Training in progress, step 50, checkpoint
Browse files
last-checkpoint/adapter_config.json
CHANGED
|
@@ -10,7 +10,7 @@
|
|
| 10 |
"layers_pattern": null,
|
| 11 |
"layers_to_transform": null,
|
| 12 |
"loftq_config": {},
|
| 13 |
-
"lora_alpha":
|
| 14 |
"lora_dropout": 0,
|
| 15 |
"megatron_config": null,
|
| 16 |
"megatron_core": "megatron.core",
|
|
@@ -20,13 +20,13 @@
|
|
| 20 |
"rank_pattern": {},
|
| 21 |
"revision": "unsloth",
|
| 22 |
"target_modules": [
|
| 23 |
-
"q_proj",
|
| 24 |
-
"k_proj",
|
| 25 |
"up_proj",
|
| 26 |
-
"
|
| 27 |
-
"v_proj",
|
| 28 |
"down_proj",
|
| 29 |
-
"
|
|
|
|
|
|
|
|
|
|
| 30 |
],
|
| 31 |
"task_type": "CAUSAL_LM",
|
| 32 |
"use_dora": false,
|
|
|
|
| 10 |
"layers_pattern": null,
|
| 11 |
"layers_to_transform": null,
|
| 12 |
"loftq_config": {},
|
| 13 |
+
"lora_alpha": 8,
|
| 14 |
"lora_dropout": 0,
|
| 15 |
"megatron_config": null,
|
| 16 |
"megatron_core": "megatron.core",
|
|
|
|
| 20 |
"rank_pattern": {},
|
| 21 |
"revision": "unsloth",
|
| 22 |
"target_modules": [
|
|
|
|
|
|
|
| 23 |
"up_proj",
|
| 24 |
+
"o_proj",
|
|
|
|
| 25 |
"down_proj",
|
| 26 |
+
"q_proj",
|
| 27 |
+
"v_proj",
|
| 28 |
+
"gate_proj",
|
| 29 |
+
"k_proj"
|
| 30 |
],
|
| 31 |
"task_type": "CAUSAL_LM",
|
| 32 |
"use_dora": false,
|
last-checkpoint/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1912664024
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d918b445e9c66a0ce0e4e84d9061fa2984ef4baa2b691afe223994e68094059d
|
| 3 |
size 1912664024
|
last-checkpoint/optimizer.pt
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 sha256:420ae61f406a1716b041f637974c538814787f0c36b6cb165d38598f7461778f
|
| 3 |
+
size 958697364
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -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 sha256:376fa8cdca401e75879a433c4cf57e5c31b5973d7635f7e690ffeb05b513101b
|
| 3 |
size 14244
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -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 sha256:43eaf6e8b9b7e97d802563efbae1976d32c9bf9869f2ff306873ea7672065324
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,537 +1,42 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 20,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"epoch": 0.0012270691453463403,
|
| 13 |
-
"grad_norm": 0.
|
| 14 |
"learning_rate": 0.00019981588314717073,
|
| 15 |
-
"loss": 2.
|
| 16 |
"step": 20
|
| 17 |
},
|
| 18 |
{
|
| 19 |
"epoch": 0.0012270691453463403,
|
| 20 |
-
"eval_loss": 2.
|
| 21 |
-
"eval_runtime": 23.
|
| 22 |
-
"eval_samples_per_second": 4.
|
| 23 |
-
"eval_steps_per_second": 0.
|
| 24 |
"step": 20
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.0024541382906926807,
|
| 28 |
-
"grad_norm": 0.
|
| 29 |
"learning_rate": 0.00019957039401006504,
|
| 30 |
-
"loss": 2.
|
| 31 |
"step": 40
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.0024541382906926807,
|
| 35 |
-
"eval_loss":
|
| 36 |
-
"eval_runtime": 23.
