Upload folder using huggingface_hub (#3)

- Upload folder using huggingface_hub (4f400093aef44295e47459519fc4d9c41c29a909)

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1b7c71f5a48f92e6510bcb8484209f58b75e91833c9a1c22c3aac3c29d200327
+oid sha256:8beb03e0dedfb5b2da0d68ebcc49dbb502cea67ba9c0d70bed474bdde253aa1d
 size 39409357

optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3319acdf659478ae0888e0b42c35dd789fcd6f3cf90e0350a5faa31dda4e7fbb
+oid sha256:745db72555229d45c6a69054bcdee18c9e9f3193c81ae335546612e8aaa6d7c4
 size 78844421

rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:370c3a07f37a8aae6ea141b54ca992b21699546baf7407eb587b6056f787333b
+oid sha256:6162bb9db25c89c41e126a7a00a5d0695219447bff9b18d08731531620758440
 size 14575

scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4f71b8f28a925f4e713f8094eb164fc93a2eae54f8641cb0bc6cba1f8b907c73
+oid sha256:461382c4e71db35cef694681fd3a0c229adb91f6f9d1e458e9d3c9be9149f8c4
 size 627

trainer_state.json

@@ -1,9 +1,9 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 11.428571428571429,
+  "epoch": 5.714285714285714,
   "eval_steps": 500,
-  "global_step": 400,
+  "global_step": 200,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -1207,1213 +1207,13 @@
       "learning_rate": 0.002642857142857143,
       "loss": 1.678,
       "step": 200
-    },
-    {
-      "epoch": 5.74,
-      "learning_rate": 0.0026410714285714284,
-      "loss": 1.7613,
-      "step": 201
-    },
-    {
-      "epoch": 5.77,
-      "learning_rate": 0.0026392857142857142,
-      "loss": 1.7541,
-      "step": 202
-    },
-    {
-      "epoch": 5.8,
-      "learning_rate": 0.0026375,
-      "loss": 1.798,
-      "step": 203
-    },
-    {
-      "epoch": 5.83,
-      "learning_rate": 0.002635714285714286,
-      "loss": 1.821,
-      "step": 204
-    },
-    {
-      "epoch": 5.86,
-      "learning_rate": 0.0026339285714285714,
-      "loss": 1.8385,
-      "step": 205
-    },
-    {
-      "epoch": 5.89,
-      "learning_rate": 0.002632142857142857,
-      "loss": 1.8613,
-      "step": 206
-    },
-    {
-      "epoch": 5.91,
-      "learning_rate": 0.002630357142857143,
-      "loss": 1.902,
-      "step": 207
-    },
-    {
-      "epoch": 5.94,
-      "learning_rate": 0.0026285714285714285,
-      "loss": 2.0848,
-      "step": 208
-    },
-    {
-      "epoch": 5.97,
-      "learning_rate": 0.0026267857142857143,
-      "loss": 2.3277,
-      "step": 209
-    },
-    {
-      "epoch": 6.0,
-      "learning_rate": 0.002625,
-      "loss": 2.8535,
-      "step": 210
-    },
-    {
-      "epoch": 6.03,
-      "learning_rate": 0.0026232142857142856,
-      "loss": 6.2197,
-      "step": 211
-    },
-    {
-      "epoch": 6.06,
-      "learning_rate": 0.0026214285714285714,
-      "loss": 10.2288,
-      "step": 212
-    },
-    {
-      "epoch": 6.09,
-      "learning_rate": 0.0026196428571428573,
-      "loss": 12.5006,
-      "step": 213
-    },
-    {
-      "epoch": 6.11,
-      "learning_rate": 0.002617857142857143,
-      "loss": 10.5184,
-      "step": 214
-    },
-    {
-      "epoch": 6.14,
-      "learning_rate": 0.0026160714285714285,
-      "loss": 9.4834,
-      "step": 215
-    },
-    {
-      "epoch": 6.17,
-      "learning_rate": 0.0026142857142857144,
-      "loss": 16.0513,
-      "step": 216
-    },
-    {
-      "epoch": 6.2,
-      "learning_rate": 0.0026125000000000002,
