Upload lora_string_concat_only/checkpoint-900/trainer_state.json with huggingface_hub
Browse files
lora_string_concat_only/checkpoint-900/trainer_state.json
ADDED
|
@@ -0,0 +1,445 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 900,
|
| 3 |
+
"best_metric": 3.339338779449463,
|
| 4 |
+
"best_model_checkpoint": "/scratch/aditya.mishra/lma/models/lora_string_concat_only/checkpoint-900",
|
| 5 |
+
"epoch": 18.0,
|
| 6 |
+
"eval_steps": 75,
|
| 7 |
+
"global_step": 900,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.4,
|
| 14 |
+
"grad_norm": 0.3263323903083801,
|
| 15 |
+
"learning_rate": 8.444444444444446e-06,
|
| 16 |
+
"loss": 7.2756,
|
| 17 |
+
"step": 20
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.8,
|
| 21 |
+
"grad_norm": 0.3472040891647339,
|
| 22 |
+
"learning_rate": 1.7333333333333336e-05,
|
| 23 |
+
"loss": 7.2397,
|
| 24 |
+
"step": 40
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 1.2,
|
| 28 |
+
"grad_norm": 0.4345553517341614,
|
| 29 |
+
"learning_rate": 2.6222222222222226e-05,
|
| 30 |
+
"loss": 7.1572,
|
| 31 |
+
"step": 60
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 1.5,
|
| 35 |
+
"eval_loss": 6.8960700035095215,
|
| 36 |
+
"eval_runtime": 2.53,
|
| 37 |
+
"eval_samples_per_second": 237.152,
|
| 38 |
+
"eval_steps_per_second": 9.881,
|
| 39 |
+
"step": 75
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 1.6,
|
| 43 |
+
"grad_norm": 0.5284088850021362,
|
| 44 |
+
"learning_rate": 3.511111111111111e-05,
|
| 45 |
+
"loss": 6.9823,
|
| 46 |
+
"step": 80
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 2.0,
|
| 50 |
+
"grad_norm": 0.5539721846580505,
|
| 51 |
+
"learning_rate": 4.4000000000000006e-05,
|
| 52 |
+
"loss": 6.6694,
|
| 53 |
+
"step": 100
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 2.4,
|
| 57 |
+
"grad_norm": 0.48029404878616333,
|
| 58 |
+
"learning_rate": 5.2888888888888885e-05,
|
| 59 |
+
"loss": 6.1555,
|
| 60 |
+
"step": 120
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 2.8,
|
| 64 |
+
"grad_norm": 0.4763217568397522,
|
| 65 |
+
"learning_rate": 6.177777777777779e-05,
|
| 66 |
+
"loss": 5.7015,
|
| 67 |
+
"step": 140
|
| 68 |
+
},
|
| 69 |
+
{
|
| 70 |
+
"epoch": 3.0,
|
| 71 |
+
"eval_loss": 5.193756103515625,
|
| 72 |
+
"eval_runtime": 1.1198,
|
| 73 |
+
"eval_samples_per_second": 535.81,
|
| 74 |
+
"eval_steps_per_second": 22.325,
|
| 75 |
+
"step": 150
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 3.2,
|
| 79 |
+
"grad_norm": 0.4804483950138092,
|
| 80 |
+
"learning_rate": 7.066666666666667e-05,
|
| 81 |
+
"loss": 5.2239,
|
| 82 |
+
"step": 160
|
| 83 |
+
},
|
| 84 |
+
{
|
| 85 |
+
"epoch": 3.6,
|
| 86 |
+
"grad_norm": 0.4357590973377228,
|
| 87 |
+
"learning_rate": 7.955555555555556e-05,
|
| 88 |
+
"loss": 4.686,
|
| 89 |
+
"step": 180
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 4.0,
|
| 93 |
+
"grad_norm": 0.3594028353691101,
|
| 94 |
+
"learning_rate": 8.844444444444445e-05,
|
| 95 |
+
"loss": 4.347,
|
| 96 |
+
"step": 200
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 4.4,
|
| 100 |
+
"grad_norm": 0.4409911036491394,
|
| 101 |
+
"learning_rate": 9.733333333333335e-05,
|
| 102 |
+
"loss": 4.122,
|
| 103 |
+
"step": 220
|
| 104 |
+
},
|
| 105 |
+
{
|
| 106 |
+
"epoch": 4.5,
|
| 107 |
+
"eval_loss": 4.00063943862915,
|
| 108 |
+
"eval_runtime": 1.1484,
|
| 109 |
+
"eval_samples_per_second": 522.462,
|
| 110 |