|
| 37 |
-
"eval_samples_per_second": 4.242,
|
| 38 |
-
"eval_steps_per_second": 0.552,
|
| 39 |
-
"step": 40
|
| 40 |
-
},
|
| 41 |
-
{
|
| 42 |
-
"epoch": 0.003681207436039021,
|
| 43 |
-
"grad_norm": 0.08049603551626205,
|
| 44 |
-
"learning_rate": 0.00019932490487295938,
|
| 45 |
-
"loss": 2.0529,
|
| 46 |
-
"step": 60
|
| 47 |
-
},
|
| 48 |
-
{
|
| 49 |
-
"epoch": 0.003681207436039021,
|
| 50 |
-
"eval_loss": 1.9338455200195312,
|
| 51 |
-
"eval_runtime": 23.1495,
|
| 52 |
-
"eval_samples_per_second": 4.32,
|
| 53 |
-
"eval_steps_per_second": 0.562,
|
| 54 |
-
"step": 60
|
| 55 |
-
},
|
| 56 |
-
{
|
| 57 |
-
"epoch": 0.004908276581385361,
|
| 58 |
-
"grad_norm": 0.08653070032596588,
|
| 59 |
-
"learning_rate": 0.00019907941573585368,
|
| 60 |
-
"loss": 1.9395,
|
| 61 |
-
"step": 80
|
| 62 |
-
},
|
| 63 |
-
{
|
| 64 |
-
"epoch": 0.004908276581385361,
|
| 65 |
-
"eval_loss": 1.8689138889312744,
|
| 66 |
-
"eval_runtime": 23.7275,
|
| 67 |
-
"eval_samples_per_second": 4.215,
|
| 68 |
-
"eval_steps_per_second": 0.548,
|
| 69 |
-
"step": 80
|
| 70 |
-
},
|
| 71 |
-
{
|
| 72 |
-
"epoch": 0.006135345726731701,
|
| 73 |
-
"grad_norm": 0.08481493592262268,
|
| 74 |
-
"learning_rate": 0.00019883392659874802,
|
| 75 |
-
"loss": 1.8773,
|
| 76 |
-
"step": 100
|
| 77 |
-
},
|
| 78 |
-
{
|
| 79 |
-
"epoch": 0.006135345726731701,
|
| 80 |
-
"eval_loss": 1.8180441856384277,
|
| 81 |
-
"eval_runtime": 23.5705,
|
| 82 |
-
"eval_samples_per_second": 4.243,
|
| 83 |
-
"eval_steps_per_second": 0.552,
|
| 84 |
-
"step": 100
|
| 85 |
-
},
|
| 86 |
-
{
|
| 87 |
-
"epoch": 0.007362414872078042,
|
| 88 |
-
"grad_norm": 0.11568216979503632,
|
| 89 |
-
"learning_rate": 0.00019858843746164233,
|
| 90 |
-
"loss": 1.7827,
|
| 91 |
-
"step": 120
|
| 92 |
-
},
|
| 93 |
-
{
|
| 94 |
-
"epoch": 0.007362414872078042,
|
| 95 |
-
"eval_loss": 1.774067997932434,
|
| 96 |
-
"eval_runtime": 23.9722,
|
| 97 |
-
"eval_samples_per_second": 4.172,
|
| 98 |
-
"eval_steps_per_second": 0.542,
|
| 99 |
-
"step": 120
|
| 100 |
-
},
|
| 101 |
-
{
|
| 102 |
-
"epoch": 0.008589484017424381,
|
| 103 |
-
"grad_norm": 0.10869361460208893,
|
| 104 |
-
"learning_rate": 0.00019834294832453666,
|
| 105 |
-
"loss": 1.812,
|
| 106 |
-
"step": 140
|
| 107 |
-
},
|
| 108 |
-
{
|
| 109 |
-
"epoch": 0.008589484017424381,
|
| 110 |
-
"eval_loss": 1.737804889678955,
|
| 111 |
-
"eval_runtime": 23.519,
|
| 112 |
-
"eval_samples_per_second": 4.252,
|
| 113 |
-
"eval_steps_per_second": 0.553,
|
| 114 |
-
"step": 140
|
| 115 |
-
},
|
| 116 |
-
{
|
| 117 |
-
"epoch": 0.009816553162770723,