-      "loss": 11.0576,
-      "step": 217
-    },
-    {
-      "epoch": 6.23,
-      "learning_rate": 0.0026107142857142857,
-      "loss": 15.3574,
-      "step": 218
-    },
-    {
-      "epoch": 6.26,
-      "learning_rate": 0.0026089285714285715,
-      "loss": 15.5239,
-      "step": 219
-    },
-    {
-      "epoch": 6.29,
-      "learning_rate": 0.0026071428571428574,
-      "loss": 15.3973,
-      "step": 220
-    },
-    {
-      "epoch": 6.31,
-      "learning_rate": 0.0026053571428571428,
-      "loss": 12.059,
-      "step": 221
-    },
-    {
-      "epoch": 6.34,
-      "learning_rate": 0.0026035714285714286,
-      "loss": 10.8352,
-      "step": 222
-    },
-    {
-      "epoch": 6.37,
-      "learning_rate": 0.0026017857142857145,
-      "loss": 10.1507,
-      "step": 223
-    },
-    {
-      "epoch": 6.4,
-      "learning_rate": 0.0026000000000000003,
-      "loss": 10.651,
-      "step": 224
-    },
-    {
-      "epoch": 6.43,
-      "learning_rate": 0.0025982142857142857,
-      "loss": 9.8363,
-      "step": 225
-    },
-    {
-      "epoch": 6.46,
-      "learning_rate": 0.0025964285714285716,
-      "loss": 9.3673,
-      "step": 226
-    },
-    {
-      "epoch": 6.49,
-      "learning_rate": 0.0025946428571428574,
-      "loss": 9.5433,
-      "step": 227
-    },
-    {
-      "epoch": 6.51,
-      "learning_rate": 0.002592857142857143,
-      "loss": 9.9206,
-      "step": 228
-    },
-    {
-      "epoch": 6.54,
-      "learning_rate": 0.0025910714285714287,
-      "loss": 9.5516,
-      "step": 229
-    },
-    {
-      "epoch": 6.57,
-      "learning_rate": 0.0025892857142857145,
-      "loss": 9.2165,
-      "step": 230
-    },
-    {
-      "epoch": 6.6,
-      "learning_rate": 0.0025875000000000004,
-      "loss": 9.0825,
-      "step": 231
-    },
-    {
-      "epoch": 6.63,
-      "learning_rate": 0.002585714285714286,
-      "loss": 8.7437,
-      "step": 232
-    },
-    {
-      "epoch": 6.66,
-      "learning_rate": 0.0025839285714285717,
-      "loss": 8.6366,
-      "step": 233
-    },
-    {
-      "epoch": 6.69,
-      "learning_rate": 0.0025821428571428575,
-      "loss": 9.7431,
-      "step": 234
-    },
-    {
-      "epoch": 6.71,
-      "learning_rate": 0.002580357142857143,
-      "loss": 8.1876,
-      "step": 235
-    },
-    {
-      "epoch": 6.74,
-      "learning_rate": 0.0025785714285714288,
-      "loss": 8.4559,
-      "step": 236
-    },
-    {
-      "epoch": 6.77,
-      "learning_rate": 0.002576785714285714,
-      "loss": 8.0092,
-      "step": 237
-    },
-    {
-      "epoch": 6.8,
-      "learning_rate": 0.002575,
-      "loss": 8.028,
-      "step": 238
-    },
-    {
-      "epoch": 6.83,
-      "learning_rate": 0.0025732142857142854,
-      "loss": 7.8379,
-      "step": 239
-    },
-    {
-      "epoch": 6.86,
-      "learning_rate": 0.0025714285714285713,
-      "loss": 7.8127,
-      "step": 240
-    },
-    {
-      "epoch": 6.89,
-      "learning_rate": 0.002569642857142857,
-      "loss": 7.8252,
-      "step": 241
-    },
-    {
-      "epoch": 6.91,
-      "learning_rate": 0.002567857142857143,
-      "loss": 7.7094,
-      "step": 242
-    },
-    {
-      "epoch": 6.94,
-      "learning_rate": 0.0025660714285714284,
-      "loss": 7.7962,
-      "step": 243
-    },
-    {
-      "epoch": 6.97,
-      "learning_rate": 0.0025642857142857143,
