learning_rate": 1.7839907341869895e-05,
|
| 577 |
+
"loss": 0.2572,
|
| 578 |
+
"step": 40000
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 1.3029679079868093,
|
| 582 |
+
"grad_norm": 1.487726092338562,
|
| 583 |
+
"learning_rate": 1.7437745318834054e-05,
|
| 584 |
+
"loss": 0.2502,
|
| 585 |
+
"step": 40500
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 1.3190541301375371,
|
| 589 |
+
"grad_norm": 1.4628674983978271,
|
| 590 |
+
"learning_rate": 1.7035583295798213e-05,
|
| 591 |
+
"loss": 0.2437,
|
| 592 |
+
"step": 41000
|
| 593 |
+
},
|
| 594 |
+
{
|
| 595 |
+
"epoch": 1.3351403522882652,
|
| 596 |
+
"grad_norm": 1.401607632637024,
|
| 597 |
+
"learning_rate": 1.663342127276237e-05,
|
| 598 |
+
"loss": 0.2421,
|
| 599 |
+
"step": 41500
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 1.351226574438993,
|
| 603 |
+
"grad_norm": 1.1497563123703003,
|
| 604 |
+
"learning_rate": 1.623125924972653e-05,
|
| 605 |
+
"loss": 0.231,
|
| 606 |
+
"step": 42000
|
| 607 |
+
},
|
| 608 |
+
{
|
| 609 |
+
"epoch": 1.367312796589721,
|
| 610 |
+
"grad_norm": 1.322836995124817,
|
| 611 |
+
"learning_rate": 1.5829097226690688e-05,
|
| 612 |
+
"loss": 0.2261,
|
| 613 |
+
"step": 42500
|
| 614 |
+
},
|
| 615 |
+
{
|
| 616 |
+
"epoch": 1.3833990187404488,
|
| 617 |
+
"grad_norm": 1.5328525304794312,
|
| 618 |
+
"learning_rate": 1.542773952770092e-05,
|
| 619 |
+
"loss": 0.2177,
|
| 620 |
+
"step": 43000
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 1.3994852408911767,
|
| 624 |
+
"grad_norm": 1.7748241424560547,
|
| 625 |
+
"learning_rate": 1.502557750466508e-05,
|
| 626 |
+
"loss": 0.2186,
|
| 627 |
+
"step": 43500
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 1.4155714630419047,
|
| 631 |
+
"grad_norm": 1.6542141437530518,
|
| 632 |
+
"learning_rate": 1.4623415481629241e-05,
|
| 633 |
+
"loss": 0.2138,
|
| 634 |
+
"step": 44000
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 1.4316576851926326,
|
| 638 |
+
"grad_norm": 1.3098843097686768,
|
| 639 |
+
"learning_rate": 1.4221253458593397e-05,
|
| 640 |
+
"loss": 0.211,
|
| 641 |
+
"step": 44500
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 1.4477439073433604,
|
| 645 |
+
"grad_norm": 1.345651626586914,
|
| 646 |
+
"learning_rate": 1.3819091435557557e-05,
|
| 647 |
+
"loss": 0.2027,
|
| 648 |
+
"step": 45000
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 1.4638301294940883,
|
| 652 |
+
"grad_norm": 1.4520297050476074,
|
| 653 |
+
"learning_rate": 1.3416929412521718e-05,
|
| 654 |
+
"loss": 0.2039,
|
| 655 |
+
"step": 45500
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 1.4799163516448162,
|
| 659 |
+
"grad_norm": 1.5913499593734741,
|
| 660 |
+
"learning_rate": 1.3014767389485877e-05,
|
| 661 |
+
"loss": 0.1939,
|
| 662 |
+
"step": 46000
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 1.4960025737955442,
|
| 666 |
+
"grad_norm": 1.1803226470947266,
|
| 667 |
+
"learning_rate": 1.2612605366450037e-05,
|
| 668 |
+
"loss": 0.1887,
|
| 669 |
+
"step": 46500
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 1.5120887959462719,
|
| 673 |
+
"grad_norm": 1.1462236642837524,
|
| 674 |
+
"learning_rate": 1.2210443343414194e-05,
|
| 675 |
+
"loss": 0.1883,
|
| 676 |
+
"step": 47000
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"epoch": 1.528175018097,
|
| 680 |
+
"grad_norm": 0.8483968377113342,
|
| 681 |
+
"learning_rate": 1.1808281320378355e-05,
|
| 682 |
+
"loss": 0.1809,
|
| 683 |
+
"step": 47500
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 1.5442612402477278,
|
| 687 |
+
"grad_norm": 1.1205823421478271,
|
| 688 |
+
"learning_rate": 1.1406119297342515e-05,
|
| 689 |
+
"loss": 0.1813,
|
| 690 |
+
"step": 48000
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"epoch": 1.5603474623984557,
|
| 694 |
+
"grad_norm": 1.417622447013855,
|
| 695 |
+
"learning_rate": 1.1003957274306672e-05,
|
| 696 |
+
"loss": 0.1788,
|
| 697 |
+
"step": 48500
|
| 698 |
+
},
|
| 699 |
+
{
|
| 700 |
+
"epoch": 1.5764336845491838,
|
| 701 |
+
"grad_norm": 1.179103970527649,
|
| 702 |
+
"learning_rate": 1.0602599575316904e-05,
|
| 703 |
+
"loss": 0.1809,
|
| 704 |
+
"step": 49000
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 1.5925199066999114,
|
| 708 |
+
"grad_norm": 1.1092889308929443,
|
| 709 |
+
"learning_rate": 1.0200437552281065e-05,
|
| 710 |
+
"loss": 0.1734,
|
| 711 |
+
"step": 49500
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 1.6086061288506395,
|
| 715 |
+
"grad_norm": 1.0196574926376343,
|
| 716 |
+
"learning_rate": 9.798275529245222e-06,
|
| 717 |
+
"loss": 0.1688,
|
| 718 |
+
"step": 50000
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 1.6246923510013673,
|
| 722 |
+
"grad_norm": 1.1376862525939941,
|
| 723 |
+
"learning_rate": 9.396113506209382e-06,
|
| 724 |
+
"loss": 0.1703,
|
| 725 |
+
"step": 50500
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 1.6407785731520952,
|
| 729 |
+
"grad_norm": 0.8885149955749512,
|
| 730 |
+
"learning_rate": 8.995560131265685e-06,
|
| 731 |
+
"loss": 0.1691,
|
| 732 |
+
"step": 51000
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 1.6568647953028233,
|
| 736 |
+
"grad_norm": 1.2574944496154785,
|
| 737 |
+
"learning_rate": 8.593398108229844e-06,
|
| 738 |
+
"loss": 0.1615,
|
| 739 |
+
"step": 51500
|
| 740 |
+
},
|
| 741 |
+
{
|
| 742 |
+
"epoch": 1.672951017453551,
|
| 743 |
+
"grad_norm": 1.2620723247528076,
|
| 744 |
+
"learning_rate": 8.191236085194004e-06,
|
| 745 |
+
"loss": 0.1593,
|
| 746 |
+
"step": 52000