|
| 118 |
-
"grad_norm": 0.0976206362247467,
|
| 119 |
-
"learning_rate": 0.00019809745918743097,
|
| 120 |
-
"loss": 1.74,
|
| 121 |
-
"step": 160
|
| 122 |
-
},
|
| 123 |
-
{
|
| 124 |
-
"epoch": 0.009816553162770723,
|
| 125 |
-
"eval_loss": 1.700899600982666,
|
| 126 |
-
"eval_runtime": 23.2347,
|
| 127 |
-
"eval_samples_per_second": 4.304,
|
| 128 |
-
"eval_steps_per_second": 0.56,
|
| 129 |
-
"step": 160
|
| 130 |
-
},
|
| 131 |
-
{
|
| 132 |
-
"epoch": 0.011043622308117063,
|
| 133 |
-
"grad_norm": 0.1123971939086914,
|
| 134 |
-
"learning_rate": 0.00019785197005032528,
|
| 135 |
-
"loss": 1.787,
|
| 136 |
-
"step": 180
|
| 137 |
-
},
|
| 138 |
-
{
|
| 139 |
-
"epoch": 0.011043622308117063,
|
| 140 |
-
"eval_loss": 1.6765294075012207,
|
| 141 |
-
"eval_runtime": 23.6403,
|
| 142 |
-
"eval_samples_per_second": 4.23,
|
| 143 |
-
"eval_steps_per_second": 0.55,
|
| 144 |
-
"step": 180
|
| 145 |
-
},
|
| 146 |
-
{
|
| 147 |
-
"epoch": 0.012270691453463402,
|
| 148 |
-
"grad_norm": 0.10320968925952911,
|
| 149 |
-
"learning_rate": 0.0001976064809132196,
|
| 150 |
-
"loss": 1.7804,
|
| 151 |
-
"step": 200
|
| 152 |
-
},
|
| 153 |
-
{
|
| 154 |
-
"epoch": 0.012270691453463402,
|
| 155 |
-
"eval_loss": 1.6563650369644165,
|
| 156 |
-
"eval_runtime": 23.6381,
|
| 157 |
-
"eval_samples_per_second": 4.23,
|
| 158 |
-
"eval_steps_per_second": 0.55,
|
| 159 |
-
"step": 200
|
| 160 |
-
},
|
| 161 |
-
{
|
| 162 |
-
"epoch": 0.013497760598809742,
|
| 163 |
-
"grad_norm": 0.14491896331310272,
|
| 164 |
-
"learning_rate": 0.00019736099177611392,
|
| 165 |
-
"loss": 1.7043,
|
| 166 |
-
"step": 220
|
| 167 |
-
},
|
| 168 |
-
{
|
| 169 |
-
"epoch": 0.013497760598809742,
|
| 170 |
-
"eval_loss": 1.6346065998077393,
|
| 171 |
-
"eval_runtime": 23.7121,
|
| 172 |
"eval_samples_per_second": 4.217,
|
| 173 |
"eval_steps_per_second": 0.548,
|
| 174 |
-
"step":
|
| 175 |
-
},
|
| 176 |
-
{
|
| 177 |
-
"epoch": 0.014724829744156084,
|
| 178 |
-
"grad_norm": 0.12502990663051605,
|
| 179 |
-
"learning_rate": 0.00019711550263900825,
|
| 180 |
-
"loss": 1.7345,
|
| 181 |
-
"step": 240
|
| 182 |
-
},
|
| 183 |
-
{
|
| 184 |
-
"epoch": 0.014724829744156084,
|
| 185 |
-
"eval_loss": 1.6147732734680176,
|
| 186 |
-
"eval_runtime": 23.5936,
|
| 187 |
-
"eval_samples_per_second": 4.238,
|
| 188 |
-
"eval_steps_per_second": 0.551,
|
| 189 |
-
"step": 240
|
| 190 |
-
},
|
| 191 |
-
{
|
| 192 |
-
"epoch": 0.015951898889502422,
|
| 193 |
-
"grad_norm": 0.1230228915810585,
|
| 194 |
-
"learning_rate": 0.00019687001350190256,
|
| 195 |
-
"loss": 1.7338,
