Upload checkpoint-600/trainer_state.json with huggingface_hub
Browse files
checkpoint-600/trainer_state.json
ADDED
|
@@ -0,0 +1,514 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 3.2609637479559113,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 600,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.054367447490814874,
|
| 14 |
+
"grad_norm": 2.0497019290924072,
|
| 15 |
+
"learning_rate": 2.2727272727272728e-06,
|
| 16 |
+
"loss": 0.3348,
|
| 17 |
+
"mean_token_accuracy": 0.036216020046958876,
|
| 18 |
+
"step": 10
|
| 19 |
+
},
|
| 20 |
+
{
|
| 21 |
+
"epoch": 0.10873489498162975,
|
| 22 |
+
"grad_norm": 3.0014805793762207,
|
| 23 |
+
"learning_rate": 4.5454545454545455e-06,
|
| 24 |
+
"loss": 0.2845,
|
| 25 |
+
"mean_token_accuracy": 0.03815823743243527,
|
| 26 |
+
"step": 20
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.16310234247244462,
|
| 30 |
+
"grad_norm": 2.813481569290161,
|
| 31 |
+
"learning_rate": 4.998433870444026e-06,
|
| 32 |
+
"loss": 0.2715,
|
| 33 |
+
"mean_token_accuracy": 0.03921769789994869,
|
| 34 |
+
"step": 30
|
| 35 |
+
},
|
| 36 |
+
{
|
| 37 |
+
"epoch": 0.2174697899632595,
|
| 38 |
+
"grad_norm": 1.4481842517852783,
|
| 39 |
+
"learning_rate": 4.992074831939997e-06,
|
| 40 |
+
"loss": 0.2675,
|
| 41 |
+
"mean_token_accuracy": 0.03833619062224898,
|
| 42 |
+
"step": 40
|
| 43 |
+
},
|
| 44 |
+
{
|
| 45 |
+
"epoch": 0.2718372374540744,
|
| 46 |
+
"grad_norm": 1.9870078563690186,
|
| 47 |
+
"learning_rate": 4.980837439924479e-06,
|
| 48 |
+
"loss": 0.2632,
|
| 49 |
+
"mean_token_accuracy": 0.039046796466209344,
|
| 50 |
+
"step": 50
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.32620468494488924,
|
| 54 |
+
"grad_norm": 1.224491834640503,
|
| 55 |
+
"learning_rate": 4.9647436921200514e-06,
|
| 56 |
+
"loss": 0.2593,
|
| 57 |
+
"mean_token_accuracy": 0.04083767885422276,
|
| 58 |
+
"step": 60
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.38057213243570415,
|
| 62 |
+
"grad_norm": 1.2895762920379639,
|
| 63 |
+
"learning_rate": 4.943825092793806e-06,
|
| 64 |
+
"loss": 0.2574,
|
| 65 |
+
"mean_token_accuracy": 0.040451897433194973,
|
| 66 |
+
"step": 70
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.434939579926519,
|
| 70 |
+
"grad_norm": 1.4011473655700684,
|
| 71 |
+
"learning_rate": 4.91812259108626e-06,
|
| 72 |
+
"loss": 0.2545,
|
| 73 |
+
"mean_token_accuracy": 0.041019549200973414,
|
| 74 |
+
"step": 80
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.4893070274173339,
|
| 78 |
+
"grad_norm": 1.6724004745483398,
|
| 79 |
+
"learning_rate": 4.887686500851499e-06,
|
| 80 |
+
"loss": 0.2487,
|
| 81 |
+
"mean_token_accuracy": 0.042838540498996734,
|
| 82 |
+
"step": 90
|
| 83 |
+
},
|
| 84 |
+
{
|
| 85 |
+
"epoch": 0.5436744749081488,
|
| 86 |
+
"grad_norm": 1.4560455083847046,
|
| 87 |
+
"learning_rate": 4.852576402165436e-06,
|
| 88 |
+
"loss": 0.2506,
|
| 89 |
+
"mean_token_accuracy": 0.042806161533553676,