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 1.689037239604279,
|
| 750 |
+
"grad_norm": 1.551480770111084,
|
| 751 |
+
"learning_rate": 7.789074062158163e-06,
|
| 752 |
+
"loss": 0.1639,
|
| 753 |
+
"step": 52500
|
| 754 |
+
},
|
| 755 |
+
{
|
| 756 |
+
"epoch": 1.7051234617550068,
|
| 757 |
+
"grad_norm": 1.5938962697982788,
|
| 758 |
+
"learning_rate": 7.386912039122322e-06,
|
| 759 |
+
"loss": 0.1587,
|
| 760 |
+
"step": 53000
|
| 761 |
+
},
|
| 762 |
+
{
|
| 763 |
+
"epoch": 1.7212096839057347,
|
| 764 |
+
"grad_norm": 1.0503953695297241,
|
| 765 |
+
"learning_rate": 6.984750016086482e-06,
|
| 766 |
+
"loss": 0.1599,
|
| 767 |
+
"step": 53500
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 1.7372959060564628,
|
| 771 |
+
"grad_norm": 1.1205036640167236,
|
| 772 |
+
"learning_rate": 6.583392317096712e-06,
|
| 773 |
+
"loss": 0.1541,
|
| 774 |
+
"step": 54000
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 1.7533821282071904,
|
| 778 |
+
"grad_norm": 0.7524433732032776,
|
| 779 |
+
"learning_rate": 6.181230294060872e-06,
|
| 780 |
+
"loss": 0.1521,
|
| 781 |
+
"step": 54500
|
| 782 |
+
},
|
| 783 |
+
{
|
| 784 |
+
"epoch": 1.7694683503579185,
|
| 785 |
+
"grad_norm": 0.9619775414466858,
|
| 786 |
+
"learning_rate": 5.779068271025031e-06,
|
| 787 |
+
"loss": 0.1521,
|
| 788 |
+
"step": 55000
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 1.7855545725086464,
|
| 792 |
+
"grad_norm": 0.9406844973564148,
|
| 793 |
+
"learning_rate": 5.37690624798919e-06,
|
| 794 |
+
"loss": 0.1509,
|
| 795 |
+
"step": 55500
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 1.8016407946593742,
|
| 799 |
+
"grad_norm": 0.9363726377487183,
|
| 800 |
+
"learning_rate": 4.975548548999421e-06,
|
| 801 |
+
"loss": 0.1513,
|
| 802 |
+
"step": 56000
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 1.8177270168101023,
|
| 806 |
+
"grad_norm": 0.9941402673721313,
|
| 807 |
+
"learning_rate": 4.573386525963581e-06,
|
| 808 |
+
"loss": 0.1484,
|
| 809 |
+
"step": 56500
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 1.83381323896083,
|
| 813 |
+
"grad_norm": 1.3756345510482788,
|
| 814 |
+
"learning_rate": 4.17122450292774e-06,
|
| 815 |
+
"loss": 0.1509,
|
| 816 |
+
"step": 57000
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 1.849899461111558,
|
| 820 |
+
"grad_norm": 1.0644595623016357,
|
| 821 |
+
"learning_rate": 3.7690624798918986e-06,
|
| 822 |
+
"loss": 0.1486,
|
| 823 |
+
"step": 57500
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"epoch": 1.8659856832622859,
|
| 827 |
+
"grad_norm": 1.070890188217163,
|
| 828 |
+
"learning_rate": 3.3669004568560584e-06,
|
| 829 |
+
"loss": 0.1462,
|
| 830 |
+
"step": 58000
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 1.8820719054130137,
|
| 834 |
+
"grad_norm": 1.3034768104553223,
|
| 835 |
+
"learning_rate": 2.9647384338202173e-06,
|
| 836 |
+
"loss": 0.1481,
|
| 837 |
+
"step": 58500
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 1.8981581275637418,
|
| 841 |
+
"grad_norm": 1.127517580986023,
|
| 842 |
+
"learning_rate": 2.5625764107843767e-06,
|
| 843 |
+
"loss": 0.1451,
|
| 844 |
+
"step": 59000
|
| 845 |
+
},
|
| 846 |
+
{
|
| 847 |
+
"epoch": 1.9142443497144694,
|
| 848 |
+
"grad_norm": 0.9431403279304504,
|
| 849 |
+
"learning_rate": 2.1604143877485364e-06,
|
| 850 |
+
"loss": 0.1458,
|
| 851 |
+
"step": 59500
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 1.9303305718651975,
|
| 855 |
+
"grad_norm": 1.271483302116394,
|
| 856 |
+
"learning_rate": 1.7590566887587673e-06,
|
| 857 |
+
"loss": 0.1463,
|
| 858 |
+
"step": 60000
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 1.9464167940159254,
|
| 862 |
+
"grad_norm": 0.7327952980995178,
|
| 863 |
+
"learning_rate": 1.3568946657229264e-06,
|
| 864 |
+
"loss": 0.1434,
|
| 865 |
+
"step": 60500
|
| 866 |
+
},
|
| 867 |
+
{
|
| 868 |
+
"epoch": 1.9625030161666532,
|
| 869 |
+
"grad_norm": 1.0670543909072876,
|
| 870 |
+
"learning_rate": 9.547326426870858e-07,
|
| 871 |
+
"loss": 0.1424,
|
| 872 |
+
"step": 61000
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 1.9785892383173813,
|
| 876 |
+
"grad_norm": 1.2705425024032593,
|
| 877 |
+
"learning_rate": 5.525706196512451e-07,
|
| 878 |
+
"loss": 0.1431,
|
| 879 |
+
"step": 61500
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 1.994675460468109,
|
| 883 |
+
"grad_norm": 0.9267213344573975,
|
| 884 |
+
"learning_rate": 1.5040859661540443e-07,
|
| 885 |
+
"loss": 0.1418,
|
| 886 |
+
"step": 62000
|
| 887 |
+
}
|
| 888 |
+
],
|
| 889 |
+
"logging_steps": 500,
|
| 890 |
+
"max_steps": 62164,
|
| 891 |
+
"num_input_tokens_seen": 0,
|
| 892 |
+
"num_train_epochs": 2,
|
| 893 |
+
"save_steps": 500,
|
| 894 |
+
"stateful_callbacks": {
|
| 895 |
+
"TrainerControl": {
|
| 896 |
+
"args": {
|
| 897 |
+
"should_epoch_stop": false,
|
| 898 |
+
"should_evaluate": false,
|
| 899 |
+
"should_log": false,
|
| 900 |
+
"should_save": true,
|
| 901 |
+
"should_training_stop": false
|
| 902 |
+
},
|
| 903 |
+
"attributes": {}
|
| 904 |
+
}
|
| 905 |
+
},
|
| 906 |
+
"total_flos": 1.3425693542685082e+17,
|
| 907 |
+
"train_batch_size": 32,
|
| 908 |
+
"trial_name": null,
|
| 909 |
+
"trial_params": null
|
| 910 |
+
}
|
checkpoints/{checkpoint-57500 → checkpoint-62000}/training_args.bin
RENAMED
|
File without changes
|
checkpoints/{checkpoint-57726 → checkpoint-62164}/config.json
RENAMED
|
File without changes
|
checkpoints/{checkpoint-57726 → checkpoint-62164}/generation_config.json
RENAMED
|
File without changes