+
"eval_steps_per_second": 21.769,
|
| 111 |
+
"step": 225
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 4.8,
|
| 115 |
+
"grad_norm": 0.3402056396007538,
|
| 116 |
+
"learning_rate": 0.00010622222222222222,
|
| 117 |
+
"loss": 3.963,
|
| 118 |
+
"step": 240
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 5.2,
|
| 122 |
+
"grad_norm": 0.37722334265708923,
|
| 123 |
+
"learning_rate": 0.00011511111111111112,
|
| 124 |
+
"loss": 3.8099,
|
| 125 |
+
"step": 260
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 5.6,
|
| 129 |
+
"grad_norm": 0.344826877117157,
|
| 130 |
+
"learning_rate": 0.000124,
|
| 131 |
+
"loss": 3.6781,
|
| 132 |
+
"step": 280
|
| 133 |
+
},
|
| 134 |
+
{
|
| 135 |
+
"epoch": 6.0,
|
| 136 |
+
"grad_norm": 0.5443089008331299,
|
| 137 |
+
"learning_rate": 0.00013288888888888888,
|
| 138 |
+
"loss": 3.622,
|
| 139 |
+
"step": 300
|
| 140 |
+
},
|
| 141 |
+
{
|
| 142 |
+
"epoch": 6.0,
|
| 143 |
+
"eval_loss": 3.5904574394226074,
|
| 144 |
+
"eval_runtime": 1.2964,
|
| 145 |
+
"eval_samples_per_second": 462.819,
|
| 146 |
+
"eval_steps_per_second": 19.284,
|
| 147 |
+
"step": 300
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 6.4,
|
| 151 |
+
"grad_norm": 0.4413987994194031,
|
| 152 |
+
"learning_rate": 0.00014177777777777778,
|
| 153 |
+
"loss": 3.5478,
|
| 154 |
+
"step": 320
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 6.8,
|
| 158 |
+
"grad_norm": 0.3995497524738312,
|
| 159 |
+
"learning_rate": 0.00015066666666666668,
|
| 160 |
+
"loss": 3.5393,
|
| 161 |
+
"step": 340
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 7.2,
|
| 165 |
+
"grad_norm": 0.6766912937164307,
|
| 166 |
+
"learning_rate": 0.00015955555555555558,
|
| 167 |
+
"loss": 3.4758,
|
| 168 |
+
"step": 360
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"epoch": 7.5,
|
| 172 |
+
"eval_loss": 3.479808807373047,
|
| 173 |
+
"eval_runtime": 1.312,
|
| 174 |
+
"eval_samples_per_second": 457.325,
|
| 175 |
+
"eval_steps_per_second": 19.055,
|
| 176 |
+
"step": 375
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 7.6,
|
| 180 |
+
"grad_norm": 0.5757963061332703,
|
| 181 |
+
"learning_rate": 0.00016844444444444445,
|
| 182 |
+
"loss": 3.4825,
|
| 183 |
+
"step": 380
|
| 184 |
+
},
|
| 185 |
+
{
|
| 186 |
+
"epoch": 8.0,
|
| 187 |
+
"grad_norm": 0.4912727475166321,
|
| 188 |
+
"learning_rate": 0.00017733333333333335,
|
| 189 |
+
"loss": 3.4496,
|
| 190 |
+
"step": 400
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 8.4,
|
| 194 |
+
"grad_norm": 0.7055399417877197,
|
| 195 |
+
"learning_rate": 0.00018622222222222223,
|
| 196 |
+
"loss": 3.4289,
|
| 197 |
+
"step": 420
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 8.8,
|
| 201 |
+
"grad_norm": 0.4926600754261017,
|
| 202 |
+
"learning_rate": 0.0001951111111111111,
|
| 203 |
+
"loss": 3.4198,
|
| 204 |
+
"step": 440
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 9.0,
|
| 208 |
+
"eval_loss": 3.4288127422332764,
|
| 209 |
+
"eval_runtime": 1.264,
|
| 210 |
+
"eval_samples_per_second": 474.677,
|
| 211 |
+
"eval_steps_per_second": 19.778,
|
| 212 |
+
"step": 450
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 9.2,
|
| 216 |
+
"grad_norm": 0.6211638450622559,
|
| 217 |
+
"learning_rate": 0.00019999981118841182,
|
| 218 |
+
"loss": 3.4014,
|
| 219 |
+
"step": 460
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 9.6,
|
| 223 |
+
