Training in progress, step 24000
Browse files- adapter_model.safetensors +1 -1
- last-checkpoint/adapter_config.json +4 -4
- last-checkpoint/adapter_model.safetensors +1 -1
- last-checkpoint/optimizer.pt +1 -1
- last-checkpoint/rng_state.pth +1 -1
- last-checkpoint/scheduler.pt +1 -1
- last-checkpoint/trainer_state.json +3 -815
- last-checkpoint/training_args.bin +1 -1
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 360740440
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:687727c0c025af3bad6b25b52789d99245d33b1d1735aa8becfc6bfe71318f16
|
| 3 |
size 360740440
|
last-checkpoint/adapter_config.json
CHANGED
|
@@ -23,13 +23,13 @@
|
|
| 23 |
"rank_pattern": {},
|
| 24 |
"revision": null,
|
| 25 |
"target_modules": [
|
| 26 |
-
"k_proj",
|
| 27 |
-
"o_proj",
|
| 28 |
-
"v_proj",
|
| 29 |
"gate_proj",
|
| 30 |
"q_proj",
|
|
|
|
| 31 |
"up_proj",
|
| 32 |
-
"
|
|
|
|
|
|
|
| 33 |
],
|
| 34 |
"task_type": "CAUSAL_LM",
|
| 35 |
"use_dora": false,
|
|
|
|
| 23 |
"rank_pattern": {},
|
| 24 |
"revision": null,
|
| 25 |
"target_modules": [
|
|
|
|
|
|
|
|
|
|
| 26 |
"gate_proj",
|
| 27 |
"q_proj",
|
| 28 |
+
"k_proj",
|
| 29 |
"up_proj",
|
| 30 |
+
"v_proj",
|
| 31 |
+
"down_proj",
|
| 32 |
+
"o_proj"
|
| 33 |
],
|
| 34 |
"task_type": "CAUSAL_LM",
|
| 35 |
"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 360740440
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2a1dd12a210017a07f561dace36dc00f59b55ff12d579593e8e5f59db1ca495d
|
| 3 |
size 360740440
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 184018770
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:09dbea6aaf941a049c9aa3656b362f19b6c30102fcdcd2d81680e63cd278a9c4
|
| 3 |
size 184018770
|
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:fe9c2e6cee455bb212d5ae0dd7c343acca65b7f37490224e2c429dd428b2c9ad
|
| 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 sha256:06eec3ac373cc6b81236635fbbf70132276a159a47d5685288ad3485f23d8131
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -28,818 +28,6 @@
|
|
| 28 |
"learning_rate": 1.9999998496226195e-05,
|
| 29 |
"loss": 2.202,
|
| 30 |
"step": 600
|
| 31 |
-
},
|
| 32 |
-
{
|
| 33 |
-
"epoch": 0.0005563858842119246,
|
| 34 |
-
"grad_norm": 7.003796577453613,
|
| 35 |
-
"learning_rate": 1.999999732662442e-05,
|
| 36 |
-
"loss": 2.0739,
|
| 37 |
-
"step": 800
|
| 38 |
-
},
|
| 39 |
-
{
|
| 40 |
-
"epoch": 0.0006954823552649058,
|
| 41 |
-
"grad_norm": 4.851119041442871,
|
| 42 |
-
"learning_rate": 1.99999958228508e-05,
|
| 43 |
-
"loss": 2.0175,
|
| 44 |
-
"step": 1000
|
| 45 |
-
},
|
| 46 |
-
{
|
| 47 |
-
"epoch": 0.0008345788263178869,
|
| 48 |
-
"grad_norm": 6.82249116897583,
|
| 49 |
-
"learning_rate": 1.999999398490542e-05,
|
| 50 |
-
"loss": 2.0206,
|
| 51 |
-
"step": 1200
|
| 52 |
-
},
|
| 53 |
-
{
|
| 54 |
-
"epoch": 0.0009736752973708681,
|
| 55 |
-
"grad_norm": 3.424342632293701,
|
| 56 |
-
"learning_rate": 1.999999181278835e-05,
|
| 57 |
-
"loss": 2.0587,
|
| 58 |
-
"step": 1400
|
| 59 |
-
},
|
| 60 |
-
{
|
| 61 |
-
"epoch": 0.001112771768423849,
|
| 62 |
-
"grad_norm": 5.291999340057373,
|
| 63 |
-
"learning_rate": 1.999998930649971e-05,
|
| 64 |
-
"loss": 2.0822,
|
| 65 |
-
"step": 1600
|
| 66 |
-
},
|
| 67 |
-
{
|
| 68 |
-
"epoch": 0.0012518682394768305,
|
| 69 |
-
"grad_norm": 5.283050537109375,
|
| 70 |
-
"learning_rate": 1.9999986466039608e-05,
|
| 71 |
-
"loss": 2.0833,
|
| 72 |
-
"step": 1800
|
| 73 |
-
},
|
| 74 |
-
{
|
| 75 |
-
"epoch": 0.0013909647105298116,
|
| 76 |
-
"grad_norm": 5.604334354400635,
|
| 77 |
-
"learning_rate": 1.999998329140819e-05,
|
| 78 |
-
"loss": 2.1159,
|
| 79 |
-
"step": 2000