|
checkpoints/{checkpoint-57726 → checkpoint-62164}/model.safetensors
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 258368552
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7e8c8770f6dcbe1b9d9475af0bda9b1c3d8b183e7518931c4ee48457079834ba
|
| 3 |
size 258368552
|
checkpoints/{checkpoint-57726 → checkpoint-62164}/optimizer.pt
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 516816826
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fd427ed1567c5a01aba6a5e07e5f19d0d2b1351120d29cd57c3503632243f624
|
| 3 |
size 516816826
|
checkpoints/{checkpoint-57000 → checkpoint-62164}/rng_state.pth
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d5400ec42823ce9dd7ec0af9b56911140e87ed03e3a335d283424176da15c490
|
| 3 |
size 14244
|
checkpoints/{checkpoint-57000 → checkpoint-62164}/scaler.pt
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 988
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:44e5b40555d97ad78f23a565d71371cb3eae76e6f0c043cf1a9db3960ced1f75
|
| 3 |
size 988
|
checkpoints/{checkpoint-57500 → checkpoint-62164}/scheduler.pt
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c868981b82a0a6bae2aa0dd7a6e1355ee650c2f6cfd3fe97b38094bc4d5e87e4
|
| 3 |
size 1064
|
checkpoints/{checkpoint-57726 → checkpoint-62164}/special_tokens_map.json
RENAMED
|
File without changes
|
checkpoints/{checkpoint-57726 → checkpoint-62164}/spiece.model
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:610d79d092a886e9af7d31b94512c46fd20a5f62557bd5b8d8a5b23f1f78650a
|
| 3 |
+
size 1042971
|
checkpoints/{checkpoint-57726 → checkpoint-62164}/tokenizer.json
RENAMED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoints/{checkpoint-57726 → checkpoint-62164}/tokenizer_config.json
RENAMED
|
File without changes
|
checkpoints/checkpoint-62164/trainer_state.json
ADDED
|
@@ -0,0 +1,910 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 1.9999517413335477,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 62164,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.016086222150727903,
|
| 14 |
+
"grad_norm": 0.5078127384185791,
|
| 15 |
+
"learning_rate": 4.959864230101023e-05,
|
| 16 |
+
"loss": 2.1432,
|
| 17 |
+
"step": 500
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.032172444301455806,
|
| 21 |
+
"grad_norm": 0.4508506655693054,
|
| 22 |
+
"learning_rate": 4.9196480277974395e-05,
|
| 23 |
+
"loss": 1.9093,
|
| 24 |
+
"step": 1000
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.048258666452183706,
|
| 28 |
+
"grad_norm": 0.4430558979511261,
|
| 29 |
+
"learning_rate": 4.879431825493855e-05,
|
| 30 |
+
"loss": 1.8418,
|
| 31 |
+
"step": 1500
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.06434488860291161,
|
| 35 |
+
"grad_norm": 0.4775325059890747,
|
| 36 |
+
"learning_rate": 4.8392156231902713e-05,
|
| 37 |
+
"loss": 1.7771,
|
| 38 |
+
"step": 2000
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.08043111075363951,
|
| 42 |
+
"grad_norm": 0.49685001373291016,
|
| 43 |
+
"learning_rate": 4.7989994208866876e-05,
|
| 44 |
+
"loss": 1.7226,
|
| 45 |
+
"step": 2500
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.09651733290436741,
|
| 49 |
+
"grad_norm": 0.5552434325218201,
|
| 50 |
+
"learning_rate": 4.7587832185831025e-05,
|
| 51 |
+
"loss": 1.6767,
|
| 52 |
+
"step": 3000
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.11260355505509531,
|
| 56 |
+
"grad_norm": 0.6779139637947083,
|
| 57 |
+
"learning_rate": 4.718567016279519e-05,
|
| 58 |
+
"loss": 1.6588,
|
| 59 |
+
"step": 3500
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.12868977720582322,
|
| 63 |
+
"grad_norm": 0.5552022457122803,
|
| 64 |
+
"learning_rate": 4.6783508139759344e-05,
|
| 65 |
+
"loss": 1.603,
|
| 66 |
+
"step": 4000
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.1447759993565511,
|
| 70 |
+
"grad_norm": 0.5302042365074158,
|
| 71 |
+
"learning_rate": 4.638134611672351e-05,
|
| 72 |
+
"loss": 1.5776,
|
| 73 |
+
"step": 4500
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.16086222150727902,
|
| 77 |
+
"grad_norm": 0.5810815691947937,
|
| 78 |
+
"learning_rate": 4.597918409368766e-05,
|
| 79 |
+
"loss": 1.5333,
|
| 80 |
+
"step": 5000
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.1769484436580069,
|
| 84 |
+
"grad_norm": 0.5819700956344604,
|
| 85 |
+
"learning_rate": 4.5577022070651826e-05,
|
| 86 |
+
"loss": 1.5168,
|
| 87 |
+
"step": 5500
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.19303466580873482,
|
| 91 |
+
"grad_norm": 0.6134072542190552,
|
| 92 |
+
"learning_rate": 4.517486004761599e-05,
|
| 93 |
+
"loss": 1.4748,
|
| 94 |
+
"step": 6000
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.2091208879594627,
|
| 98 |
+
"grad_norm": 0.5746152400970459,
|
| 99 |
+
"learning_rate": 4.4772698024580144e-05,
|
| 100 |
+
"loss": 1.4622,
|
| 101 |
+
"step": 6500
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.22520711011019062,
|
| 105 |
+
"grad_norm": 0.7663710713386536,
|
| 106 |
+
"learning_rate": 4.437053600154431e-05,
|
| 107 |
+
"loss": 1.4767,
|
| 108 |
+
"step": 7000
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.24129333226091854,
|
| 112 |
+
"grad_norm": 0.7993176579475403,
|
| 113 |
+
"learning_rate": 4.396837397850846e-05,
|
| 114 |
+
"loss": 1.4527,
|
| 115 |
+
"step": 7500
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.25737955441164645,
|
| 119 |
+
"grad_norm": 0.6892676949501038,
|
| 120 |
+
"learning_rate": 4.3566211955472626e-05,
|
| 121 |
+
"loss": 1.4325,
|
| 122 |
+
"step": 8000