|
| 90 |
+
"step": 100
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 0.5980419223989636,
|
| 94 |
+
"grad_norm": 1.4531595706939697,
|
| 95 |
+
"learning_rate": 4.812861024695024e-06,
|
| 96 |
+
"loss": 0.2459,
|
| 97 |
+
"mean_token_accuracy": 0.04187717779204832,
|
| 98 |
+
"step": 110
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.6524093698897785,
|
| 102 |
+
"grad_norm": 1.185799241065979,
|
| 103 |
+
"learning_rate": 4.768618113156695e-06,
|
| 104 |
+
"loss": 0.2475,
|
| 105 |
+
"mean_token_accuracy": 0.04067943783993542,
|
| 106 |
+
"step": 120
|
| 107 |
+
},
|
| 108 |
+
{
|
| 109 |
+
"epoch": 0.7067768173805934,
|
| 110 |
+
"grad_norm": 1.970852017402649,
|
| 111 |
+
"learning_rate": 4.719934275127435e-06,
|
| 112 |
+
"loss": 0.247,
|
| 113 |
+
"mean_token_accuracy": 0.04101934070331481,
|
| 114 |
+
"step": 130
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.7611442648714083,
|
| 118 |
+
"grad_norm": 1.789883017539978,
|
| 119 |
+
"learning_rate": 4.666904811506382e-06,
|
| 120 |
+
"loss": 0.2442,
|
| 121 |
+
"mean_token_accuracy": 0.04314688319063862,
|
| 122 |
+
"step": 140
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.8155117123622231,
|
| 126 |
+
"grad_norm": 1.707210659980774,
|
| 127 |
+
"learning_rate": 4.609633529958841e-06,
|
| 128 |
+
"loss": 0.2453,
|
| 129 |
+
"mean_token_accuracy": 0.04226606881111365,
|
| 130 |
+
"step": 150
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 0.869879159853038,
|
| 134 |
+
"grad_norm": 1.4981927871704102,
|
| 135 |
+
"learning_rate": 4.5482325417079045e-06,
|
| 136 |
+
"loss": 0.2455,
|
| 137 |
+
"mean_token_accuracy": 0.04208922391790111,
|
| 138 |
+
"step": 160
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"epoch": 0.9242466073438529,
|
| 142 |
+
"grad_norm": 1.4780975580215454,
|
| 143 |
+
"learning_rate": 4.482822042071466e-06,
|
| 144 |
+
"loss": 0.242,
|
| 145 |
+
"mean_token_accuracy": 0.043558270454741435,
|
| 146 |
+
"step": 170
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.9786140548346678,
|
| 150 |
+
"grad_norm": 1.8650013208389282,
|
| 151 |
+
"learning_rate": 4.413530075174245e-06,
|
| 152 |
+
"loss": 0.2445,
|
| 153 |
+
"mean_token_accuracy": 0.04371535095342551,
|
| 154 |
+
"step": 180
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 1.0326204684944889,
|
| 158 |
+
"grad_norm": 1.2570964097976685,
|
| 159 |
+
"learning_rate": 4.340492283295396e-06,
|
| 160 |
+
"loss": 0.2383,
|
| 161 |
+
"mean_token_accuracy": 0.04298217990664554,
|
| 162 |
+
"step": 190
|
| 163 |
+
},
|
| 164 |
+
{
|
| 165 |
+
"epoch": 1.0869879159853038,
|
| 166 |
+
"grad_norm": 1.0197902917861938,
|
| 167 |
+
"learning_rate": 4.263851641342383e-06,
|
| 168 |
+
"loss": 0.236,
|
| 169 |
+
"mean_token_accuracy": 0.04432974819592346,
|
| 170 |
+
"step": 200
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 1.1413553634761187,
|
| 174 |
+
"grad_norm": 1.1943126916885376,
|
| 175 |
+