|
| 80 |
-
},
|
| 81 |
-
{
|
| 82 |
-
"epoch": 0.0015300611815827927,
|
| 83 |
-
"grad_norm": 4.539398670196533,
|
| 84 |
-
"learning_rate": 1.99999797826056e-05,
|
| 85 |
-
"loss": 2.0815,
|
| 86 |
-
"step": 2200
|
| 87 |
-
},
|
| 88 |
-
{
|
| 89 |
-
"epoch": 0.0016691576526357739,
|
| 90 |
-
"grad_norm": 5.886259078979492,
|
| 91 |
-
"learning_rate": 1.9999975939632005e-05,
|
| 92 |
-
"loss": 2.0135,
|
| 93 |
-
"step": 2400
|
| 94 |
-
},
|
| 95 |
-
{
|
| 96 |
-
"epoch": 0.001808254123688755,
|
| 97 |
-
"grad_norm": 5.076303482055664,
|
| 98 |
-
"learning_rate": 1.9999971762487593e-05,
|
| 99 |
-
"loss": 2.0704,
|
| 100 |
-
"step": 2600
|
| 101 |
-
},
|
| 102 |
-
{
|
| 103 |
-
"epoch": 0.0019473505947417361,
|
| 104 |
-
"grad_norm": 5.898621559143066,
|
| 105 |
-
"learning_rate": 1.999996725117256e-05,
|
| 106 |
-
"loss": 2.0405,
|
| 107 |
-
"step": 2800
|
| 108 |
-
},
|
| 109 |
-
{
|
| 110 |
-
"epoch": 0.0020864470657947173,
|
| 111 |
-
"grad_norm": 4.124399662017822,
|
| 112 |
-
"learning_rate": 1.9999962405687122e-05,
|
| 113 |
-
"loss": 1.9888,
|
| 114 |
-
"step": 3000
|
| 115 |
-
},
|
| 116 |
-
{
|
| 117 |
-
"epoch": 0.002225543536847698,
|
| 118 |
-
"grad_norm": 5.221329689025879,
|
| 119 |
-
"learning_rate": 1.9999957226031515e-05,
|
| 120 |
-
"loss": 2.0433,
|
| 121 |
-
"step": 3200
|
| 122 |
-
},
|
| 123 |
-
{
|
| 124 |
-
"epoch": 0.0023646400079006796,
|
| 125 |
-
"grad_norm": 5.066499710083008,
|
| 126 |
-
"learning_rate": 1.9999951712205977e-05,
|
| 127 |
-
"loss": 1.9347,
|
| 128 |
-
"step": 3400
|
| 129 |
-
},
|
| 130 |
-
{
|
| 131 |
-
"epoch": 0.002503736478953661,
|
| 132 |
-
"grad_norm": 5.852139949798584,
|
| 133 |
-
"learning_rate": 1.9999945864210777e-05,
|
| 134 |
-
"loss": 2.0016,
|
| 135 |
-
"step": 3600
|
| 136 |
-
},
|
| 137 |
-
{
|
| 138 |
-
"epoch": 0.002642832950006642,
|
| 139 |
-
"grad_norm": 3.3446638584136963,
|
| 140 |
-
"learning_rate": 1.999993968204619e-05,
|
| 141 |
-
"loss": 2.0887,
|
| 142 |
-
"step": 3800
|
| 143 |
-
},
|
| 144 |
-
{
|
| 145 |
-
"epoch": 0.002781929421059623,
|
| 146 |
-
"grad_norm": 4.233782768249512,
|
| 147 |
-
"learning_rate": 1.9999933165712525e-05,
|
| 148 |
-
"loss": 2.0563,
|
| 149 |
-
"step": 4000
|
| 150 |
-
},
|
| 151 |
-
{
|
| 152 |
-
"epoch": 0.002921025892112604,
|
| 153 |
-
"grad_norm": 5.62469482421875,
|
| 154 |
-
"learning_rate": 1.999992631521008e-05,
|
| 155 |
-
"loss": 1.9745,
|
| 156 |
-
"step": 4200
|
| 157 |
-
},
|
| 158 |
-
{
|
| 159 |
-
"epoch": 0.0030601223631655855,
|
| 160 |
-
"grad_norm": 5.743213176727295,
|
| 161 |
-
"learning_rate": 1.999991913053918e-05,
|
| 162 |
-
"loss": 1.9969,
|
| 163 |
-
"step": 4400
|
| 164 |
-
},
|
| 165 |
-
{
|
| 166 |
-
"epoch": 0.0031992188342185664,
|
| 167 |
-
"grad_norm": 6.02440071105957,
|
| 168 |
-
"learning_rate": 1.9999911611700172e-05,
|
| 169 |
-
"loss": 1.9583,
|
| 170 |
-
"step": 4600
|
| 171 |
-
},
|
| 172 |
-
{
|
| 173 |
-
"epoch": 0.0033383153052715477,
|
| 174 |
-
"grad_norm": 6.839730739593506,
|
| 175 |
-
"learning_rate": 1.999990375869342e-05,
|
| 176 |
-
"loss": 2.0131,
|
| 177 |
-
"step": 4800
|
| 178 |
-
},
|
| 179 |
-
{
|
| 180 |
-
"epoch": 0.0034774117763245287,
|
| 181 |
-
"grad_norm": 6.269118309020996,
|
| 182 |
-
"learning_rate": 1.9999895571519294e-05,
|
| 183 |
-
"loss": 2.0302,
|
| 184 |
-
"step": 5000
|
| 185 |
-
},
|
| 186 |
-
{
|
| 187 |
-
"epoch": 0.00361650824737751,
|
| 188 |
-
"grad_norm": 5.001865863800049,
|
| 189 |
-
"learning_rate": 1.9999887050178187e-05,
|
| 190 |
-
"loss": 1.946,
|
| 191 |
-
"step": 5200
|
| 192 |
-
},
|
| 193 |
-
{
|
| 194 |
-
"epoch": 0.003755604718430491,
|