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.2734657765623743,
|
| 126 |
+
"grad_norm": 0.6928556561470032,
|
| 127 |
+
"learning_rate": 4.316404993243678e-05,
|
| 128 |
+
"loss": 1.4038,
|
| 129 |
+
"step": 8500
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.2895519987131022,
|
| 133 |
+
"grad_norm": 0.7578593492507935,
|
| 134 |
+
"learning_rate": 4.2761887909400944e-05,
|
| 135 |
+
"loss": 1.3945,
|
| 136 |
+
"step": 9000
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.30563822086383013,
|
| 140 |
+
"grad_norm": 0.7504703402519226,
|
| 141 |
+
"learning_rate": 4.23597258863651e-05,
|
| 142 |
+
"loss": 1.3644,
|
| 143 |
+
"step": 9500
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.32172444301455805,
|
| 147 |
+
"grad_norm": 0.8370710611343384,
|
| 148 |
+
"learning_rate": 4.1957563863329256e-05,
|
| 149 |
+
"loss": 1.3619,
|
| 150 |
+
"step": 10000
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.3378106651652859,
|
| 154 |
+
"grad_norm": 0.8501142263412476,
|
| 155 |
+
"learning_rate": 4.155540184029342e-05,
|
| 156 |
+
"loss": 1.3448,
|
| 157 |
+
"step": 10500
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.3538968873160138,
|
| 161 |
+
"grad_norm": 0.9001900553703308,
|
| 162 |
+
"learning_rate": 4.1153239817257575e-05,
|
| 163 |
+
"loss": 1.3004,
|
| 164 |
+
"step": 11000
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.36998310946674173,
|
| 168 |
+
"grad_norm": 1.0658681392669678,
|
| 169 |
+
"learning_rate": 4.075107779422174e-05,
|
| 170 |
+
"loss": 1.2789,
|
| 171 |
+
"step": 11500
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.38606933161746965,
|
| 175 |
+
"grad_norm": 1.1038371324539185,
|
| 176 |
+
"learning_rate": 4.0348915771185894e-05,
|
| 177 |
+
"loss": 1.2651,
|
| 178 |
+
"step": 12000
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.40215555376819756,
|
| 182 |
+
"grad_norm": 1.2004213333129883,
|
| 183 |
+
"learning_rate": 3.994755807219613e-05,
|
| 184 |
+
"loss": 1.2216,
|
| 185 |
+
"step": 12500
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.4182417759189254,
|
| 189 |
+
"grad_norm": 1.235543966293335,
|
| 190 |
+
"learning_rate": 3.9545396049160286e-05,
|
| 191 |
+
"loss": 1.1955,
|
| 192 |
+
"step": 13000
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.43432799806965333,
|
| 196 |
+
"grad_norm": 1.5088828802108765,
|
| 197 |
+
"learning_rate": 3.914323402612445e-05,
|
| 198 |
+
"loss": 1.1836,
|
| 199 |
+
"step": 13500
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.45041422022038125,
|
| 203 |
+
"grad_norm": 1.264153242111206,
|
| 204 |
+
"learning_rate": 3.8741072003088605e-05,
|
| 205 |
+
"loss": 1.1658,
|
| 206 |
+
"step": 14000
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.46650044237110916,
|
| 210 |
+
"grad_norm": 1.3023343086242676,
|
| 211 |
+
"learning_rate": 3.833971430409884e-05,
|
| 212 |
+
"loss": 1.1481,
|
| 213 |
+
"step": 14500
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.48258666452183707,
|
| 217 |
+
"grad_norm": 1.3824670314788818,
|
| 218 |
+
"learning_rate": 3.7938356605109064e-05,
|
| 219 |
+
"loss": 1.1221,
|
| 220 |
+
"step": 15000
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.49867288667256493,
|
| 224 |
+
"grad_norm": 1.4364969730377197,
|
| 225 |
+
"learning_rate": 3.75369989061193e-05,
|
| 226 |
+
"loss": 1.1057,
|
| 227 |
+
"step": 15500
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.5147591088232929,
|
| 231 |
+
"grad_norm": 2.051701545715332,
|
| 232 |
+
"learning_rate": 3.7134836883083456e-05,
|
| 233 |
+
"loss": 1.0873,
|
| 234 |
+
"step": 16000
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.5308453309740208,
|
| 238 |
+
"grad_norm": 1.4329720735549927,
|
| 239 |
+
"learning_rate": 3.673267486004762e-05,
|
| 240 |
+
"loss": 1.0607,
|
| 241 |
+
"step": 16500
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.5469315531247486,
|
| 245 |
+
"grad_norm": 1.4981014728546143,
|
| 246 |
+
"learning_rate": 3.6330512837011775e-05,
|
| 247 |
+
"loss": 1.0516,
|
| 248 |
+
"step": 17000
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.5630177752754766,
|
| 252 |
+
"grad_norm": 1.3012079000473022,
|
| 253 |
+
"learning_rate": 3.592835081397594e-05,
|
| 254 |
+
"loss": 1.0317,
|
| 255 |
+
"step": 17500
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.5791039974262044,
|
| 259 |
+
"grad_norm": 1.401825189590454,
|
| 260 |
+
"learning_rate": 3.552699311498617e-05,
|
| 261 |
+
"loss": 1.0183,
|
| 262 |
+
"step": 18000
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.5951902195769324,
|
| 266 |
+
"grad_norm": 2.0783369541168213,
|
| 267 |
+
"learning_rate": 3.512483109195033e-05,
|
| 268 |
+
"loss": 0.9985,
|
| 269 |
+
"step": 18500
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.6112764417276603,
|
| 273 |
+
"grad_norm": 2.3940794467926025,
|
| 274 |
+
"learning_rate": 3.4722669068914486e-05,
|
| 275 |
+
"loss": 0.9698,
|
| 276 |
+
"step": 19000
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.6273626638783881,
|
| 280 |
+
"grad_norm": 1.4747998714447021,
|
| 281 |
+
"learning_rate": 3.432050704587865e-05,
|
| 282 |
+
"loss": 0.9657,
|
| 283 |
+
"step": 19500
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.6434488860291161,
|
| 287 |
+
"grad_norm": 3.0782012939453125,
|
| 288 |
+
"learning_rate": 3.391914934688888e-05,
|
| 289 |
+
"loss": 0.9379,
|
| 290 |
+
"step": 20000
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.659535108179844,
|
| 294 |
+
"grad_norm": 2.4914307594299316,
|
| 295 |
+
"learning_rate": 3.3516987323853034e-05,