|
| 196 |
-
"step": 260
|
| 197 |
-
},
|
| 198 |
-
{
|
| 199 |
-
"epoch": 0.015951898889502422,
|
| 200 |
-
"eval_loss": 1.5957908630371094,
|
| 201 |
-
"eval_runtime": 23.389,
|
| 202 |
-
"eval_samples_per_second": 4.276,
|
| 203 |
-
"eval_steps_per_second": 0.556,
|
| 204 |
-
"step": 260
|
| 205 |
-
},
|
| 206 |
-
{
|
| 207 |
-
"epoch": 0.017178968034848762,
|
| 208 |
-
"grad_norm": 0.12000931799411774,
|
| 209 |
-
"learning_rate": 0.00019662452436479687,
|
| 210 |
-
"loss": 1.7143,
|
| 211 |
-
"step": 280
|
| 212 |
-
},
|
| 213 |
-
{
|
| 214 |
-
"epoch": 0.017178968034848762,
|
| 215 |
-
"eval_loss": 1.585697889328003,
|
| 216 |
-
"eval_runtime": 23.566,
|
| 217 |
-
"eval_samples_per_second": 4.243,
|
| 218 |
-
"eval_steps_per_second": 0.552,
|
| 219 |
-
"step": 280
|
| 220 |
-
},
|
| 221 |
-
{
|
| 222 |
-
"epoch": 0.018406037180195105,
|
| 223 |
-
"grad_norm": 0.1442350149154663,
|
| 224 |
-
"learning_rate": 0.00019637903522769118,
|
| 225 |
-
"loss": 1.6406,
|
| 226 |
-
"step": 300
|
| 227 |
-
},
|
| 228 |
-
{
|
| 229 |
-
"epoch": 0.018406037180195105,
|
| 230 |
-
"eval_loss": 1.5710804462432861,
|
| 231 |
-
"eval_runtime": 23.5083,
|
| 232 |
-
"eval_samples_per_second": 4.254,
|
| 233 |
-
"eval_steps_per_second": 0.553,
|
| 234 |
-
"step": 300
|
| 235 |
-
},
|
| 236 |
-
{
|
| 237 |
-
"epoch": 0.019633106325541445,
|
| 238 |
-
"grad_norm": 0.09555982798337936,
|
| 239 |
-
"learning_rate": 0.00019613354609058549,
|
| 240 |
-
"loss": 1.6213,
|
| 241 |
-
"step": 320
|
| 242 |
-
},
|
| 243 |
-
{
|
| 244 |
-
"epoch": 0.019633106325541445,
|
| 245 |
-
"eval_loss": 1.5556869506835938,
|
| 246 |
-
"eval_runtime": 23.5239,
|
| 247 |
-
"eval_samples_per_second": 4.251,
|
| 248 |
-
"eval_steps_per_second": 0.553,
|
| 249 |
-
"step": 320
|
| 250 |
-
},
|
| 251 |
-
{
|
| 252 |
-
"epoch": 0.020860175470887785,
|
| 253 |
-
"grad_norm": 0.13320715725421906,
|
| 254 |
-
"learning_rate": 0.00019588805695347982,
|
| 255 |
-
"loss": 1.6956,
|
| 256 |
-
"step": 340
|
| 257 |
-
},
|
| 258 |
-
{
|
| 259 |
-
"epoch": 0.020860175470887785,
|
| 260 |
-
"eval_loss": 1.5424914360046387,
|
| 261 |
-
"eval_runtime": 23.6064,
|
| 262 |
-
"eval_samples_per_second": 4.236,
|
| 263 |
-
"eval_steps_per_second": 0.551,
|
| 264 |
-
"step": 340
|
| 265 |
-
},
|
| 266 |
-
{
|
| 267 |
-
"epoch": 0.022087244616234125,
|
| 268 |
-
"grad_norm": 0.12061001360416412,
|
| 269 |
-
"learning_rate": 0.00019564256781637413,
|
| 270 |
-
"loss": 1.6589,
|
| 271 |
-
"step": 360
|
| 272 |
-
},
|
| 273 |
-
{
|
| 274 |
-
"epoch": 0.022087244616234125,
|
| 275 |