| 195 |
-
"grad_norm": 4.900404930114746,
|
| 196 |
-
"learning_rate": 1.99998781946705e-05,
|
| 197 |
-
"loss": 1.9367,
|
| 198 |
-
"step": 5400
|
| 199 |
-
},
|
| 200 |
-
{
|
| 201 |
-
"epoch": 0.0038947011894834723,
|
| 202 |
-
"grad_norm": 15.09891128540039,
|
| 203 |
-
"learning_rate": 1.9999869004996664e-05,
|
| 204 |
-
"loss": 2.0014,
|
| 205 |
-
"step": 5600
|
| 206 |
-
},
|
| 207 |
-
{
|
| 208 |
-
"epoch": 0.004033797660536453,
|
| 209 |
-
"grad_norm": 5.8560285568237305,
|
| 210 |
-
"learning_rate": 1.9999859481157112e-05,
|
| 211 |
-
"loss": 1.9926,
|
| 212 |
-
"step": 5800
|
| 213 |
-
},
|
| 214 |
-
{
|
| 215 |
-
"epoch": 0.004172894131589435,
|
| 216 |
-
"grad_norm": 4.16021203994751,
|
| 217 |
-
"learning_rate": 1.99998496231523e-05,
|
| 218 |
-
"loss": 1.9833,
|
| 219 |
-
"step": 6000
|
| 220 |
-
},
|
| 221 |
-
{
|
| 222 |
-
"epoch": 0.004311990602642416,
|
| 223 |
-
"grad_norm": 5.219597339630127,
|
| 224 |
-
"learning_rate": 1.9999839430982698e-05,
|
| 225 |
-
"loss": 2.0193,
|
| 226 |
-
"step": 6200
|
| 227 |
-
},
|
| 228 |
-
{
|
| 229 |
-
"epoch": 0.004451087073695396,
|
| 230 |
-
"grad_norm": 5.076187610626221,
|
| 231 |
-
"learning_rate": 1.9999828904648794e-05,
|
| 232 |
-
"loss": 1.9929,
|
| 233 |
-
"step": 6400
|
| 234 |
-
},
|
| 235 |
-
{
|
| 236 |
-
"epoch": 0.004590183544748378,
|
| 237 |
-
"grad_norm": 4.857876777648926,
|
| 238 |
-
"learning_rate": 1.9999818044151088e-05,
|
| 239 |
-
"loss": 2.0109,
|
| 240 |
-
"step": 6600
|
| 241 |
-
},
|
| 242 |
-
{
|
| 243 |
-
"epoch": 0.004729280015801359,
|
| 244 |
-
"grad_norm": 3.6183769702911377,
|
| 245 |
-
"learning_rate": 1.9999806849490103e-05,
|
| 246 |
-
"loss": 2.019,
|
| 247 |
-
"step": 6800
|
| 248 |
-
},
|
| 249 |
-
{
|
| 250 |
-
"epoch": 0.0048683764868543405,
|
| 251 |
-
"grad_norm": 3.081662178039551,
|
| 252 |
-
"learning_rate": 1.9999795320666373e-05,
|
| 253 |
-
"loss": 2.0102,
|
| 254 |
-
"step": 7000
|
| 255 |
-
},
|
| 256 |
-
{
|
| 257 |
-
"epoch": 0.005007472957907322,
|
| 258 |
-
"grad_norm": 3.6525964736938477,
|
| 259 |
-
"learning_rate": 1.9999783457680448e-05,
|
| 260 |
-
"loss": 2.0319,
|
| 261 |
-
"step": 7200
|
| 262 |
-
},
|
| 263 |
-
{
|
| 264 |
-
"epoch": 0.005146569428960302,
|
| 265 |
-
"grad_norm": 5.330043315887451,
|
| 266 |
-
"learning_rate": 1.9999771260532886e-05,
|
| 267 |
-
"loss": 1.9925,
|
| 268 |
-
"step": 7400
|
| 269 |
-
},
|
| 270 |
-
{
|
| 271 |
-
"epoch": 0.005285665900013284,
|
| 272 |
-
"grad_norm": 4.449774742126465,
|
| 273 |
-
"learning_rate": 1.9999758729224277e-05,
|
| 274 |
-
"loss": 2.033,
|
| 275 |
-
"step": 7600
|
| 276 |
-
},
|
| 277 |
-
{
|
| 278 |
-
"epoch": 0.005424762371066265,
|
| 279 |
-
"grad_norm": 4.431877136230469,
|
| 280 |
-
"learning_rate": 1.9999745863755225e-05,
|
| 281 |
-
"loss": 2.0076,
|
| 282 |
-
"step": 7800
|
| 283 |
-
},
|
| 284 |
-
{
|
| 285 |
-
"epoch": 0.005563858842119246,
|
| 286 |
-
"grad_norm": 4.620669841766357,
|
| 287 |
-
"learning_rate": 1.999973266412633e-05,
|
| 288 |
-
"loss": 1.9021,
|
| 289 |
-
"step": 8000
|
| 290 |
-
},
|
| 291 |
-
{
|
| 292 |
-
"epoch": 0.005702955313172227,
|
| 293 |
-
"grad_norm": 6.169932842254639,
|
| 294 |
-
"learning_rate": 1.9999719130338235e-05,
|
| 295 |
-
"loss": 1.9601,
|
| 296 |
-
"step": 8200
|
| 297 |
-
},
|
| 298 |
-
{
|
| 299 |
-
"epoch": 0.005842051784225208,
|
| 300 |
-
"grad_norm": 5.546196460723877,
|
| 301 |
-
"learning_rate": 1.9999705262391574e-05,
|
| 302 |
-
"loss": 2.0071,
|
| 303 |
-
"step": 8400
|
| 304 |
-
},
|
| 305 |
-
{
|
| 306 |
-
"epoch": 0.00598114825527819,
|
| 307 |
-
"grad_norm": 6.319796562194824,
|
| 308 |
-