|
| 296 |
+
"loss": 0.915,
|
| 297 |
+
"step": 20500
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.6756213303305718,
|
| 301 |
+
"grad_norm": 2.772120237350464,
|
| 302 |
+
"learning_rate": 3.3115629624863264e-05,
|
| 303 |
+
"loss": 0.9047,
|
| 304 |
+
"step": 21000
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 0.6917075524812998,
|
| 308 |
+
"grad_norm": 2.519575595855713,
|
| 309 |
+
"learning_rate": 3.271346760182743e-05,
|
| 310 |
+
"loss": 0.8688,
|
| 311 |
+
"step": 21500
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.7077937746320276,
|
| 315 |
+
"grad_norm": 4.085098743438721,
|
| 316 |
+
"learning_rate": 3.231130557879158e-05,
|
| 317 |
+
"loss": 0.8581,
|
| 318 |
+
"step": 22000
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.7238799967827556,
|
| 322 |
+
"grad_norm": 1.4670002460479736,
|
| 323 |
+
"learning_rate": 3.1909143555755745e-05,
|
| 324 |
+
"loss": 0.8354,
|
| 325 |
+
"step": 22500
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.7399662189334835,
|
| 329 |
+
"grad_norm": 2.4749488830566406,
|
| 330 |
+
"learning_rate": 3.1507785856765975e-05,
|
| 331 |
+
"loss": 0.8108,
|
| 332 |
+
"step": 23000
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.7560524410842113,
|
| 336 |
+
"grad_norm": 1.8635029792785645,
|
| 337 |
+
"learning_rate": 3.110562383373014e-05,
|
| 338 |
+
"loss": 0.7773,
|
| 339 |
+
"step": 23500
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.7721386632349393,
|
| 343 |
+
"grad_norm": 3.5713748931884766,
|
| 344 |
+
"learning_rate": 3.0703461810694294e-05,
|
| 345 |
+
"loss": 0.756,
|
| 346 |
+
"step": 24000
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.7882248853856672,
|
| 350 |
+
"grad_norm": 1.8903526067733765,
|
| 351 |
+
"learning_rate": 3.0301299787658456e-05,
|
| 352 |
+
"loss": 0.7326,
|
| 353 |
+
"step": 24500
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.8043111075363951,
|
| 357 |
+
"grad_norm": 8.286703109741211,
|
| 358 |
+
"learning_rate": 2.9899942088668686e-05,
|
| 359 |
+
"loss": 0.6948,
|
| 360 |
+
"step": 25000
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.820397329687123,
|
| 364 |
+
"grad_norm": 2.2209272384643555,
|
| 365 |
+
"learning_rate": 2.9497780065632845e-05,
|
| 366 |
+
"loss": 0.6914,
|
| 367 |
+
"step": 25500
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.8364835518378508,
|
| 371 |
+
"grad_norm": 2.2284536361694336,
|
| 372 |
+
"learning_rate": 2.9095618042597e-05,
|
| 373 |
+
"loss": 0.6585,
|
| 374 |
+
"step": 26000
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.8525697739885788,
|
| 378 |
+
"grad_norm": 3.4615938663482666,
|
| 379 |
+
"learning_rate": 2.869345601956116e-05,
|
| 380 |
+
"loss": 0.633,
|
| 381 |
+
"step": 26500
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.8686559961393067,
|
| 385 |
+
"grad_norm": 3.1158838272094727,
|
| 386 |
+
"learning_rate": 2.829209832057139e-05,
|
| 387 |
+
"loss": 0.6181,
|
| 388 |
+
"step": 27000
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.8847422182900346,
|
| 392 |
+
"grad_norm": 2.3320417404174805,
|
| 393 |
+
"learning_rate": 2.7889936297535553e-05,
|
| 394 |
+
"loss": 0.5993,
|
| 395 |
+
"step": 27500
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.9008284404407625,
|
| 399 |
+
"grad_norm": 1.8331427574157715,
|
| 400 |
+
"learning_rate": 2.7487774274499712e-05,
|
| 401 |
+
"loss": 0.5839,
|
| 402 |
+
"step": 28000
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.9169146625914903,
|
| 406 |
+
"grad_norm": 3.2398369312286377,
|
| 407 |
+
"learning_rate": 2.708561225146387e-05,
|
| 408 |
+
"loss": 0.562,
|
| 409 |
+
"step": 28500
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.9330008847422183,
|
| 413 |
+
"grad_norm": 1.6575061082839966,
|
| 414 |
+
"learning_rate": 2.66842545524741e-05,
|
| 415 |
+
"loss": 0.5313,
|
| 416 |
+
"step": 29000
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.9490871068929462,
|
| 420 |
+
"grad_norm": 2.1604230403900146,
|
| 421 |
+
"learning_rate": 2.6282092529438264e-05,
|
| 422 |
+
"loss": 0.5203,
|
| 423 |
+
"step": 29500
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.9651733290436741,
|
| 427 |
+
"grad_norm": 3.3743808269500732,
|
| 428 |
+
"learning_rate": 2.5879930506402423e-05,
|
| 429 |
+
"loss": 0.4938,
|
| 430 |
+
"step": 30000
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 0.981259551194402,
|
| 434 |
+
"grad_norm": 3.766514301300049,
|
| 435 |
+
"learning_rate": 2.5477768483366583e-05,
|
| 436 |
+
"loss": 0.4724,
|
| 437 |
+
"step": 30500
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 0.9973457733451299,
|
| 441 |
+
"grad_norm": 2.26712703704834,
|
| 442 |
+
"learning_rate": 2.5075606460330742e-05,
|
| 443 |
+
"loss": 0.4656,
|
| 444 |
+
"step": 31000
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 1.0,
|
| 448 |
+
"eval_loss": 0.26554691791534424,
|
| 449 |
+
"eval_runtime": 1917.4803,
|
| 450 |
+
"eval_samples_per_second": 345.81,
|
| 451 |
+
"eval_steps_per_second": 43.227,
|
| 452 |
+
"step": 31083
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 1.0134159092737072,
|
| 456 |
+
"grad_norm": 2.1041958332061768,
|
| 457 |
+
"learning_rate": 2.467424876134097e-05,
|
| 458 |
+
"loss": 0.4381,
|
| 459 |
+
"step": 31500
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 1.029502131424435,
|
| 463 |
+
"grad_norm": 1.7629106044769287,
|
| 464 |
+
"learning_rate": 2.427208673830513e-05,
|
| 465 |
+
"loss": 0.4298,
|
| 466 |
+
"step": 32000
