Training checkpoint at step 1000
Browse files- trainer_state.json +369 -8
trainer_state.json
CHANGED
|
@@ -1,19 +1,380 @@
|
|
| 1 |
{
|
| 2 |
-
"best_global_step":
|
| 3 |
-
"best_metric":
|
| 4 |
-
"best_model_checkpoint":
|
| 5 |
-
"epoch":
|
| 6 |
"eval_steps": 100,
|
| 7 |
-
"global_step":
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
| 11 |
-
"log_history": [
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 12 |
"logging_steps": 25,
|
| 13 |
"max_steps": 50000,
|
| 14 |
"num_input_tokens_seen": 0,
|
| 15 |
"num_train_epochs": 9223372036854775807,
|
| 16 |
-
"save_steps":
|
| 17 |
"stateful_callbacks": {
|
| 18 |
"TrainerControl": {
|
| 19 |
"args": {
|
|
@@ -26,7 +387,7 @@
|
|
| 26 |
"attributes": {}
|
| 27 |
}
|
| 28 |
},
|
| 29 |
-
"total_flos":
|
| 30 |
"train_batch_size": 1,
|
| 31 |
"trial_name": null,
|
| 32 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_global_step": 1000,
|
| 3 |
+
"best_metric": 2.463880777359009,
|
| 4 |
+
"best_model_checkpoint": "../runs/karpathy/fineweb-edu-100b-shuffle/meta-llama/Llama-3.2-1B/linear_adamw_wd1e-03_8x1024_mem32_bs64_hf_armt_dmem64/run_20/checkpoint-1000",
|
| 5 |
+
"epoch": 0.02,
|
| 6 |
"eval_steps": 100,
|
| 7 |
+
"global_step": 1000,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.0005,
|
| 14 |
+
"grad_norm": 39.75564521032967,
|
| 15 |
+
"learning_rate": 4.8e-08,
|
| 16 |
+
"loss": 3.6517,
|
| 17 |
+
"step": 25
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.001,
|
| 21 |
+
"grad_norm": 28.937531835097435,
|
| 22 |
+
"learning_rate": 9.8e-08,
|
| 23 |
+
"loss": 3.5931,
|
| 24 |
+
"step": 50
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.0015,
|
| 28 |
+
"grad_norm": 21.922720332659644,
|
| 29 |
+
"learning_rate": 1.4800000000000003e-07,
|
| 30 |
+
"loss": 3.3397,
|
| 31 |
+
"step": 75
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.002,
|
| 35 |
+
"grad_norm": 8.739610199908325,
|
| 36 |
+
"learning_rate": 1.9800000000000003e-07,
|
| 37 |
+
"loss": 3.1289,
|
| 38 |
+
"step": 100
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.002,
|
| 42 |
+
"eval_loss": 2.9243295192718506,
|
| 43 |
+
"eval_runtime": 264.3302,
|
| 44 |
+
"eval_samples_per_second": 3.11,
|
| 45 |
+
"eval_steps_per_second": 1.555,
|
| 46 |
+
"step": 100
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 0.0025,
|
| 50 |
+
"grad_norm": 4.433912600039677,
|
| 51 |
+
"learning_rate": 2.48e-07,
|
| 52 |
+
"loss": 2.8957,
|
| 53 |
+
"step": 125
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 0.003,
|
| 57 |
+
"grad_norm": 3.2874790066620303,
|
| 58 |
+
"learning_rate": 2.9800000000000005e-07,
|
| 59 |
+
"loss": 2.763,
|
| 60 |
+
"step": 150
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 0.0035,
|
| 64 |
+
"grad_norm": 1.5203472215469231,
|
| 65 |
+
"learning_rate": 3.48e-07,
|
| 66 |
+
"loss": 2.676,
|
| 67 |
+
"step": 175
|
| 68 |
+
},
|
| 69 |
+
{
|
| 70 |
+
"epoch": 0.004,
|
| 71 |
+
"grad_norm": 1.1945541683905954,
|
| 72 |
+
"learning_rate": 3.9800000000000004e-07,
|
| 73 |
+
"loss": 2.635,
|
| 74 |
+
"step": 200
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.004,