"learning_rate": 1.999969106028702e-05,
|
| 309 |
-
"loss": 1.9881,
|
| 310 |
-
"step": 8600
|
| 311 |
-
},
|
| 312 |
-
{
|
| 313 |
-
"epoch": 0.006120244726331171,
|
| 314 |
-
"grad_norm": 5.991957187652588,
|
| 315 |
-
"learning_rate": 1.999967652402525e-05,
|
| 316 |
-
"loss": 1.9871,
|
| 317 |
-
"step": 8800
|
| 318 |
-
},
|
| 319 |
-
{
|
| 320 |
-
"epoch": 0.006259341197384151,
|
| 321 |
-
"grad_norm": 8.088932037353516,
|
| 322 |
-
"learning_rate": 1.9999661653606947e-05,
|
| 323 |
-
"loss": 2.0261,
|
| 324 |
-
"step": 9000
|
| 325 |
-
},
|
| 326 |
-
{
|
| 327 |
-
"epoch": 0.006398437668437133,
|
| 328 |
-
"grad_norm": 4.159659385681152,
|
| 329 |
-
"learning_rate": 1.9999646449032833e-05,
|
| 330 |
-
"loss": 2.0084,
|
| 331 |
-
"step": 9200
|
| 332 |
-
},
|
| 333 |
-
{
|
| 334 |
-
"epoch": 0.006537534139490114,
|
| 335 |
-
"grad_norm": 4.520460605621338,
|
| 336 |
-
"learning_rate": 1.9999630910303628e-05,
|
| 337 |
-
"loss": 1.9969,
|
| 338 |
-
"step": 9400
|
| 339 |
-
},
|
| 340 |
-
{
|
| 341 |
-
"epoch": 0.0066766306105430955,
|
| 342 |
-
"grad_norm": 5.970917701721191,
|
| 343 |
-
"learning_rate": 1.9999615037420078e-05,
|
| 344 |
-
"loss": 2.0213,
|
| 345 |
-
"step": 9600
|
| 346 |
-
},
|
| 347 |
-
{
|
| 348 |
-
"epoch": 0.006815727081596076,
|
| 349 |
-
"grad_norm": 4.681369304656982,
|
| 350 |
-
"learning_rate": 1.9999598830382934e-05,
|
| 351 |
-
"loss": 1.9711,
|
| 352 |
-
"step": 9800
|
| 353 |
-
},
|
| 354 |
-
{
|
| 355 |
-
"epoch": 0.006954823552649057,
|
| 356 |
-
"grad_norm": 4.877756118774414,
|
| 357 |
-
"learning_rate": 1.9999582289192974e-05,
|
| 358 |
-
"loss": 1.9427,
|
| 359 |
-
"step": 10000
|
| 360 |
-
},
|
| 361 |
-
{
|
| 362 |
-
"epoch": 0.007093920023702039,
|
| 363 |
-
"grad_norm": 6.686002254486084,
|
| 364 |
-
"learning_rate": 1.9999565413850993e-05,
|
| 365 |
-
"loss": 2.0468,
|
| 366 |
-
"step": 10200
|
| 367 |
-
},
|
| 368 |
-
{
|
| 369 |
-
"epoch": 0.00723301649475502,
|
| 370 |
-
"grad_norm": 4.4461236000061035,
|
| 371 |
-
"learning_rate": 1.9999548204357783e-05,
|
| 372 |
-
"loss": 1.9352,
|
| 373 |
-
"step": 10400
|
| 374 |
-
},
|
| 375 |
-
{
|
| 376 |
-
"epoch": 0.007372112965808001,
|
| 377 |
-
"grad_norm": 9.74795150756836,
|
| 378 |
-
"learning_rate": 1.9999530660714175e-05,
|
| 379 |
-
"loss": 1.9799,
|
| 380 |
-
"step": 10600
|
| 381 |
-
},
|
| 382 |
-
{
|
| 383 |
-
"epoch": 0.007511209436860982,
|
| 384 |
-
"grad_norm": 7.161931037902832,
|
| 385 |
-
"learning_rate": 1.999951278292101e-05,
|
| 386 |
-
"loss": 2.012,
|
| 387 |
-
"step": 10800
|
| 388 |
-
},
|
| 389 |
-
{
|
| 390 |
-
"epoch": 0.007650305907913963,
|
| 391 |
-
"grad_norm": 3.948525905609131,
|
| 392 |
-
"learning_rate": 1.999949457097913e-05,
|
| 393 |
-
"loss": 1.9932,
|
| 394 |
-
"step": 11000
|
| 395 |
-
},
|
| 396 |
-
{
|
| 397 |
-
"epoch": 0.007789402378966945,
|
| 398 |
-
"grad_norm": 4.876201629638672,
|
| 399 |
-
"learning_rate": 1.9999476024889414e-05,
|
| 400 |
-
"loss": 1.9907,
|
| 401 |
-
"step": 11200
|
| 402 |
-
},
|
| 403 |
-
{
|
| 404 |
-
"epoch": 0.007928498850019925,
|
| 405 |
-
"grad_norm": 6.009729862213135,
|
| 406 |
-
"learning_rate": 1.9999457144652746e-05,
|
| 407 |
-
"loss": 1.9347,
|
| 408 |
-
"step": 11400
|
| 409 |
-
},
|
| 410 |
-
{
|
| 411 |
-
"epoch": 0.008067595321072906,
|
| 412 |
-
"grad_norm": 5.02329158782959,
|
| 413 |
-
"learning_rate": 1.999943793027002e-05,
|
| 414 |
-
"loss": 1.9714,
|
| 415 |
-
"step": 11600
|
| 416 |
-
},
|
| 417 |
-
{
|
| 418 |
-
"epoch": 0.008206691792125888,
|
| 419 |
-
"grad_norm": 5.213016986846924,
|
| 420 |
-
"learning_rate": 1.999941838174216e-05,
|
| 421 |
-