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 1.0455883535751629,
|
| 470 |
+
"grad_norm": 2.5032904148101807,
|
| 471 |
+
"learning_rate": 2.386992471526929e-05,
|
| 472 |
+
"loss": 0.4188,
|
| 473 |
+
"step": 32500
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 1.0616745757258907,
|
| 477 |
+
"grad_norm": 1.6467881202697754,
|
| 478 |
+
"learning_rate": 2.3467762692233446e-05,
|
| 479 |
+
"loss": 0.3986,
|
| 480 |
+
"step": 33000
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"epoch": 1.0777607978766186,
|
| 484 |
+
"grad_norm": 1.957220435142517,
|
| 485 |
+
"learning_rate": 2.3065600669197606e-05,
|
| 486 |
+
"loss": 0.382,
|
| 487 |
+
"step": 33500
|
| 488 |
+
},
|
| 489 |
+
{
|
| 490 |
+
"epoch": 1.0938470200273467,
|
| 491 |
+
"grad_norm": 1.6566946506500244,
|
| 492 |
+
"learning_rate": 2.2663438646161765e-05,
|
| 493 |
+
"loss": 0.3689,
|
| 494 |
+
"step": 34000
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 1.1099332421780745,
|
| 498 |
+
"grad_norm": 2.081613540649414,
|
| 499 |
+
"learning_rate": 2.2261276623125928e-05,
|
| 500 |
+
"loss": 0.3603,
|
| 501 |
+
"step": 34500
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 1.1260194643288024,
|
| 505 |
+
"grad_norm": 2.155226945877075,
|
| 506 |
+
"learning_rate": 2.1859918924136157e-05,
|
| 507 |
+
"loss": 0.3478,
|
| 508 |
+
"step": 35000
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 1.1421056864795303,
|
| 512 |
+
"grad_norm": 1.9459590911865234,
|
| 513 |
+
"learning_rate": 2.1457756901100317e-05,
|
| 514 |
+
"loss": 0.3315,
|
| 515 |
+
"step": 35500
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 1.1581919086302581,
|
| 519 |
+
"grad_norm": 2.3381567001342773,
|
| 520 |
+
"learning_rate": 2.1055594878064476e-05,
|
| 521 |
+
"loss": 0.3259,
|
| 522 |
+
"step": 36000
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 1.1742781307809862,
|
| 526 |
+
"grad_norm": 1.4302254915237427,
|
| 527 |
+
"learning_rate": 2.0653432855028635e-05,
|
| 528 |
+
"loss": 0.3168,
|
| 529 |
+
"step": 36500
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 1.190364352931714,
|
| 533 |
+
"grad_norm": 1.1770597696304321,
|
| 534 |
+
"learning_rate": 2.0251270831992795e-05,
|
| 535 |
+
"loss": 0.3082,
|
| 536 |
+
"step": 37000
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 1.206450575082442,
|
| 540 |
+
"grad_norm": 1.7475298643112183,
|
| 541 |
+
"learning_rate": 1.9849913133003024e-05,
|
| 542 |
+
"loss": 0.3014,
|
| 543 |
+
"step": 37500
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 1.2225367972331698,
|
| 547 |
+
"grad_norm": 1.2397468090057373,
|
| 548 |
+
"learning_rate": 1.9447751109967187e-05,
|
| 549 |
+
"loss": 0.288,
|
| 550 |
+
"step": 38000
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 1.2386230193838976,
|
| 554 |
+
"grad_norm": 1.6603740453720093,
|
| 555 |
+
"learning_rate": 1.9045589086931343e-05,
|
| 556 |
+
"loss": 0.2797,
|
| 557 |
+
"step": 38500
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 1.2547092415346257,
|
| 561 |
+
"grad_norm": 1.7009538412094116,
|
| 562 |
+
"learning_rate": 1.8643427063895502e-05,
|
| 563 |
+
"loss": 0.275,
|
| 564 |
+
"step": 39000
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 1.2707954636853536,
|
| 568 |
+
"grad_norm": 1.4941717386245728,
|
| 569 |
+
"learning_rate": 1.8241265040859662e-05,
|
| 570 |
+
"loss": 0.2623,
|
| 571 |
+
"step": 39500
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 1.2868816858360814,
|
| 575 |
+
"grad_norm": 1.941115140914917,
|
| 576 |
+
"learning_rate": 1.7839907341869895e-05,
|
| 577 |
+
"loss": 0.2572,
|
| 578 |
+
"step": 40000
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 1.3029679079868093,
|
| 582 |
+
"grad_norm": 1.487726092338562,
|
| 583 |
+
"learning_rate": 1.7437745318834054e-05,
|
| 584 |
+
"loss": 0.2502,
|
| 585 |
+
"step": 40500
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 1.3190541301375371,
|
| 589 |
+
"grad_norm": 1.4628674983978271,
|
| 590 |
+
"learning_rate": 1.7035583295798213e-05,
|
| 591 |
+
"loss": 0.2437,
|
| 592 |
+
"step": 41000
|
| 593 |
+
},
|
| 594 |
+
{
|
| 595 |
+
"epoch": 1.3351403522882652,
|
| 596 |
+
"grad_norm": 1.401607632637024,
|
| 597 |
+
"learning_rate": 1.663342127276237e-05,
|
| 598 |
+
"loss": 0.2421,
|
| 599 |
+
"step": 41500
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 1.351226574438993,
|
| 603 |
+
"grad_norm": 1.1497563123703003,
|
| 604 |
+
"learning_rate": 1.623125924972653e-05,
|
| 605 |
+
"loss": 0.231,
|
| 606 |
+
"step": 42000
|
| 607 |
+
},
|
| 608 |
+
{
|
| 609 |
+
"epoch": 1.367312796589721,
|
| 610 |
+
"grad_norm": 1.322836995124817,
|
| 611 |
+
"learning_rate": 1.5829097226690688e-05,
|
| 612 |
+
"loss": 0.2261,
|
| 613 |
+
"step": 42500
|
| 614 |
+
},
|
| 615 |
+
{
|
| 616 |
+
"epoch": 1.3833990187404488,
|
| 617 |
+
"grad_norm": 1.5328525304794312,
|
| 618 |
+
"learning_rate": 1.542773952770092e-05,
|
| 619 |
+
"loss": 0.2177,
|
| 620 |
+
"step": 43000
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 1.3994852408911767,
|
| 624 |
+
"grad_norm": 1.7748241424560547,
|
| 625 |
+
"learning_rate": 1.502557750466508e-05,
|
| 626 |
+
"loss": 0.2186,
|
| 627 |
+
"step": 43500
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 1.4155714630419047,
|
| 631 |
+
"grad_norm": 1.6542141437530518,
|
| 632 |
+
"learning_rate": 1.4623415481629241e-05,
|
| 633 |
+
"loss": 0.2138,
|
| 634 |
+
"step": 44000
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 1.4316576851926326,
|
| 638 |
+
"grad_norm": 1.3098843097686768,
|
| 639 |