-
"eval_loss": 1.528477430343628,
|
| 276 |
-
"eval_runtime": 23.6796,
|
| 277 |
-
"eval_samples_per_second": 4.223,
|
| 278 |
-
"eval_steps_per_second": 0.549,
|
| 279 |
-
"step": 360
|
| 280 |
-
},
|
| 281 |
-
{
|
| 282 |
-
"epoch": 0.023314313761580465,
|
| 283 |
-
"grad_norm": 0.14327766001224518,
|
| 284 |
-
"learning_rate": 0.00019539707867926844,
|
| 285 |
-
"loss": 1.5946,
|
| 286 |
-
"step": 380
|
| 287 |
-
},
|
| 288 |
-
{
|
| 289 |
-
"epoch": 0.023314313761580465,
|
| 290 |
-
"eval_loss": 1.52202570438385,
|
| 291 |
-
"eval_runtime": 23.6756,
|
| 292 |
-
"eval_samples_per_second": 4.224,
|
| 293 |
-
"eval_steps_per_second": 0.549,
|
| 294 |
-
"step": 380
|
| 295 |
-
},
|
| 296 |
-
{
|
| 297 |
-
"epoch": 0.024541382906926805,
|
| 298 |
-
"grad_norm": 0.12291988730430603,
|
| 299 |
-
"learning_rate": 0.00019515158954216277,
|
| 300 |
-
"loss": 1.5366,
|
| 301 |
-
"step": 400
|
| 302 |
-
},
|
| 303 |
-
{
|
| 304 |
-
"epoch": 0.024541382906926805,
|
| 305 |
-
"eval_loss": 1.507960319519043,
|
| 306 |
-
"eval_runtime": 23.6216,
|
| 307 |
-
"eval_samples_per_second": 4.233,
|
| 308 |
-
"eval_steps_per_second": 0.55,
|
| 309 |
-
"step": 400
|
| 310 |
-
},
|
| 311 |
-
{
|
| 312 |
-
"epoch": 0.025768452052273145,
|
| 313 |
-
"grad_norm": 0.15288175642490387,
|
| 314 |
-
"learning_rate": 0.00019490610040505708,
|
| 315 |
-
"loss": 1.5829,
|
| 316 |
-
"step": 420
|
| 317 |
-
},
|
| 318 |
-
{
|
| 319 |
-
"epoch": 0.025768452052273145,
|
| 320 |
-
"eval_loss": 1.4994325637817383,
|
| 321 |
-
"eval_runtime": 23.6368,
|
| 322 |
-
"eval_samples_per_second": 4.231,
|
| 323 |
-
"eval_steps_per_second": 0.55,
|
| 324 |
-
"step": 420
|
| 325 |
-
},
|
| 326 |
-
{
|
| 327 |
-
"epoch": 0.026995521197619485,
|
| 328 |
-
"grad_norm": 0.13319191336631775,
|
| 329 |
-
"learning_rate": 0.0001946606112679514,
|
| 330 |
-
"loss": 1.5523,
|
| 331 |
-
"step": 440
|
| 332 |
-
},
|
| 333 |
-
{
|
| 334 |
-
"epoch": 0.026995521197619485,
|
| 335 |
-
"eval_loss": 1.4956778287887573,
|
| 336 |
-
"eval_runtime": 23.6921,
|
| 337 |
-
"eval_samples_per_second": 4.221,
|
| 338 |
-
"eval_steps_per_second": 0.549,
|
| 339 |
-
"step": 440
|
| 340 |
-
},
|
| 341 |
-
{
|
| 342 |
-
"epoch": 0.028222590342965825,
|
| 343 |
-
"grad_norm": 0.14759239554405212,
|
| 344 |
-
"learning_rate": 0.00019441512213084572,
|
| 345 |
-
"loss": 1.5735,
|
| 346 |
-
"step": 460
|
| 347 |
-
},
|
| 348 |
-
{
|
| 349 |
-
"epoch": 0.028222590342965825,
|
| 350 |
-
"eval_loss": 1.486402988433838,
|
| 351 |
-
"eval_runtime": 23.2911,
|
| 352 |
-
"eval_samples_per_second": 4.293,