"loss": 2.0812,
|
| 422 |
-
"step": 11800
|
| 423 |
-
},
|
| 424 |
-
{
|
| 425 |
-
"epoch": 0.00834578826317887,
|
| 426 |
-
"grad_norm": 6.0479416847229,
|
| 427 |
-
"learning_rate": 1.9999398499070103e-05,
|
| 428 |
-
"loss": 1.9836,
|
| 429 |
-
"step": 12000
|
| 430 |
-
},
|
| 431 |
-
{
|
| 432 |
-
"epoch": 0.00848488473423185,
|
| 433 |
-
"grad_norm": 5.763875484466553,
|
| 434 |
-
"learning_rate": 1.9999378282254787e-05,
|
| 435 |
-
"loss": 2.0187,
|
| 436 |
-
"step": 12200
|
| 437 |
-
},
|
| 438 |
-
{
|
| 439 |
-
"epoch": 0.008623981205284832,
|
| 440 |
-
"grad_norm": 5.157899379730225,
|
| 441 |
-
"learning_rate": 1.999935773129719e-05,
|
| 442 |
-
"loss": 1.9386,
|
| 443 |
-
"step": 12400
|
| 444 |
-
},
|
| 445 |
-
{
|
| 446 |
-
"epoch": 0.008763077676337813,
|
| 447 |
-
"grad_norm": 2.9542014598846436,
|
| 448 |
-
"learning_rate": 1.999933684619828e-05,
|
| 449 |
-
"loss": 2.0046,
|
| 450 |
-
"step": 12600
|
| 451 |
-
},
|
| 452 |
-
{
|
| 453 |
-
"epoch": 0.008902174147390793,
|
| 454 |
-
"grad_norm": 5.380853176116943,
|
| 455 |
-
"learning_rate": 1.9999315626959067e-05,
|
| 456 |
-
"loss": 2.0102,
|
| 457 |
-
"step": 12800
|
| 458 |
-
},
|
| 459 |
-
{
|
| 460 |
-
"epoch": 0.009041270618443774,
|
| 461 |
-
"grad_norm": 5.02028226852417,
|
| 462 |
-
"learning_rate": 1.999929407358055e-05,
|
| 463 |
-
"loss": 1.9841,
|
| 464 |
-
"step": 13000
|
| 465 |
-
},
|
| 466 |
-
{
|
| 467 |
-
"epoch": 0.009180367089496756,
|
| 468 |
-
"grad_norm": 6.345952987670898,
|
| 469 |
-
"learning_rate": 1.9999272186063767e-05,
|
| 470 |
-
"loss": 1.9903,
|
| 471 |
-
"step": 13200
|
| 472 |
-
},
|
| 473 |
-
{
|
| 474 |
-
"epoch": 0.009319463560549737,
|
| 475 |
-
"grad_norm": 5.001527309417725,
|
| 476 |
-
"learning_rate": 1.9999249964409763e-05,
|
| 477 |
-
"loss": 1.9106,
|
| 478 |
-
"step": 13400
|
| 479 |
-
},
|
| 480 |
-
{
|
| 481 |
-
"epoch": 0.009458560031602718,
|
| 482 |
-
"grad_norm": 5.8275651931762695,
|
| 483 |
-
"learning_rate": 1.9999227408619597e-05,
|
| 484 |
-
"loss": 1.946,
|
| 485 |
-
"step": 13600
|
| 486 |
-
},
|
| 487 |
-
{
|
| 488 |
-
"epoch": 0.0095976565026557,
|
| 489 |
-
"grad_norm": 4.7476396560668945,
|
| 490 |
-
"learning_rate": 1.9999204518694348e-05,
|
| 491 |
-
"loss": 1.99,
|
| 492 |
-
"step": 13800
|
| 493 |
-
},
|
| 494 |
-
{
|
| 495 |
-
"epoch": 0.009736752973708681,
|
| 496 |
-
"grad_norm": 4.389928817749023,
|
| 497 |
-
"learning_rate": 1.9999181294635103e-05,
|
| 498 |
-
"loss": 1.9211,
|
| 499 |
-
"step": 14000
|
| 500 |
-
},
|
| 501 |
-
{
|
| 502 |
-
"epoch": 0.009875849444761662,
|
| 503 |
-
"grad_norm": 3.244771957397461,
|
| 504 |
-
"learning_rate": 1.9999157736442973e-05,
|
| 505 |
-
"loss": 1.9621,
|
| 506 |
-
"step": 14200
|
| 507 |
-
},
|
| 508 |
-
{
|
| 509 |
-
"epoch": 0.010014945915814644,
|
| 510 |
-
"grad_norm": 5.625466823577881,
|
| 511 |
-
"learning_rate": 1.999913384411909e-05,
|
| 512 |
-
"loss": 1.9118,
|
| 513 |
-
"step": 14400
|
| 514 |
-
},
|
| 515 |
-
{
|
| 516 |
-
"epoch": 0.010154042386867623,
|
| 517 |
-
"grad_norm": 4.992525577545166,
|
| 518 |
-
"learning_rate": 1.9999109617664585e-05,
|
| 519 |
-
"loss": 1.9967,
|
| 520 |
-
"step": 14600
|
| 521 |
-
},
|
| 522 |
-
{
|
| 523 |
-
"epoch": 0.010293138857920605,
|
| 524 |
-
"grad_norm": 4.458539962768555,
|
| 525 |
-
"learning_rate": 1.9999085057080614e-05,
|
| 526 |
-
"loss": 1.9181,
|
| 527 |
-
"step": 14800
|
| 528 |
-
},
|
| 529 |
-
{
|
| 530 |
-
"epoch": 0.010432235328973586,
|
| 531 |
-
"grad_norm": 4.348485946655273,
|
| 532 |
-
"learning_rate": 1.999906016236836e-05,
|
| 533 |
-
"loss": 1.954,
|
| 534 |
-
"step": 15000
|
| 535 |
-