"learning_rate": 4.1837581769708755e-06,
|
| 176 |
+
"loss": 0.238,
|
| 177 |
+
"mean_token_accuracy": 0.043376582934797625,
|
| 178 |
+
"step": 210
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 1.1957228109669336,
|
| 182 |
+
"grad_norm": 1.7524161338806152,
|
| 183 |
+
"learning_rate": 4.100368676898575e-06,
|
| 184 |
+
"loss": 0.2374,
|
| 185 |
+
"mean_token_accuracy": 0.04293899511212658,
|
| 186 |
+
"step": 220
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 1.2500902584577485,
|
| 190 |
+
"grad_norm": 1.563820719718933,
|
| 191 |
+
"learning_rate": 4.013846379987847e-06,
|
| 192 |
+
"loss": 0.2371,
|
| 193 |
+
"mean_token_accuracy": 0.043539136235267506,
|
| 194 |
+
"step": 230
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 1.3044577059485634,
|
| 198 |
+
"grad_norm": 1.222346544265747,
|
| 199 |
+
"learning_rate": 3.924360657697987e-06,
|
| 200 |
+
"loss": 0.2388,
|
| 201 |
+
"mean_token_accuracy": 0.043110936877928906,
|
| 202 |
+
"step": 240
|
| 203 |
+
},
|
| 204 |
+
{
|
| 205 |
+
"epoch": 1.358825153439378,
|
| 206 |
+
"grad_norm": 1.827516794204712,
|
| 207 |
+
"learning_rate": 3.832086682532633e-06,
|
| 208 |
+
"loss": 0.2366,
|
| 209 |
+
"mean_token_accuracy": 0.04413067772566137,
|
| 210 |
+
"step": 250
|
| 211 |
+
},
|
| 212 |
+
{
|
| 213 |
+
"epoch": 1.4131926009301932,
|
| 214 |
+
"grad_norm": 1.1535050868988037,
|
| 215 |
+
"learning_rate": 3.7372050851313597e-06,
|
| 216 |
+
"loss": 0.2317,
|
| 217 |
+
"mean_token_accuracy": 0.04412070568432682,
|
| 218 |
+
"step": 260
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 1.4675600484210078,
|
| 222 |
+
"grad_norm": 2.219189405441284,
|
| 223 |
+
"learning_rate": 3.639901600676725e-06,
|
| 224 |
+
"loss": 0.2293,
|
| 225 |
+
"mean_token_accuracy": 0.043295259247861394,
|
| 226 |
+
"step": 270
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 1.5219274959118227,
|
| 230 |
+
"grad_norm": 1.609122633934021,
|
| 231 |
+
"learning_rate": 3.5403667053089263e-06,
|
| 232 |
+
"loss": 0.2279,
|
| 233 |
+
"mean_token_accuracy": 0.046050211629699335,
|
| 234 |
+
"step": 280
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 1.5762949434026376,
|
| 238 |
+
"grad_norm": 1.0812249183654785,
|
| 239 |
+
"learning_rate": 3.4387952432598102e-06,
|
| 240 |
+
"loss": 0.2327,
|
| 241 |
+
"mean_token_accuracy": 0.04309764919016743,
|
| 242 |
+
"step": 290
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 1.6306623908934526,
|
| 246 |
+
"grad_norm": 1.4412078857421875,
|
| 247 |
+
"learning_rate": 3.3353860454361398e-06,
|
| 248 |
+
"loss": 0.2345,
|
| 249 |
+
"mean_token_accuracy": 0.042526886053747145,
|
| 250 |
+
"step": 300
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 1.6850298383842675,
|
| 254 |
+
"grad_norm": 1.4374982118606567,
|
| 255 |
+
"learning_rate": 3.2303415401987543e-06,
|
| 256 |
+
"loss": 0.2326,
|
| 257 |
+
"mean_token_accuracy": 0.042373882613537715,
|
| 258 |
+
"step": 310