-      "loss": 7.4966,
-      "step": 244
-    },
-    {
-      "epoch": 7.0,
-      "learning_rate": 0.0025625,
-      "loss": 7.4851,
-      "step": 245
-    },
-    {
-      "epoch": 7.03,
-      "learning_rate": 0.0025607142857142855,
-      "loss": 7.5188,
-      "step": 246
-    },
-    {
-      "epoch": 7.06,
-      "learning_rate": 0.0025589285714285714,
-      "loss": 7.7866,
-      "step": 247
-    },
-    {
-      "epoch": 7.09,
-      "learning_rate": 0.0025571428571428572,
-      "loss": 7.5743,
-      "step": 248
-    },
-    {
-      "epoch": 7.11,
-      "learning_rate": 0.0025553571428571426,
-      "loss": 7.4608,
-      "step": 249
-    },
-    {
-      "epoch": 7.14,
-      "learning_rate": 0.0025535714285714285,
-      "loss": 7.4655,
-      "step": 250
-    },
-    {
-      "epoch": 7.17,
-      "learning_rate": 0.0025517857142857143,
-      "loss": 7.5474,
-      "step": 251
-    },
-    {
-      "epoch": 7.2,
-      "learning_rate": 0.00255,
-      "loss": 7.6983,
-      "step": 252
-    },
-    {
-      "epoch": 7.23,
-      "learning_rate": 0.0025482142857142856,
-      "loss": 7.4936,
-      "step": 253
-    },
-    {
-      "epoch": 7.26,
-      "learning_rate": 0.0025464285714285714,
-      "loss": 7.6966,
-      "step": 254
-    },
-    {
-      "epoch": 7.29,
-      "learning_rate": 0.0025446428571428573,
-      "loss": 7.4701,
-      "step": 255
-    },
-    {
-      "epoch": 7.31,
-      "learning_rate": 0.0025428571428571427,
-      "loss": 7.511,
-      "step": 256
-    },
-    {
-      "epoch": 7.34,
-      "learning_rate": 0.0025410714285714286,
-      "loss": 7.3709,
-      "step": 257
-    },
-    {
-      "epoch": 7.37,
-      "learning_rate": 0.0025392857142857144,
-      "loss": 7.4582,
-      "step": 258
-    },
-    {
-      "epoch": 7.4,
-      "learning_rate": 0.0025375,
-      "loss": 7.4263,
-      "step": 259
-    },
-    {
-      "epoch": 7.43,
-      "learning_rate": 0.0025357142857142857,
-      "loss": 7.3134,
-      "step": 260
-    },
-    {
-      "epoch": 7.46,
-      "learning_rate": 0.0025339285714285715,
-      "loss": 7.3849,
-      "step": 261
-    },
-    {
-      "epoch": 7.49,
-      "learning_rate": 0.0025321428571428574,
-      "loss": 7.292,
-      "step": 262
-    },
-    {
-      "epoch": 7.51,
-      "learning_rate": 0.002530357142857143,
-      "loss": 7.343,
-      "step": 263
-    },
-    {
-      "epoch": 7.54,
-      "learning_rate": 0.0025285714285714286,
-      "loss": 7.3166,
-      "step": 264
-    },
-    {
-      "epoch": 7.57,
-      "learning_rate": 0.0025267857142857145,
-      "loss": 7.2676,
-      "step": 265
-    },
-    {
-      "epoch": 7.6,
-      "learning_rate": 0.002525,
-      "loss": 7.2955,
-      "step": 266
-    },
-    {
-      "epoch": 7.63,
-      "learning_rate": 0.0025232142857142857,
-      "loss": 7.3386,
-      "step": 267
-    },
-    {
-      "epoch": 7.66,
-      "learning_rate": 0.0025214285714285716,
-      "loss": 7.2682,
-      "step": 268
-    },
-    {
-      "epoch": 7.69,
-      "learning_rate": 0.0025196428571428574,
-      "loss": 7.2359,
-      "step": 269
-    },
-    {
-      "epoch": 7.71,
-      "learning_rate": 0.002517857142857143,
-      "loss": 7.1849,
-      "step": 270
-    },
-    {
-      "epoch": 7.74,
-      "learning_rate": 0.0025160714285714287,
-      "loss": 7.2421,