|
| 353 |
-
"eval_steps_per_second": 0.558,
|
| 354 |
-
"step": 460
|
| 355 |
-
},
|
| 356 |
-
{
|
| 357 |
-
"epoch": 0.029449659488312168,
|
| 358 |
-
"grad_norm": 0.11428073793649673,
|
| 359 |
-
"learning_rate": 0.00019416963299374006,
|
| 360 |
-
"loss": 1.5788,
|
| 361 |
-
"step": 480
|
| 362 |
-
},
|
| 363 |
-
{
|
| 364 |
-
"epoch": 0.029449659488312168,
|
| 365 |
-
"eval_loss": 1.4712104797363281,
|
| 366 |
-
"eval_runtime": 23.4851,
|
| 367 |
-
"eval_samples_per_second": 4.258,
|
| 368 |
-
"eval_steps_per_second": 0.554,
|
| 369 |
-
"step": 480
|
| 370 |
-
},
|
| 371 |
-
{
|
| 372 |
-
"epoch": 0.030676728633658508,
|
| 373 |
-
"grad_norm": 0.11649870127439499,
|
| 374 |
-
"learning_rate": 0.00019392414385663436,
|
| 375 |
-
"loss": 1.5667,
|
| 376 |
-
"step": 500
|
| 377 |
-
},
|
| 378 |
-
{
|
| 379 |
-
"epoch": 0.030676728633658508,
|
| 380 |
-
"eval_loss": 1.4620152711868286,
|
| 381 |
-
"eval_runtime": 23.5455,
|
| 382 |
-
"eval_samples_per_second": 4.247,
|
| 383 |
-
"eval_steps_per_second": 0.552,
|
| 384 |
-
"step": 500
|
| 385 |
-
},
|
| 386 |
-
{
|
| 387 |
-
"epoch": 0.031903797779004844,
|
| 388 |
-
"grad_norm": 0.16019868850708008,
|
| 389 |
-
"learning_rate": 0.00019367865471952867,
|
| 390 |
-
"loss": 1.4778,
|
| 391 |
-
"step": 520
|
| 392 |
-
},
|
| 393 |
-
{
|
| 394 |
-
"epoch": 0.031903797779004844,
|
| 395 |
-
"eval_loss": 1.4597880840301514,
|
| 396 |
-
"eval_runtime": 23.624,
|
| 397 |
-
"eval_samples_per_second": 4.233,
|
| 398 |
-
"eval_steps_per_second": 0.55,
|
| 399 |
-
"step": 520
|
| 400 |
-
},
|
| 401 |
-
{
|
| 402 |
-
"epoch": 0.03313086692435119,
|
| 403 |
-
"grad_norm": 0.1370091289281845,
|
| 404 |
-
"learning_rate": 0.00019343316558242298,
|
| 405 |
-
"loss": 1.5531,
|
| 406 |
-
"step": 540
|
| 407 |
-
},
|
| 408 |
-
{
|
| 409 |
-
"epoch": 0.03313086692435119,
|
| 410 |
-
"eval_loss": 1.443243384361267,
|
| 411 |
-
"eval_runtime": 23.5537,
|
| 412 |
-
"eval_samples_per_second": 4.246,
|
| 413 |
-
"eval_steps_per_second": 0.552,
|
| 414 |
-
"step": 540
|
| 415 |
-
},
|
| 416 |
-
{
|
| 417 |
-
"epoch": 0.034357936069697524,
|
| 418 |
-
"grad_norm": 0.1211417093873024,
|
| 419 |
-
"learning_rate": 0.0001931876764453173,
|
| 420 |
-
"loss": 1.5879,
|
| 421 |
-
"step": 560
|
| 422 |
-
},
|
| 423 |
-
{
|
| 424 |
-
"epoch": 0.034357936069697524,
|
| 425 |
-
"eval_loss": 1.4466437101364136,
|
| 426 |
-
"eval_runtime": 23.8508,
|
| 427 |
-
"eval_samples_per_second": 4.193,
|
| 428 |
-
"eval_steps_per_second": 0.545,
|
| 429 |
-
"step": 560
|
| 430 |
-
},
|
| 431 |
-
{
|
| 432 |
-