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 1.7393972858750821,
|
| 262 |
+
"grad_norm": 1.3343548774719238,
|
| 263 |
+
"learning_rate": 3.1238673570995526e-06,
|
| 264 |
+
"loss": 0.2308,
|
| 265 |
+
"mean_token_accuracy": 0.04209477992808388,
|
| 266 |
+
"step": 320
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 1.7937647333658973,
|
| 270 |
+
"grad_norm": 1.3959535360336304,
|
| 271 |
+
"learning_rate": 3.0161719243519848e-06,
|
| 272 |
+
"loss": 0.2265,
|
| 273 |
+
"mean_token_accuracy": 0.04400216135891242,
|
| 274 |
+
"step": 330
|
| 275 |
+
},
|
| 276 |
+
{
|
| 277 |
+
"epoch": 1.848132180856712,
|
| 278 |
+
"grad_norm": 1.9636735916137695,
|
| 279 |
+
"learning_rate": 2.907466060823037e-06,
|
| 280 |
+
"loss": 0.227,
|
| 281 |
+
"mean_token_accuracy": 0.044824491128019874,
|
| 282 |
+
"step": 340
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 1.902499628347527,
|
| 286 |
+
"grad_norm": 1.432839274406433,
|
| 287 |
+
"learning_rate": 2.7979625633454005e-06,
|
| 288 |
+
"loss": 0.2281,
|
| 289 |
+
"mean_token_accuracy": 0.043323908064667196,
|
| 290 |
+
"step": 350
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 1.9568670758383417,
|
| 294 |
+
"grad_norm": 1.2604482173919678,
|
| 295 |
+
"learning_rate": 2.6878757901576775e-06,
|
| 296 |
+
"loss": 0.2254,
|
| 297 |
+
"mean_token_accuracy": 0.04304394573555328,
|
| 298 |
+
"step": 360
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 2.010873489498163,
|
| 302 |
+
"grad_norm": 1.0908658504486084,
|
| 303 |
+
"learning_rate": 2.5774212412880636e-06,
|
| 304 |
+
"loss": 0.2256,
|
| 305 |
+
"mean_token_accuracy": 0.04459014333798497,
|
| 306 |
+
"step": 370
|
| 307 |
+
},
|
| 308 |
+
{
|
| 309 |
+
"epoch": 2.0652409369889777,
|
| 310 |
+
"grad_norm": 1.6308084726333618,
|
| 311 |
+
"learning_rate": 2.4668151367029235e-06,
|
| 312 |
+
"loss": 0.2245,
|
| 313 |
+
"mean_token_accuracy": 0.044322372208444,
|
| 314 |
+
"step": 380
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 2.119608384479793,
|
| 318 |
+
"grad_norm": 1.3369026184082031,
|
| 319 |
+
"learning_rate": 2.35627399304605e-06,
|
| 320 |
+
"loss": 0.2188,
|
| 321 |
+
"mean_token_accuracy": 0.0456739331535573,
|
| 322 |
+
"step": 390
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 2.1739758319706075,
|
| 326 |
+
"grad_norm": 1.295975685119629,
|
| 327 |
+
"learning_rate": 2.2460141997971695e-06,
|
| 328 |
+
"loss": 0.2193,
|
| 329 |
+
"mean_token_accuracy": 0.04523389639671223,
|
| 330 |
+
"step": 400
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 2.2283432794614226,
|
| 334 |
+
"grad_norm": 1.357338309288025,
|
| 335 |
+
"learning_rate": 2.1362515956793717e-06,
|
| 336 |
+
"loss": 0.2192,
|
| 337 |
+
"mean_token_accuracy": 0.04524684594507562,
|
| 338 |
+
"step": 410
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 2.2827107269522373,
|
| 342 |
+
"grad_norm": 1.4117406606674194,
|
| 343 |
+