+
"learning_rate": 1.4221253458593397e-05,
|
| 640 |
+
"loss": 0.211,
|
| 641 |
+
"step": 44500
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 1.4477439073433604,
|
| 645 |
+
"grad_norm": 1.345651626586914,
|
| 646 |
+
"learning_rate": 1.3819091435557557e-05,
|
| 647 |
+
"loss": 0.2027,
|
| 648 |
+
"step": 45000
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 1.4638301294940883,
|
| 652 |
+
"grad_norm": 1.4520297050476074,
|
| 653 |
+
"learning_rate": 1.3416929412521718e-05,
|
| 654 |
+
"loss": 0.2039,
|
| 655 |
+
"step": 45500
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 1.4799163516448162,
|
| 659 |
+
"grad_norm": 1.5913499593734741,
|
| 660 |
+
"learning_rate": 1.3014767389485877e-05,
|
| 661 |
+
"loss": 0.1939,
|
| 662 |
+
"step": 46000
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 1.4960025737955442,
|
| 666 |
+
"grad_norm": 1.1803226470947266,
|
| 667 |
+
"learning_rate": 1.2612605366450037e-05,
|
| 668 |
+
"loss": 0.1887,
|
| 669 |
+
"step": 46500
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 1.5120887959462719,
|
| 673 |
+
"grad_norm": 1.1462236642837524,
|
| 674 |
+
"learning_rate": 1.2210443343414194e-05,
|
| 675 |
+
"loss": 0.1883,
|
| 676 |
+
"step": 47000
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"epoch": 1.528175018097,
|
| 680 |
+
"grad_norm": 0.8483968377113342,
|
| 681 |
+
"learning_rate": 1.1808281320378355e-05,
|
| 682 |
+
"loss": 0.1809,
|
| 683 |
+
"step": 47500
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 1.5442612402477278,
|
| 687 |
+
"grad_norm": 1.1205823421478271,
|
| 688 |
+
"learning_rate": 1.1406119297342515e-05,
|
| 689 |
+
"loss": 0.1813,
|
| 690 |
+
"step": 48000
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"epoch": 1.5603474623984557,
|
| 694 |
+
"grad_norm": 1.417622447013855,
|
| 695 |
+
"learning_rate": 1.1003957274306672e-05,
|
| 696 |
+
"loss": 0.1788,
|
| 697 |
+
"step": 48500
|
| 698 |
+
},
|
| 699 |
+
{
|
| 700 |
+
"epoch": 1.5764336845491838,
|
| 701 |
+
"grad_norm": 1.179103970527649,
|
| 702 |
+
"learning_rate": 1.0602599575316904e-05,
|
| 703 |
+
"loss": 0.1809,
|
| 704 |
+
"step": 49000
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 1.5925199066999114,
|
| 708 |
+
"grad_norm": 1.1092889308929443,
|
| 709 |
+
"learning_rate": 1.0200437552281065e-05,
|
| 710 |
+
"loss": 0.1734,
|
| 711 |
+
"step": 49500
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 1.6086061288506395,
|
| 715 |
+
"grad_norm": 1.0196574926376343,
|
| 716 |
+
"learning_rate": 9.798275529245222e-06,
|
| 717 |
+
"loss": 0.1688,
|
| 718 |
+
"step": 50000
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 1.6246923510013673,
|
| 722 |
+
"grad_norm": 1.1376862525939941,
|
| 723 |
+
"learning_rate": 9.396113506209382e-06,
|
| 724 |
+
"loss": 0.1703,
|
| 725 |
+
"step": 50500
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 1.6407785731520952,
|
| 729 |
+
"grad_norm": 0.8885149955749512,
|
| 730 |
+
"learning_rate": 8.995560131265685e-06,
|
| 731 |
+
"loss": 0.1691,
|
| 732 |
+
"step": 51000
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 1.6568647953028233,
|
| 736 |
+
"grad_norm": 1.2574944496154785,
|
| 737 |
+
"learning_rate": 8.593398108229844e-06,
|
| 738 |
+
"loss": 0.1615,
|
| 739 |
+
"step": 51500
|
| 740 |
+
},
|
| 741 |
+
{
|
| 742 |
+
"epoch": 1.672951017453551,
|
| 743 |
+
"grad_norm": 1.2620723247528076,
|
| 744 |
+
"learning_rate": 8.191236085194004e-06,
|
| 745 |
+
"loss": 0.1593,
|
| 746 |
+
"step": 52000
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 1.689037239604279,
|
| 750 |
+
"grad_norm": 1.551480770111084,
|
| 751 |
+
"learning_rate": 7.789074062158163e-06,
|
| 752 |
+
"loss": 0.1639,
|
| 753 |
+
"step": 52500
|
| 754 |
+
},
|
| 755 |
+
{
|
| 756 |
+
"epoch": 1.7051234617550068,
|
| 757 |
+
"grad_norm": 1.5938962697982788,
|
| 758 |
+
"learning_rate": 7.386912039122322e-06,
|
| 759 |
+
"loss": 0.1587,
|
| 760 |
+
"step": 53000
|
| 761 |
+
},
|
| 762 |
+
{
|
| 763 |
+
"epoch": 1.7212096839057347,
|
| 764 |
+
"grad_norm": 1.0503953695297241,
|
| 765 |
+
"learning_rate": 6.984750016086482e-06,
|
| 766 |
+
"loss": 0.1599,
|
| 767 |
+
"step": 53500
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 1.7372959060564628,
|
| 771 |
+
"grad_norm": 1.1205036640167236,
|
| 772 |
+
"learning_rate": 6.583392317096712e-06,
|
| 773 |
+
"loss": 0.1541,
|
| 774 |
+
"step": 54000
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 1.7533821282071904,
|
| 778 |
+
"grad_norm": 0.7524433732032776,
|
| 779 |
+
"learning_rate": 6.181230294060872e-06,
|
| 780 |
+
"loss": 0.1521,
|
| 781 |
+
"step": 54500
|
| 782 |
+
},
|
| 783 |
+
{
|
| 784 |
+
"epoch": 1.7694683503579185,
|
| 785 |
+
"grad_norm": 0.9619775414466858,
|
| 786 |
+
"learning_rate": 5.779068271025031e-06,
|
| 787 |
+
"loss": 0.1521,
|
| 788 |
+
"step": 55000
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 1.7855545725086464,
|
| 792 |
+
"grad_norm": 0.9406844973564148,
|
| 793 |
+
"learning_rate": 5.37690624798919e-06,
|
| 794 |
+
"loss": 0.1509,
|
| 795 |
+
"step": 55500
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 1.8016407946593742,
|
| 799 |
+
"grad_norm": 0.9363726377487183,
|
| 800 |
+
"learning_rate": 4.975548548999421e-06,
|
| 801 |
+
"loss": 0.1513,
|
| 802 |
+
"step": 56000
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 1.8177270168101023,
|
| 806 |
+
"grad_norm": 0.9941402673721313,
|
| 807 |
+
"learning_rate": 4.573386525963581e-06,
|
| 808 |
+
"loss": 0.1484,
|
| 809 |
+
"step": 56500