"grad_norm": 0.4444306194782257,
|
| 224 |
+
"learning_rate": 0.00019999803962868136,
|
| 225 |
+
"loss": 3.3888,
|
| 226 |
+
"step": 480
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 10.0,
|
| 230 |
+
"grad_norm": 0.37307801842689514,
|
| 231 |
+
"learning_rate": 0.00019999440330200995,
|
| 232 |
+
"loss": 3.3927,
|
| 233 |
+
"step": 500
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 10.4,
|
| 237 |
+
"grad_norm": 0.5229858160018921,
|
| 238 |
+
"learning_rate": 0.00019998890227620783,
|
| 239 |
+
"loss": 3.3702,
|
| 240 |
+
"step": 520
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 10.5,
|
| 244 |
+
"eval_loss": 3.3896546363830566,
|
| 245 |
+
"eval_runtime": 1.0711,
|
| 246 |
+
"eval_samples_per_second": 560.186,
|
| 247 |
+
"eval_steps_per_second": 23.341,
|
| 248 |
+
"step": 525
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 10.8,
|
| 252 |
+
"grad_norm": 0.4428179860115051,
|
| 253 |
+
"learning_rate": 0.00019998153665385843,
|
| 254 |
+
"loss": 3.3547,
|
| 255 |
+
"step": 540
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 11.2,
|
| 259 |
+
"grad_norm": 0.44106265902519226,
|
| 260 |
+
"learning_rate": 0.00019997230657231612,
|
| 261 |
+
"loss": 3.3539,
|
| 262 |
+
"step": 560
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 11.6,
|
| 266 |
+
"grad_norm": 0.45065343379974365,
|
| 267 |
+
"learning_rate": 0.00019996121220370374,
|
| 268 |
+
"loss": 3.3313,
|
| 269 |
+
"step": 580
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 12.0,
|
| 273 |
+
"grad_norm": 0.5424341559410095,
|
| 274 |
+
"learning_rate": 0.00019994825375490948,
|
| 275 |
+
"loss": 3.3438,
|
| 276 |
+
"step": 600
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 12.0,
|
| 280 |
+
"eval_loss": 3.372157096862793,
|
| 281 |
+
"eval_runtime": 1.0488,
|
| 282 |
+
"eval_samples_per_second": 572.093,
|
| 283 |
+
"eval_steps_per_second": 23.837,
|
| 284 |
+
"step": 600
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 12.4,
|
| 288 |
+
"grad_norm": 0.496880441904068,
|
| 289 |
+
"learning_rate": 0.00019993343146758297,
|
| 290 |
+
"loss": 3.3136,
|
| 291 |
+
"step": 620
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 12.8,
|
| 295 |
+
"grad_norm": 0.5735253095626831,
|
| 296 |
+
"learning_rate": 0.0001999167456181307,
|
| 297 |
+
"loss": 3.331,
|
| 298 |
+
"step": 640
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 13.2,
|
| 302 |
+
"grad_norm": 0.46690699458122253,
|
| 303 |
+
"learning_rate": 0.000199898196517711,
|
| 304 |
+
"loss": 3.3355,
|
| 305 |
+
"step": 660
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 13.5,
|
| 309 |
+
"eval_loss": 3.3583269119262695,
|
| 310 |
+
"eval_runtime": 1.2011,
|
| 311 |
+
"eval_samples_per_second": 499.558,
|
| 312 |
+
"eval_steps_per_second": 20.815,
|
| 313 |
+
"step": 675
|
| 314 |
+
},
|
| 315 |
+
{
|
| 316 |
+
"epoch": 13.6,
|
| 317 |
+
"grad_norm": 0.702662467956543,
|
| 318 |
+
"learning_rate": 0.00019987778451222815,
|
| 319 |
+
"loss": 3.3135,
|
| 320 |
+
"step": 680
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 14.0,
|
| 324 |
+
"grad_norm": 0.4215978980064392,
|
| 325 |
+
"learning_rate": 0.00019985550998232596,
|
| 326 |
+
"loss": 3.2992,
|
| 327 |
+
"step": 700
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 14.4,
|
| 331 |
+
"grad_norm": 0.6096284985542297,
|
| 332 |
+
"learning_rate": 0.0001998313733433806,
|
| 333 |
+
"loss": 3.3,