|
| 78 |
+
"eval_loss": 2.6094932556152344,
|
| 79 |
+
"eval_runtime": 265.7702,
|
| 80 |
+
"eval_samples_per_second": 3.093,
|
| 81 |
+
"eval_steps_per_second": 1.546,
|
| 82 |
+
"step": 200
|
| 83 |
+
},
|
| 84 |
+
{
|
| 85 |
+
"epoch": 0.0045,
|
| 86 |
+
"grad_norm": 1.0852713304633745,
|
| 87 |
+
"learning_rate": 4.4800000000000004e-07,
|
| 88 |
+
"loss": 2.6016,
|
| 89 |
+
"step": 225
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 0.005,
|
| 93 |
+
"grad_norm": 1.0733940346699529,
|
| 94 |
+
"learning_rate": 4.98e-07,
|
| 95 |
+
"loss": 2.5797,
|
| 96 |
+
"step": 250
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 0.0055,
|
| 100 |
+
"grad_norm": 0.9273949035031271,
|
| 101 |
+
"learning_rate": 5.480000000000001e-07,
|
| 102 |
+
"loss": 2.5607,
|
| 103 |
+
"step": 275
|
| 104 |
+
},
|
| 105 |
+
{
|
| 106 |
+
"epoch": 0.006,
|
| 107 |
+
"grad_norm": 0.9289300678591714,
|
| 108 |
+
"learning_rate": 5.98e-07,
|
| 109 |
+
"loss": 2.552,
|
| 110 |
+
"step": 300
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.006,
|
| 114 |
+
"eval_loss": 2.541522264480591,
|
| 115 |
+
"eval_runtime": 266.7478,
|
| 116 |
+
"eval_samples_per_second": 3.082,
|
| 117 |
+
"eval_steps_per_second": 1.541,
|
| 118 |
+
"step": 300
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 0.0065,
|
| 122 |
+
"grad_norm": 1.1328584507449984,
|
| 123 |
+
"learning_rate": 6.48e-07,
|
| 124 |
+
"loss": 2.5402,
|
| 125 |
+
"step": 325
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 0.007,
|
| 129 |
+
"grad_norm": 0.8593307029257858,
|
| 130 |
+
"learning_rate": 6.98e-07,
|
| 131 |
+
"loss": 2.5286,
|
| 132 |
+
"step": 350
|
| 133 |
+
},
|
| 134 |
+
{
|
| 135 |
+
"epoch": 0.0075,
|
| 136 |
+
"grad_norm": 0.895615604067586,
|
| 137 |
+
"learning_rate": 7.480000000000001e-07,
|
| 138 |
+
"loss": 2.5311,
|
| 139 |
+
"step": 375
|
| 140 |
+
},
|
| 141 |
+
{
|
| 142 |
+
"epoch": 0.008,
|
| 143 |
+
"grad_norm": 0.912306580242149,
|
| 144 |
+
"learning_rate": 7.98e-07,
|
| 145 |
+
"loss": 2.5037,
|
| 146 |
+
"step": 400
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.008,
|
| 150 |
+
"eval_loss": 2.514389991760254,
|
| 151 |
+
"eval_runtime": 266.4899,
|
| 152 |
+
"eval_samples_per_second": 3.085,
|
| 153 |
+
"eval_steps_per_second": 1.542,
|
| 154 |
+
"step": 400
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 0.0085,
|
| 158 |
+
"grad_norm": 1.1866535514670034,
|
| 159 |
+
"learning_rate": 8.480000000000001e-07,
|
| 160 |
+
"loss": 2.5011,
|
| 161 |
+
"step": 425
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 0.009,
|
| 165 |
+
"grad_norm": 1.211342504193914,
|
| 166 |
+
"learning_rate": 8.980000000000001e-07,
|
| 167 |
+
"loss": 2.503,
|
| 168 |
+
"step": 450
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"epoch": 0.0095,
|
| 172 |
+
"grad_norm": 1.113763817383069,
|
| 173 |
+
"learning_rate": 9.480000000000001e-07,
|
| 174 |
+
"loss": 2.4999,
|
| 175 |
+
"step": 475
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"epoch": 0.01,
|
| 179 |
+
"grad_norm": 1.2585585589647226,
|
| 180 |
+
"learning_rate": 9.98e-07,