"learning_rate": 2.027201046144677e-06,
|
| 344 |
+
"loss": 0.2191,
|
| 345 |
+
"mean_token_accuracy": 0.044970535605898476,
|
| 346 |
+
"step": 420
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 2.337078174443052,
|
| 350 |
+
"grad_norm": 1.1702640056610107,
|
| 351 |
+
"learning_rate": 1.9190760227648183e-06,
|
| 352 |
+
"loss": 0.2199,
|
| 353 |
+
"mean_token_accuracy": 0.04270703578586108,
|
| 354 |
+
"step": 430
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 2.391445621933867,
|
| 358 |
+
"grad_norm": 1.1582852602005005,
|
| 359 |
+
"learning_rate": 1.8120881853506179e-06,
|
| 360 |
+
"loss": 0.2177,
|
| 361 |
+
"mean_token_accuracy": 0.04548128471906239,
|
| 362 |
+
"step": 440
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 2.445813069424682,
|
| 366 |
+
"grad_norm": 1.424065351486206,
|
| 367 |
+
"learning_rate": 1.7064469676179682e-06,
|
| 368 |
+
"loss": 0.219,
|
| 369 |
+
"mean_token_accuracy": 0.04496145398898079,
|
| 370 |
+
"step": 450
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 2.500180516915497,
|
| 374 |
+
"grad_norm": 1.851121425628662,
|
| 375 |
+
"learning_rate": 1.6023591672114992e-06,
|
| 376 |
+
"loss": 0.2163,
|
| 377 |
+
"mean_token_accuracy": 0.04483195860047999,
|
| 378 |
+
"step": 460
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 2.5545479644063116,
|
| 382 |
+
"grad_norm": 1.0778027772903442,
|
| 383 |
+
"learning_rate": 1.5000285408884734e-06,
|
| 384 |
+
"loss": 0.2163,
|
| 385 |
+
"mean_token_accuracy": 0.04510914630936895,
|
| 386 |
+
"step": 470
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 2.6089154118971267,
|
| 390 |
+
"grad_norm": 1.2224993705749512,
|
| 391 |
+
"learning_rate": 1.3996554056553723e-06,
|
| 392 |
+
"loss": 0.2141,
|
| 393 |
+
"mean_token_accuracy": 0.04724993375784834,
|
| 394 |
+
"step": 480
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 2.6632828593879414,
|
| 398 |
+
"grad_norm": 1.496194839477539,
|
| 399 |
+
"learning_rate": 1.3014362466379407e-06,
|
| 400 |
+
"loss": 0.2193,
|
| 401 |
+
"mean_token_accuracy": 0.045281845046338275,
|
| 402 |
+
"step": 490
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 2.717650306878756,
|
| 406 |
+
"grad_norm": 1.2077394723892212,
|
| 407 |
+
"learning_rate": 1.2055633324523324e-06,
|
| 408 |
+
"loss": 0.2149,
|
| 409 |
+
"mean_token_accuracy": 0.044335345691433756,
|
| 410 |
+
"step": 500
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 2.772017754369571,
|
| 414 |
+
"grad_norm": 1.171938180923462,
|
| 415 |
+
"learning_rate": 1.1122243388302622e-06,
|
| 416 |
+
"loss": 0.2174,
|
| 417 |
+
"mean_token_accuracy": 0.04475436422835628,
|
| 418 |
+
"step": 510
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"epoch": 2.8263852018603863,
|
| 422 |
+
"grad_norm": 1.2405449151992798,
|
| 423 |
+
"learning_rate": 1.0216019812349508e-06,
|
| 424 |
+
"loss": 0.2163,
|
| 425 |
+
"mean_token_accuracy": 0.04440462300572108,
|
| 426 |
+
"step": 520