},
|
| 536 |
-
{
|
| 537 |
-
"epoch": 0.010571331800026567,
|
| 538 |
-
"grad_norm": 6.683957099914551,
|
| 539 |
-
"learning_rate": 1.9999034933529003e-05,
|
| 540 |
-
"loss": 1.957,
|
| 541 |
-
"step": 15200
|
| 542 |
-
},
|
| 543 |
-
{
|
| 544 |
-
"epoch": 0.010710428271079549,
|
| 545 |
-
"grad_norm": 4.755821704864502,
|
| 546 |
-
"learning_rate": 1.9999009370563746e-05,
|
| 547 |
-
"loss": 1.9285,
|
| 548 |
-
"step": 15400
|
| 549 |
-
},
|
| 550 |
-
{
|
| 551 |
-
"epoch": 0.01084952474213253,
|
| 552 |
-
"grad_norm": 4.806936264038086,
|
| 553 |
-
"learning_rate": 1.9998983473473816e-05,
|
| 554 |
-
"loss": 1.9129,
|
| 555 |
-
"step": 15600
|
| 556 |
-
},
|
| 557 |
-
{
|
| 558 |
-
"epoch": 0.010988621213185511,
|
| 559 |
-
"grad_norm": 2.8975718021392822,
|
| 560 |
-
"learning_rate": 1.9998957242260447e-05,
|
| 561 |
-
"loss": 1.9598,
|
| 562 |
-
"step": 15800
|
| 563 |
-
},
|
| 564 |
-
{
|
| 565 |
-
"epoch": 0.011127717684238493,
|
| 566 |
-
"grad_norm": 4.4350409507751465,
|
| 567 |
-
"learning_rate": 1.9998930676924887e-05,
|
| 568 |
-
"loss": 1.9818,
|
| 569 |
-
"step": 16000
|
| 570 |
-
},
|
| 571 |
-
{
|
| 572 |
-
"epoch": 0.011266814155291472,
|
| 573 |
-
"grad_norm": 4.06223726272583,
|
| 574 |
-
"learning_rate": 1.999890377746841e-05,
|
| 575 |
-
"loss": 2.0254,
|
| 576 |
-
"step": 16200
|
| 577 |
-
},
|
| 578 |
-
{
|
| 579 |
-
"epoch": 0.011405910626344454,
|
| 580 |
-
"grad_norm": 4.433406829833984,
|
| 581 |
-
"learning_rate": 1.9998876543892297e-05,
|
| 582 |
-
"loss": 1.9962,
|
| 583 |
-
"step": 16400
|
| 584 |
-
},
|
| 585 |
-
{
|
| 586 |
-
"epoch": 0.011545007097397435,
|
| 587 |
-
"grad_norm": 4.087177753448486,
|
| 588 |
-
"learning_rate": 1.9998848976197845e-05,
|
| 589 |
-
"loss": 1.9585,
|
| 590 |
-
"step": 16600
|
| 591 |
-
},
|
| 592 |
-
{
|
| 593 |
-
"epoch": 0.011684103568450416,
|
| 594 |
-
"grad_norm": 5.861230850219727,
|
| 595 |
-
"learning_rate": 1.9998821074386376e-05,
|
| 596 |
-
"loss": 1.9837,
|
| 597 |
-
"step": 16800
|
| 598 |
-
},
|
| 599 |
-
{
|
| 600 |
-
"epoch": 0.011823200039503398,
|
| 601 |
-
"grad_norm": 7.665340423583984,
|
| 602 |
-
"learning_rate": 1.9998792838459226e-05,
|
| 603 |
-
"loss": 2.0083,
|
| 604 |
-
"step": 17000
|
| 605 |
-
},
|
| 606 |
-
{
|
| 607 |
-
"epoch": 0.01196229651055638,
|
| 608 |
-
"grad_norm": 3.315056800842285,
|
| 609 |
-
"learning_rate": 1.9998764268417728e-05,
|
| 610 |
-
"loss": 2.0438,
|
| 611 |
-
"step": 17200
|
| 612 |
-
},
|
| 613 |
-
{
|
| 614 |
-
"epoch": 0.01210139298160936,
|
| 615 |
-
"grad_norm": 4.6072258949279785,
|
| 616 |
-
"learning_rate": 1.9998735364263257e-05,
|
| 617 |
-
"loss": 1.9738,
|
| 618 |
-
"step": 17400
|
| 619 |
-
},
|
| 620 |
-
{
|
| 621 |
-
"epoch": 0.012240489452662342,
|
| 622 |
-
"grad_norm": 4.772326946258545,
|
| 623 |
-
"learning_rate": 1.9998706125997193e-05,
|
| 624 |
-
"loss": 1.9731,
|
| 625 |
-
"step": 17600
|
| 626 |
-
},
|
| 627 |
-
{
|
| 628 |
-
"epoch": 0.012379585923715323,
|
| 629 |
-
"grad_norm": 6.146634578704834,
|
| 630 |
-
"learning_rate": 1.9998676553620927e-05,
|
| 631 |
-
"loss": 1.9572,
|
| 632 |
-
"step": 17800
|
| 633 |
-
},
|
| 634 |
-
{
|
| 635 |
-
"epoch": 0.012518682394768303,
|
| 636 |
-
"grad_norm": 5.802914619445801,
|
| 637 |
-
"learning_rate": 1.9998646647135877e-05,
|
| 638 |
-
"loss": 2.0147,
|
| 639 |
-
"step": 18000
|
| 640 |
-
},
|
| 641 |
-
{
|
| 642 |
-
"epoch": 0.012657778865821284,
|
| 643 |
-
"grad_norm": 5.375220775604248,
|
| 644 |
-
"learning_rate": 1.9998616406543457e-05,
|
| 645 |
-
"loss": 1.9244,
|
| 646 |
-
"step": 18200
|
| 647 |
-
},
|
| 648 |
-
{
|
| 649 |
-