|
| 181 |
+
"loss": 2.4872,
|
| 182 |
+
"step": 500
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.01,
|
| 186 |
+
"eval_loss": 2.497868061065674,
|
| 187 |
+
"eval_runtime": 265.7962,
|
| 188 |
+
"eval_samples_per_second": 3.093,
|
| 189 |
+
"eval_steps_per_second": 1.546,
|
| 190 |
+
"step": 500
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 0.0105,
|
| 194 |
+
"grad_norm": 1.2585825718084245,
|
| 195 |
+
"learning_rate": 1.0480000000000002e-06,
|
| 196 |
+
"loss": 2.4852,
|
| 197 |
+
"step": 525
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 0.011,
|
| 201 |
+
"grad_norm": 1.4101257437846046,
|
| 202 |
+
"learning_rate": 1.0980000000000001e-06,
|
| 203 |
+
"loss": 2.4892,
|
| 204 |
+
"step": 550
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.0115,
|
| 208 |
+
"grad_norm": 1.1975234150707363,
|
| 209 |
+
"learning_rate": 1.148e-06,
|
| 210 |
+
"loss": 2.4861,
|
| 211 |
+
"step": 575
|
| 212 |
+
},
|
| 213 |
+
{
|
| 214 |
+
"epoch": 0.012,
|
| 215 |
+
"grad_norm": 1.3662769225582332,
|
| 216 |
+
"learning_rate": 1.1980000000000002e-06,
|
| 217 |
+
"loss": 2.4882,
|
| 218 |
+
"step": 600
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.012,
|
| 222 |
+
"eval_loss": 2.4879231452941895,
|
| 223 |
+
"eval_runtime": 267.0005,
|
| 224 |
+
"eval_samples_per_second": 3.079,
|
| 225 |
+
"eval_steps_per_second": 1.539,
|
| 226 |
+
"step": 600
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.0125,
|
| 230 |
+
"grad_norm": 1.3086724275194024,
|
| 231 |
+
"learning_rate": 1.248e-06,
|
| 232 |
+
"loss": 2.4745,
|
| 233 |
+
"step": 625
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.013,
|
| 237 |
+
"grad_norm": 1.317023206802888,
|
| 238 |
+
"learning_rate": 1.2980000000000001e-06,
|
| 239 |
+
"loss": 2.4727,
|
| 240 |
+
"step": 650
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.0135,
|
| 244 |
+
"grad_norm": 1.5284967544483212,
|
| 245 |
+
"learning_rate": 1.348e-06,
|
| 246 |
+
"loss": 2.469,
|
| 247 |
+
"step": 675
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.014,
|
| 251 |
+
"grad_norm": 1.1047595217316941,
|
| 252 |
+
"learning_rate": 1.3980000000000002e-06,
|
| 253 |
+
"loss": 2.4695,
|
| 254 |
+
"step": 700
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.014,
|
| 258 |
+
"eval_loss": 2.480103015899658,
|
| 259 |
+
"eval_runtime": 263.5022,
|
| 260 |
+
"eval_samples_per_second": 3.12,
|
| 261 |
+
"eval_steps_per_second": 1.56,
|
| 262 |
+
"step": 700
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.0145,
|
| 266 |
+
"grad_norm": 1.2077328209863791,
|
| 267 |
+
"learning_rate": 1.4480000000000002e-06,
|
| 268 |
+
"loss": 2.4654,
|
| 269 |
+
"step": 725
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.015,
|
| 273 |
+
"grad_norm": 1.209220841771836,
|
| 274 |
+
"learning_rate": 1.498e-06,
|
| 275 |
+
"loss": 2.4663,
|
| 276 |
+
"step": 750
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.0155,
|
| 280 |
+
"grad_norm": 1.3063169829879686,
|
| 281 |
+
"learning_rate": 1.548e-06,
|
| 282 |
+
"loss": 2.4704,
|
| 283 |
+
"step": 775