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 2.880752649351201,
|
| 430 |
+
"grad_norm": 1.1590036153793335,
|
| 431 |
+
"learning_rate": 9.338736571870205e-07,
|
| 432 |
+
"loss": 0.2153,
|
| 433 |
+
"mean_token_accuracy": 0.043993269855855034,
|
| 434 |
+
"step": 530
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 2.9351200968420157,
|
| 438 |
+
"grad_norm": 1.6510144472122192,
|
| 439 |
+
"learning_rate": 8.492110990005228e-07,
|
| 440 |
+
"loss": 0.2127,
|
| 441 |
+
"mean_token_accuracy": 0.045991620956010594,
|
| 442 |
+
"step": 540
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 2.989487544332831,
|
| 446 |
+
"grad_norm": 1.291462779045105,
|
| 447 |
+
"learning_rate": 7.677800376088657e-07,
|
| 448 |
+
"loss": 0.2109,
|
| 449 |
+
"mean_token_accuracy": 0.045755728109907065,
|
| 450 |
+
"step": 550
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 3.043493957992652,
|
| 454 |
+
"grad_norm": 1.295513391494751,
|
| 455 |
+
"learning_rate": 6.897398781387299e-07,
|
| 456 |
+
"loss": 0.209,
|
| 457 |
+
"mean_token_accuracy": 0.04398122681887118,
|
| 458 |
+
"step": 560
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 3.097861405483467,
|
| 462 |
+
"grad_norm": 1.6088683605194092,
|
| 463 |
+
"learning_rate": 6.152433878670485e-07,
|
| 464 |
+
"loss": 0.2053,
|
| 465 |
+
"mean_token_accuracy": 0.04704667659716506,
|
| 466 |
+
"step": 570
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 3.1522288529742815,
|
| 470 |
+
"grad_norm": 1.117344617843628,
|
| 471 |
+
"learning_rate": 5.444363971718875e-07,
|
| 472 |
+
"loss": 0.2127,
|
| 473 |
+
"mean_token_accuracy": 0.04461234014015645,
|
| 474 |
+
"step": 580
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 3.2065963004650966,
|
| 478 |
+
"grad_norm": 1.4021047353744507,
|
| 479 |
+
"learning_rate": 4.774575140626317e-07,
|
| 480 |
+
"loss": 0.2132,
|
| 481 |
+
"mean_token_accuracy": 0.044966605161243935,
|
| 482 |
+
"step": 590
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 3.2609637479559113,
|
| 486 |
+
"grad_norm": 1.2901127338409424,
|
| 487 |
+
"learning_rate": 4.144378528483009e-07,
|
| 488 |
+
"loss": 0.2096,
|
| 489 |
+
"mean_token_accuracy": 0.04288893376269698,
|
| 490 |
+
"step": 600
|
| 491 |
+
}
|
| 492 |
+
],
|
| 493 |
+
"logging_steps": 10,
|
| 494 |
+
"max_steps": 732,
|
| 495 |
+
"num_input_tokens_seen": 0,
|
| 496 |
+
"num_train_epochs": 4,
|
| 497 |
+
"save_steps": 300,
|
| 498 |
+
"stateful_callbacks": {
|
| 499 |
+
"TrainerControl": {
|
| 500 |
+
"args": {
|
| 501 |
+
"should_epoch_stop": false,
|
| 502 |
+
"should_evaluate": false,
|
| 503 |
+
"should_log": false,
|
| 504 |
+
"should_save": true,
|
| 505 |
+
"should_training_stop": false
|
| 506 |
+
},
|
| 507 |
+
"attributes": {}
|
| 508 |
+
}
|
| 509 |
+
},
|
| 510 |
+
"total_flos": 8.169216325675168e+18,
|
| 511 |
+
"train_batch_size": 1,
|
| 512 |
+
"trial_name": null,
|
| 513 |
+
"trial_params": null
|
| 514 |
+
}
|