"epoch": 0.012796875336874266,
|
| 650 |
-
"grad_norm": 5.7547101974487305,
|
| 651 |
-
"learning_rate": 1.999858583184513e-05,
|
| 652 |
-
"loss": 2.0012,
|
| 653 |
-
"step": 18400
|
| 654 |
-
},
|
| 655 |
-
{
|
| 656 |
-
"epoch": 0.012935971807927247,
|
| 657 |
-
"grad_norm": 5.727509498596191,
|
| 658 |
-
"learning_rate": 1.9998554923042343e-05,
|
| 659 |
-
"loss": 2.0101,
|
| 660 |
-
"step": 18600
|
| 661 |
-
},
|
| 662 |
-
{
|
| 663 |
-
"epoch": 0.013075068278980228,
|
| 664 |
-
"grad_norm": 7.512138366699219,
|
| 665 |
-
"learning_rate": 1.9998523680136575e-05,
|
| 666 |
-
"loss": 1.9783,
|
| 667 |
-
"step": 18800
|
| 668 |
-
},
|
| 669 |
-
{
|
| 670 |
-
"epoch": 0.01321416475003321,
|
| 671 |
-
"grad_norm": 5.997193813323975,
|
| 672 |
-
"learning_rate": 1.9998492103129314e-05,
|
| 673 |
-
"loss": 1.9289,
|
| 674 |
-
"step": 19000
|
| 675 |
-
},
|
| 676 |
-
{
|
| 677 |
-
"epoch": 0.013353261221086191,
|
| 678 |
-
"grad_norm": 3.422043800354004,
|
| 679 |
-
"learning_rate": 1.9998460192022073e-05,
|
| 680 |
-
"loss": 2.039,
|
| 681 |
-
"step": 19200
|
| 682 |
-
},
|
| 683 |
-
{
|
| 684 |
-
"epoch": 0.013492357692139172,
|
| 685 |
-
"grad_norm": 4.149094581604004,
|
| 686 |
-
"learning_rate": 1.999842794681637e-05,
|
| 687 |
-
"loss": 1.9124,
|
| 688 |
-
"step": 19400
|
| 689 |
-
},
|
| 690 |
-
{
|
| 691 |
-
"epoch": 0.013631454163192152,
|
| 692 |
-
"grad_norm": 4.264293670654297,
|
| 693 |
-
"learning_rate": 1.9998395367513753e-05,
|
| 694 |
-
"loss": 2.0029,
|
| 695 |
-
"step": 19600
|
| 696 |
-
},
|
| 697 |
-
{
|
| 698 |
-
"epoch": 0.013770550634245133,
|
| 699 |
-
"grad_norm": 6.137360095977783,
|
| 700 |
-
"learning_rate": 1.9998362454115767e-05,
|
| 701 |
-
"loss": 1.9444,
|
| 702 |
-
"step": 19800
|
| 703 |
-
},
|
| 704 |
-
{
|
| 705 |
-
"epoch": 0.013909647105298115,
|
| 706 |
-
"grad_norm": 6.3221588134765625,
|
| 707 |
-
"learning_rate": 1.999832920662399e-05,
|
| 708 |
-
"loss": 2.0269,
|
| 709 |
-
"step": 20000
|
| 710 |
-
},
|
| 711 |
-
{
|
| 712 |
-
"epoch": 0.014048743576351096,
|
| 713 |
-
"grad_norm": 6.676060676574707,
|
| 714 |
-
"learning_rate": 1.9998295625040006e-05,
|
| 715 |
-
"loss": 1.9565,
|
| 716 |
-
"step": 20200
|
| 717 |
-
},
|
| 718 |
-
{
|
| 719 |
-
"epoch": 0.014187840047404077,
|
| 720 |
-
"grad_norm": 7.500442981719971,
|
| 721 |
-
"learning_rate": 1.9998261709365422e-05,
|
| 722 |
-
"loss": 1.9674,
|
| 723 |
-
"step": 20400
|
| 724 |
-
},
|
| 725 |
-
{
|
| 726 |
-
"epoch": 0.014326936518457059,
|
| 727 |
-
"grad_norm": 4.544761657714844,
|
| 728 |
-
"learning_rate": 1.9998227459601847e-05,
|
| 729 |
-
"loss": 2.0291,
|
| 730 |
-
"step": 20600
|
| 731 |
-
},
|
| 732 |
-
{
|
| 733 |
-
"epoch": 0.01446603298951004,
|
| 734 |
-
"grad_norm": 4.8733744621276855,
|
| 735 |
-
"learning_rate": 1.9998192875750928e-05,
|
| 736 |
-
"loss": 1.9198,
|
| 737 |
-
"step": 20800
|
| 738 |
-
},
|
| 739 |
-
{
|
| 740 |
-
"epoch": 0.014605129460563021,
|
| 741 |
-
"grad_norm": 4.430062294006348,
|
| 742 |
-
"learning_rate": 1.999815795781431e-05,
|
| 743 |
-
"loss": 1.9795,
|
| 744 |
-
"step": 21000
|
| 745 |
-
},
|
| 746 |
-
{
|
| 747 |
-
"epoch": 0.014744225931616003,
|
| 748 |
-
"grad_norm": 4.341994762420654,
|
| 749 |
-
"learning_rate": 1.9998122705793667e-05,
|
| 750 |
-
"loss": 1.9514,
|
| 751 |
-
"step": 21200
|
| 752 |
-
},
|
| 753 |
-
{
|
| 754 |
-
"epoch": 0.014883322402668982,
|
| 755 |
-
"grad_norm": 5.507206916809082,
|
| 756 |
-
"learning_rate": 1.9998087119690667e-05,
|
| 757 |
-
"loss": 2.0101,
|
| 758 |
-
"step": 21400
|
| 759 |
-
},
|
| 760 |
-
{
|
| 761 |
-
"epoch": 0.015022418873721964,
|
| 762 |
-