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 1.83381323896083,
|
| 813 |
+
"grad_norm": 1.3756345510482788,
|
| 814 |
+
"learning_rate": 4.17122450292774e-06,
|
| 815 |
+
"loss": 0.1509,
|
| 816 |
+
"step": 57000
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 1.849899461111558,
|
| 820 |
+
"grad_norm": 1.0644595623016357,
|
| 821 |
+
"learning_rate": 3.7690624798918986e-06,
|
| 822 |
+
"loss": 0.1486,
|
| 823 |
+
"step": 57500
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"epoch": 1.8659856832622859,
|
| 827 |
+
"grad_norm": 1.070890188217163,
|
| 828 |
+
"learning_rate": 3.3669004568560584e-06,
|
| 829 |
+
"loss": 0.1462,
|
| 830 |
+
"step": 58000
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 1.8820719054130137,
|
| 834 |
+
"grad_norm": 1.3034768104553223,
|
| 835 |
+
"learning_rate": 2.9647384338202173e-06,
|
| 836 |
+
"loss": 0.1481,
|
| 837 |
+
"step": 58500
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 1.8981581275637418,
|
| 841 |
+
"grad_norm": 1.127517580986023,
|
| 842 |
+
"learning_rate": 2.5625764107843767e-06,
|
| 843 |
+
"loss": 0.1451,
|
| 844 |
+
"step": 59000
|
| 845 |
+
},
|
| 846 |
+
{
|
| 847 |
+
"epoch": 1.9142443497144694,
|
| 848 |
+
"grad_norm": 0.9431403279304504,
|
| 849 |
+
"learning_rate": 2.1604143877485364e-06,
|
| 850 |
+
"loss": 0.1458,
|
| 851 |
+
"step": 59500
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 1.9303305718651975,
|
| 855 |
+
"grad_norm": 1.271483302116394,
|
| 856 |
+
"learning_rate": 1.7590566887587673e-06,
|
| 857 |
+
"loss": 0.1463,
|
| 858 |
+
"step": 60000
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 1.9464167940159254,
|
| 862 |
+
"grad_norm": 0.7327952980995178,
|
| 863 |
+
"learning_rate": 1.3568946657229264e-06,
|
| 864 |
+
"loss": 0.1434,
|
| 865 |
+
"step": 60500
|
| 866 |
+
},
|
| 867 |
+
{
|
| 868 |
+
"epoch": 1.9625030161666532,
|
| 869 |
+
"grad_norm": 1.0670543909072876,
|
| 870 |
+
"learning_rate": 9.547326426870858e-07,
|
| 871 |
+
"loss": 0.1424,
|
| 872 |
+
"step": 61000
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 1.9785892383173813,
|
| 876 |
+
"grad_norm": 1.2705425024032593,
|
| 877 |
+
"learning_rate": 5.525706196512451e-07,
|
| 878 |
+
"loss": 0.1431,
|
| 879 |
+
"step": 61500
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 1.994675460468109,
|
| 883 |
+
"grad_norm": 0.9267213344573975,
|
| 884 |
+
"learning_rate": 1.5040859661540443e-07,
|
| 885 |
+
"loss": 0.1418,
|
| 886 |
+
"step": 62000
|
| 887 |
+
}
|
| 888 |
+
],
|
| 889 |
+
"logging_steps": 500,
|
| 890 |
+
"max_steps": 62164,
|
| 891 |
+
"num_input_tokens_seen": 0,
|
| 892 |
+
"num_train_epochs": 2,
|
| 893 |
+
"save_steps": 500,
|
| 894 |
+
"stateful_callbacks": {
|
| 895 |
+
"TrainerControl": {
|
| 896 |
+
"args": {
|
| 897 |
+
"should_epoch_stop": false,
|
| 898 |
+
"should_evaluate": false,
|
| 899 |
+
"should_log": false,
|
| 900 |
+
"should_save": true,
|
| 901 |
+
"should_training_stop": true
|
| 902 |
+
},
|
| 903 |
+
"attributes": {}
|
| 904 |
+
}
|
| 905 |
+
},
|
| 906 |
+
"total_flos": 1.3461207231391334e+17,
|
| 907 |
+
"train_batch_size": 32,
|
| 908 |
+
"trial_name": null,
|
| 909 |
+
"trial_params": null
|
| 910 |
+
}
|
checkpoints/{checkpoint-57726 → checkpoint-62164}/training_args.bin
RENAMED
|
File without changes
|
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 258368552
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7e8c8770f6dcbe1b9d9475af0bda9b1c3d8b183e7518931c4ee48457079834ba
|
| 3 |
size 258368552
|
spiece.model
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:610d79d092a886e9af7d31b94512c46fd20a5f62557bd5b8d8a5b23f1f78650a
|
| 3 |
+
size 1042971
|
src/data/generate_cyr_lat_pairs.py
CHANGED
|
@@ -110,7 +110,7 @@ with open(output_path, 'w', encoding = "utf-8") as out_file:
|
|
| 110 |
|
| 111 |
with open(output_path, 'a', encoding = "utf-8") as out_file:
|
| 112 |
with open("src/data/kk.txt", 'r', encoding = "utf-8") as f:
|
| 113 |
-
for line in tqdm(islice(f,
|
| 114 |
try:
|
| 115 |
cyr_text = line.strip()
|
| 116 |
lat_text = convert_to_latin(cyr_text).strip()
|
|
|
|
| 110 |
|
| 111 |
with open(output_path, 'a', encoding = "utf-8") as out_file:
|
| 112 |
with open("src/data/kk.txt", 'r', encoding = "utf-8") as f:
|
| 113 |
+
for line in tqdm(islice(f, 2_200_000), total = 2_200_000, desc = "Lines in CC100-Kazakh"):
|
| 114 |
try:
|
| 115 |
cyr_text = line.strip()
|
| 116 |
lat_text = convert_to_latin(cyr_text).strip()
|
src/tokeniser/dalat5_sp.model
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:610d79d092a886e9af7d31b94512c46fd20a5f62557bd5b8d8a5b23f1f78650a
|
| 3 |
+
size 1042971
|
src/tokeniser/dalat5_sp.vocab
CHANGED
|
The diff for this file is too large to render.
See raw diff
|
|
|
src/tokeniser/spiece.model
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:610d79d092a886e9af7d31b94512c46fd20a5f62557bd5b8d8a5b23f1f78650a
|
| 3 |
+
size 1042971
|
src/train_tokeniser.py
CHANGED
|
@@ -26,6 +26,7 @@ spm.SentencePieceTrainer.Train(
|
|
| 26 |
vocab_size = 40000,
|
| 27 |
model_type = "unigram", # worth testing with "bpe"
|
| 28 |
character_coverage = 1.0, # to preserve rare characters like ä, ñ, etc.
|
|
|
|
| 29 |
pad_id = 0,
|
| 30 |
unk_id = 1,
|
| 31 |
bos_id = 2,
|
|
|
|
| 26 |
vocab_size = 40000,
|
| 27 |
model_type = "unigram", # worth testing with "bpe"
|
| 28 |
character_coverage = 1.0, # to preserve rare characters like ä, ñ, etc.
|
| 29 |
+
max_sentence_length = 8384,
|
| 30 |
pad_id = 0,
|
| 31 |
unk_id = 1,
|
| 32 |
bos_id = 2,
|