"epoch": 0.03558500521504387,
|
| 433 |
-
"grad_norm": 0.14397528767585754,
|
| 434 |
-
"learning_rate": 0.00019294218730821162,
|
| 435 |
-
"loss": 1.5352,
|
| 436 |
-
"step": 580
|
| 437 |
-
},
|
| 438 |
-
{
|
| 439 |
-
"epoch": 0.03558500521504387,
|
| 440 |
-
"eval_loss": 1.4339115619659424,
|
| 441 |
-
"eval_runtime": 23.649,
|
| 442 |
-
"eval_samples_per_second": 4.229,
|
| 443 |
-
"eval_steps_per_second": 0.55,
|
| 444 |
-
"step": 580
|
| 445 |
-
},
|
| 446 |
-
{
|
| 447 |
-
"epoch": 0.03681207436039021,
|
| 448 |
-
"grad_norm": 0.12468410283327103,
|
| 449 |
-
"learning_rate": 0.00019269669817110593,
|
| 450 |
-
"loss": 1.5045,
|
| 451 |
-
"step": 600
|
| 452 |
-
},
|
| 453 |
-
{
|
| 454 |
-
"epoch": 0.03681207436039021,
|
| 455 |
-
"eval_loss": 1.4277862310409546,
|
| 456 |
-
"eval_runtime": 23.647,
|
| 457 |
-
"eval_samples_per_second": 4.229,
|
| 458 |
-
"eval_steps_per_second": 0.55,
|
| 459 |
-
"step": 600
|
| 460 |
-
},
|
| 461 |
-
{
|
| 462 |
-
"epoch": 0.03803914350573655,
|
| 463 |
-
"grad_norm": 0.1577584445476532,
|
| 464 |
-
"learning_rate": 0.00019245120903400024,
|
| 465 |
-
"loss": 1.5497,
|
| 466 |
-
"step": 620
|
| 467 |
-
},
|
| 468 |
-
{
|
| 469 |
-
"epoch": 0.03803914350573655,
|
| 470 |
-
"eval_loss": 1.4203659296035767,
|
| 471 |
-
"eval_runtime": 23.8622,
|
| 472 |
-
"eval_samples_per_second": 4.191,
|
| 473 |
-
"eval_steps_per_second": 0.545,
|
| 474 |
-
"step": 620
|
| 475 |
-
},
|
| 476 |
-
{
|
| 477 |
-
"epoch": 0.03926621265108289,
|
| 478 |
-
"grad_norm": 0.12410438805818558,
|
| 479 |
-
"learning_rate": 0.00019220571989689457,
|
| 480 |
-
"loss": 1.503,
|
| 481 |
-
"step": 640
|
| 482 |
-
},
|
| 483 |
-
{
|
| 484 |
-
"epoch": 0.03926621265108289,
|
| 485 |
-
"eval_loss": 1.4154139757156372,
|
| 486 |
-
"eval_runtime": 23.4706,
|
| 487 |
-
"eval_samples_per_second": 4.261,
|
| 488 |
-
"eval_steps_per_second": 0.554,
|
| 489 |
-
"step": 640
|
| 490 |
-
},
|
| 491 |
-
{
|
| 492 |
-
"epoch": 0.04049328179642923,
|
| 493 |
-
"grad_norm": 0.13563913106918335,
|
| 494 |
-
"learning_rate": 0.00019196023075978888,
|
| 495 |
-
"loss": 1.4851,
|
| 496 |
-
"step": 660
|
| 497 |
-
},
|
| 498 |
-
{
|
| 499 |
-
"epoch": 0.04049328179642923,
|
| 500 |
-
"eval_loss": 1.414802074432373,
|
| 501 |
-
"eval_runtime": 23.3961,
|
| 502 |
-
"eval_samples_per_second": 4.274,
|
| 503 |
-
"eval_steps_per_second": 0.556,
|
| 504 |
-
"step": 660
|
| 505 |
-
},
|
| 506 |
-
{
|
| 507 |
-
"epoch": 0.04172035094177557,
|
| 508 |
-
"grad_norm": 0.13915061950683594,
|
| 509 |
-
"learning_rate": 0.00019171474162268321,