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.016,
|
| 287 |
+
"grad_norm": 1.3180183352683195,
|
| 288 |
+
"learning_rate": 1.5980000000000002e-06,
|
| 289 |
+
"loss": 2.4583,
|
| 290 |
+
"step": 800
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.016,
|
| 294 |
+
"eval_loss": 2.473590850830078,
|
| 295 |
+
"eval_runtime": 305.9875,
|
| 296 |
+
"eval_samples_per_second": 2.686,
|
| 297 |
+
"eval_steps_per_second": 1.343,
|
| 298 |
+
"step": 800
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.0165,
|
| 302 |
+
"grad_norm": 1.1674852380778837,
|
| 303 |
+
"learning_rate": 1.6480000000000001e-06,
|
| 304 |
+
"loss": 2.467,
|
| 305 |
+
"step": 825
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 0.017,
|
| 309 |
+
"grad_norm": 1.2497656349941002,
|
| 310 |
+
"learning_rate": 1.6980000000000003e-06,
|
| 311 |
+
"loss": 2.4612,
|
| 312 |
+
"step": 850
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 0.0175,
|
| 316 |
+
"grad_norm": 1.3358614980967494,
|
| 317 |
+
"learning_rate": 1.7480000000000002e-06,
|
| 318 |
+
"loss": 2.4636,
|
| 319 |
+
"step": 875
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 0.018,
|
| 323 |
+
"grad_norm": 1.252489857653356,
|
| 324 |
+
"learning_rate": 1.798e-06,
|
| 325 |
+
"loss": 2.454,
|
| 326 |
+
"step": 900
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 0.018,
|
| 330 |
+
"eval_loss": 2.4681763648986816,
|
| 331 |
+
"eval_runtime": 264.702,
|
| 332 |
+
"eval_samples_per_second": 3.105,
|
| 333 |
+
"eval_steps_per_second": 1.553,
|
| 334 |
+
"step": 900
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 0.0185,
|
| 338 |
+
"grad_norm": 1.2815437998994337,
|
| 339 |
+
"learning_rate": 1.8480000000000001e-06,
|
| 340 |
+
"loss": 2.4571,
|
| 341 |
+
"step": 925
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 0.019,
|
| 345 |
+
"grad_norm": 1.0902475329451575,
|
| 346 |
+
"learning_rate": 1.898e-06,
|
| 347 |
+
"loss": 2.451,
|
| 348 |
+
"step": 950
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 0.0195,
|
| 352 |
+
"grad_norm": 1.1502696024965324,
|
| 353 |
+
"learning_rate": 1.9480000000000002e-06,
|
| 354 |
+
"loss": 2.4527,
|
| 355 |
+
"step": 975
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 0.02,
|
| 359 |
+
"grad_norm": 1.2336661855806117,
|
| 360 |
+
"learning_rate": 1.998e-06,
|
| 361 |
+
"loss": 2.4496,
|
| 362 |
+
"step": 1000
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.02,
|
| 366 |
+
"eval_loss": 2.463880777359009,
|
| 367 |
+
"eval_runtime": 275.7426,
|
| 368 |
+
"eval_samples_per_second": 2.981,
|
| 369 |
+
"eval_steps_per_second": 1.491,
|
| 370 |
+
"step": 1000
|
| 371 |
+
}
|
| 372 |
+
],
|
| 373 |
"logging_steps": 25,
|
| 374 |
"max_steps": 50000,
|
| 375 |
"num_input_tokens_seen": 0,
|
| 376 |
"num_train_epochs": 9223372036854775807,
|
| 377 |
+
"save_steps": 1000,
|
| 378 |
"stateful_callbacks": {
|
| 379 |
"TrainerControl": {
|
| 380 |
"args": {
|
|
|
|
| 387 |
"attributes": {}
|
| 388 |
}
|
| 389 |
},
|
| 390 |
+
"total_flos": 3.183202298327204e+18,
|
| 391 |
"train_batch_size": 1,
|
| 392 |
"trial_name": null,
|
| 393 |
"trial_params": null
|