"grad_norm": 3.7597243785858154,
|
| 763 |
-
"learning_rate": 1.9998051199507023e-05,
|
| 764 |
-
"loss": 1.9755,
|
| 765 |
-
"step": 21600
|
| 766 |
-
},
|
| 767 |
-
{
|
| 768 |
-
"epoch": 0.015161515344774945,
|
| 769 |
-
"grad_norm": 7.106645107269287,
|
| 770 |
-
"learning_rate": 1.9998014945244445e-05,
|
| 771 |
-
"loss": 2.0197,
|
| 772 |
-
"step": 21800
|
| 773 |
-
},
|
| 774 |
-
{
|
| 775 |
-
"epoch": 0.015300611815827926,
|
| 776 |
-
"grad_norm": 7.3320231437683105,
|
| 777 |
-
"learning_rate": 1.9997978356904658e-05,
|
| 778 |
-
"loss": 1.9894,
|
| 779 |
-
"step": 22000
|
| 780 |
-
},
|
| 781 |
-
{
|
| 782 |
-
"epoch": 0.015439708286880908,
|
| 783 |
-
"grad_norm": 5.327560901641846,
|
| 784 |
-
"learning_rate": 1.999794143448942e-05,
|
| 785 |
-
"loss": 1.9868,
|
| 786 |
-
"step": 22200
|
| 787 |
-
},
|
| 788 |
-
{
|
| 789 |
-
"epoch": 0.01557880475793389,
|
| 790 |
-
"grad_norm": 6.348668098449707,
|
| 791 |
-
"learning_rate": 1.9997904178000485e-05,
|
| 792 |
-
"loss": 2.0534,
|
| 793 |
-
"step": 22400
|
| 794 |
-
},
|
| 795 |
-
{
|
| 796 |
-
"epoch": 0.01571790122898687,
|
| 797 |
-
"grad_norm": 5.529504299163818,
|
| 798 |
-
"learning_rate": 1.9997866587439633e-05,
|
| 799 |
-
"loss": 1.9518,
|
| 800 |
-
"step": 22600
|
| 801 |
-
},
|
| 802 |
-
{
|
| 803 |
-
"epoch": 0.01585699770003985,
|
| 804 |
-
"grad_norm": 5.461117267608643,
|
| 805 |
-
"learning_rate": 1.9997828662808665e-05,
|
| 806 |
-
"loss": 1.9358,
|
| 807 |
-
"step": 22800
|
| 808 |
-
},
|
| 809 |
-
{
|
| 810 |
-
"epoch": 0.015996094171092833,
|
| 811 |
-
"grad_norm": 4.327660083770752,
|
| 812 |
-
"learning_rate": 1.999779040410938e-05,
|
| 813 |
-
"loss": 1.9166,
|
| 814 |
-
"step": 23000
|
| 815 |
-
},
|
| 816 |
-
{
|
| 817 |
-
"epoch": 0.016135190642145813,
|
| 818 |
-
"grad_norm": 4.492312908172607,
|
| 819 |
-
"learning_rate": 1.9997751811343617e-05,
|
| 820 |
-
"loss": 2.0067,
|
| 821 |
-
"step": 23200
|
| 822 |
-
},
|
| 823 |
-
{
|
| 824 |
-
"epoch": 0.016274287113198796,
|
| 825 |
-
"grad_norm": 4.487472057342529,
|
| 826 |
-
"learning_rate": 1.9997712884513206e-05,
|
| 827 |
-
"loss": 1.9651,
|
| 828 |
-
"step": 23400
|
| 829 |
-
},
|
| 830 |
-
{
|
| 831 |
-
"epoch": 0.016413383584251776,
|
| 832 |
-
"grad_norm": 4.890496253967285,
|
| 833 |
-
"learning_rate": 1.9997673623620018e-05,
|
| 834 |
-
"loss": 1.952,
|
| 835 |
-
"step": 23600
|
| 836 |
-
},
|
| 837 |
-
{
|
| 838 |
-
"epoch": 0.016552480055304755,
|
| 839 |
-
"grad_norm": 3.2920827865600586,
|
| 840 |
-
"learning_rate": 1.9997634028665915e-05,
|
| 841 |
-
"loss": 2.0047,
|
| 842 |
-
"step": 23800
|
| 843 |
}
|
| 844 |
],
|
| 845 |
"logging_steps": 200,
|
|
@@ -859,7 +47,7 @@
|
|
| 859 |
"attributes": {}
|
| 860 |
}
|
| 861 |
},
|
| 862 |
-
"total_flos":
|
| 863 |
"train_batch_size": 1,
|
| 864 |
"trial_name": null,
|
| 865 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.00041728941315894347,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 600,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 28 |
"learning_rate": 1.9999998496226195e-05,
|
| 29 |
"loss": 2.202,
|
| 30 |
"step": 600
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 31 |
}
|
| 32 |
],
|
| 33 |
"logging_steps": 200,
|
|
|
|
| 47 |
"attributes": {}
|
| 48 |
}
|
| 49 |
},
|
| 50 |
+
"total_flos": 1138117833400320.0,
|
| 51 |
"train_batch_size": 1,
|
| 52 |
"trial_name": null,
|
| 53 |
"trial_params": null
|
last-checkpoint/training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 6840
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f73bef2970b56da564b1d8c87d27fe806335e746653d451535ecd6b817d641ba
|
| 3 |
size 6840
|