|
| 510 |
-
"loss": 1.4847,
|
| 511 |
-
"step": 680
|
| 512 |
-
},
|
| 513 |
-
{
|
| 514 |
-
"epoch": 0.04172035094177557,
|
| 515 |
-
"eval_loss": 1.4029760360717773,
|
| 516 |
-
"eval_runtime": 23.6066,
|
| 517 |
-
"eval_samples_per_second": 4.236,
|
| 518 |
-
"eval_steps_per_second": 0.551,
|
| 519 |
-
"step": 680
|
| 520 |
-
},
|
| 521 |
-
{
|
| 522 |
-
"epoch": 0.04294742008712191,
|
| 523 |
-
"grad_norm": 0.14418162405490875,
|
| 524 |
-
"learning_rate": 0.00019146925248557752,
|
| 525 |
-
"loss": 1.4724,
|
| 526 |
-
"step": 700
|
| 527 |
-
},
|
| 528 |
-
{
|
| 529 |
-
"epoch": 0.04294742008712191,
|
| 530 |
-
"eval_loss": 1.4029196500778198,
|
| 531 |
-
"eval_runtime": 23.5244,
|
| 532 |
-
"eval_samples_per_second": 4.251,
|
| 533 |
-
"eval_steps_per_second": 0.553,
|
| 534 |
-
"step": 700
|
| 535 |
}
|
| 536 |
],
|
| 537 |
"logging_steps": 20,
|
|
@@ -539,7 +44,7 @@
|
|
| 539 |
"num_input_tokens_seen": 0,
|
| 540 |
"num_train_epochs": 1,
|
| 541 |
"save_steps": 50,
|
| 542 |
-
"total_flos":
|
| 543 |
"train_batch_size": 2,
|
| 544 |
"trial_name": null,
|
| 545 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.0030676728633658506,
|
| 5 |
"eval_steps": 20,
|
| 6 |
+
"global_step": 50,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"epoch": 0.0012270691453463403,
|
| 13 |
+
"grad_norm": 0.07455573976039886,
|
| 14 |
"learning_rate": 0.00019981588314717073,
|
| 15 |
+
"loss": 2.6233,
|
| 16 |
"step": 20
|
| 17 |
},
|
| 18 |
{
|
| 19 |
"epoch": 0.0012270691453463403,
|
| 20 |
+
"eval_loss": 2.2018821239471436,
|
| 21 |
+
"eval_runtime": 23.583,
|
| 22 |
+
"eval_samples_per_second": 4.24,
|
| 23 |
+
"eval_steps_per_second": 0.551,
|
| 24 |
"step": 20
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.0024541382906926807,
|
| 28 |
+
"grad_norm": 0.0842185765504837,
|
| 29 |
"learning_rate": 0.00019957039401006504,
|
| 30 |
+
"loss": 2.1109,
|
| 31 |
"step": 40
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.0024541382906926807,
|
| 35 |
+
"eval_loss": 1.9889661073684692,
|
| 36 |
+
"eval_runtime": 23.7122,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 37 |
"eval_samples_per_second": 4.217,
|
| 38 |
"eval_steps_per_second": 0.548,
|
| 39 |
+
"step": 40
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 40 |
}
|
| 41 |
],
|
| 42 |
"logging_steps": 20,
|
|
|
|
| 44 |
"num_input_tokens_seen": 0,
|
| 45 |
"num_train_epochs": 1,
|
| 46 |
"save_steps": 50,
|
| 47 |
+
"total_flos": 3276518679244800.0,
|
| 48 |
"train_batch_size": 2,
|
| 49 |
"trial_name": null,
|
| 50 |
"trial_params": null
|