|
| 334 |
+
"step": 720
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 14.8,
|
| 338 |
+
"grad_norm": 0.42653387784957886,
|
| 339 |
+
"learning_rate": 0.00019980537504549298,
|
| 340 |
+
"loss": 3.3013,
|
| 341 |
+
"step": 740
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 15.0,
|
| 345 |
+
"eval_loss": 3.351369857788086,
|
| 346 |
+
"eval_runtime": 1.128,
|
| 347 |
+
"eval_samples_per_second": 531.938,
|
| 348 |
+
"eval_steps_per_second": 22.164,
|
| 349 |
+
"step": 750
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 15.2,
|
| 353 |
+
"grad_norm": 0.5360826849937439,
|
| 354 |
+
"learning_rate": 0.00019977751557348025,
|
| 355 |
+
"loss": 3.2973,
|
| 356 |
+
"step": 760
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 15.6,
|
| 360 |
+
"grad_norm": 0.43838226795196533,
|
| 361 |
+
"learning_rate": 0.00019974779544686683,
|
| 362 |
+
"loss": 3.2853,
|
| 363 |
+
"step": 780
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 16.0,
|
| 367 |
+
"grad_norm": 0.594366729259491,
|
| 368 |
+
"learning_rate": 0.0001997162152198747,
|
| 369 |
+
"loss": 3.2927,
|
| 370 |
+
"step": 800
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 16.4,
|
| 374 |
+
"grad_norm": 0.5382407307624817,
|
| 375 |
+
"learning_rate": 0.00019968277548141302,
|
| 376 |
+
"loss": 3.2757,
|
| 377 |
+
"step": 820
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 16.5,
|
| 381 |
+
"eval_loss": 3.3513808250427246,
|
| 382 |
+
"eval_runtime": 1.264,
|
| 383 |
+
"eval_samples_per_second": 474.674,
|
| 384 |
+
"eval_steps_per_second": 19.778,
|
| 385 |
+
"step": 825
|
| 386 |
+
},
|
| 387 |
+
{
|
| 388 |
+
"epoch": 16.8,
|
| 389 |
+
"grad_norm": 0.6264561414718628,
|
| 390 |
+
"learning_rate": 0.0001996474768550672,
|
| 391 |
+
"loss": 3.2876,
|
| 392 |
+
"step": 840
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 17.2,
|
| 396 |
+
"grad_norm": 0.5054410099983215,
|
| 397 |
+
"learning_rate": 0.00019961031999908725,
|
| 398 |
+
"loss": 3.2794,
|
| 399 |
+
"step": 860
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 17.6,
|
| 403 |
+
"grad_norm": 0.4844222366809845,
|
| 404 |
+
"learning_rate": 0.00019957130560637552,
|
| 405 |
+
"loss": 3.2632,
|
| 406 |
+
"step": 880
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 18.0,
|
| 410 |
+
"grad_norm": 0.7173214554786682,
|
| 411 |
+
"learning_rate": 0.0001995304344044738,
|
| 412 |
+
"loss": 3.2738,
|
| 413 |
+
"step": 900
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 18.0,
|
| 417 |
+
"eval_loss": 3.339338779449463,
|
| 418 |
+
"eval_runtime": 1.2575,
|
| 419 |
+
"eval_samples_per_second": 477.137,
|
| 420 |
+
"eval_steps_per_second": 19.881,
|
| 421 |
+
"step": 900
|
| 422 |
+
}
|
| 423 |
+
],
|
| 424 |
+
"logging_steps": 20,
|
| 425 |
+
"max_steps": 15000,
|
| 426 |
+
"num_input_tokens_seen": 0,
|
| 427 |
+
"num_train_epochs": 300,
|
| 428 |
+
"save_steps": 150,
|
| 429 |
+
"stateful_callbacks": {
|
| 430 |
+
"TrainerControl": {
|
| 431 |
+
"args": {
|
| 432 |
+
"should_epoch_stop": false,
|
| 433 |
+
"should_evaluate": false,
|
| 434 |
+
"should_log": false,
|
| 435 |
+
"should_save": true,
|
| 436 |
+
"should_training_stop": false
|
| 437 |
+
},
|
| 438 |
+
"attributes": {}
|
| 439 |
+
}
|
| 440 |
+
},
|
| 441 |
+
"total_flos": 1702112210196480.0,
|
| 442 |
+
"train_batch_size": 96,
|
| 443 |
+
"trial_name": null,
|
| 444 |
+
"trial_params": null
|
| 445 |
+
}
|