-      "step": 271
-    },
-    {
-      "epoch": 7.77,
-      "learning_rate": 0.0025142857142857146,
-      "loss": 7.2341,
-      "step": 272
-    },
-    {
-      "epoch": 7.8,
-      "learning_rate": 0.0025125,
-      "loss": 7.2901,
-      "step": 273
-    },
-    {
-      "epoch": 7.83,
-      "learning_rate": 0.002510714285714286,
-      "loss": 7.1931,
-      "step": 274
-    },
-    {
-      "epoch": 7.86,
-      "learning_rate": 0.0025089285714285717,
-      "loss": 7.1907,
-      "step": 275
-    },
-    {
-      "epoch": 7.89,
-      "learning_rate": 0.002507142857142857,
-      "loss": 7.2369,
-      "step": 276
-    },
-    {
-      "epoch": 7.91,
-      "learning_rate": 0.002505357142857143,
-      "loss": 7.1764,
-      "step": 277
-    },
-    {
-      "epoch": 7.94,
-      "learning_rate": 0.002503571428571429,
-      "loss": 7.1928,
-      "step": 278
-    },
-    {
-      "epoch": 7.97,
-      "learning_rate": 0.0025017857142857146,
-      "loss": 7.2114,
-      "step": 279
-    },
-    {
-      "epoch": 8.0,
-      "learning_rate": 0.0025,
-      "loss": 7.2307,
-      "step": 280
-    },
-    {
-      "epoch": 8.03,
-      "learning_rate": 0.002498214285714286,
-      "loss": 7.2477,
-      "step": 281
-    },
-    {
-      "epoch": 8.06,
-      "learning_rate": 0.0024964285714285718,
-      "loss": 7.2069,
-      "step": 282
-    },
-    {
-      "epoch": 8.09,
-      "learning_rate": 0.002494642857142857,
-      "loss": 7.1484,
-      "step": 283
-    },
-    {
-      "epoch": 8.11,
-      "learning_rate": 0.002492857142857143,
-      "loss": 7.1076,
-      "step": 284
-    },
-    {
-      "epoch": 8.14,
-      "learning_rate": 0.002491071428571429,
-      "loss": 7.0819,
-      "step": 285
-    },
-    {
-      "epoch": 8.17,
-      "learning_rate": 0.0024892857142857143,
-      "loss": 7.0708,
-      "step": 286
-    },
-    {
-      "epoch": 8.2,
-      "learning_rate": 0.0024875,
-      "loss": 7.0763,
-      "step": 287
-    },
-    {
-      "epoch": 8.23,
-      "learning_rate": 0.002485714285714286,
-      "loss": 7.0792,
-      "step": 288
-    },
-    {
-      "epoch": 8.26,
-      "learning_rate": 0.0024839285714285714,
-      "loss": 7.1397,
-      "step": 289
-    },
-    {
-      "epoch": 8.29,
-      "learning_rate": 0.0024821428571428572,
-      "loss": 7.0893,
-      "step": 290
-    },
-    {
-      "epoch": 8.31,
-      "learning_rate": 0.0024803571428571427,
-      "loss": 7.1263,
-      "step": 291
-    },
-    {
-      "epoch": 8.34,
-      "learning_rate": 0.0024785714285714285,
-      "loss": 7.0226,
-      "step": 292
-    },
-    {
-      "epoch": 8.37,
-      "learning_rate": 0.0024767857142857144,
-      "loss": 7.1017,
-      "step": 293
-    },
-    {
-      "epoch": 8.4,
-      "learning_rate": 0.0024749999999999998,
-      "loss": 7.0161,
-      "step": 294
-    },
-    {
-      "epoch": 8.43,
-      "learning_rate": 0.0024732142857142856,
-      "loss": 7.117,
-      "step": 295
-    },
-    {
-      "epoch": 8.46,
-      "learning_rate": 0.0024714285714285715,
-      "loss": 7.0234,
-      "step": 296
-    },
-    {
-      "epoch": 8.49,
-      "learning_rate": 0.002469642857142857,
-      "loss": 7.0663,
-      "step": 297
-    },
-    {
-      "epoch": 8.51,
-      "learning_rate": 0.0024678571428571427,
-      "loss": 7.1604,
-      "step": 298
-    },
-    {
-      "epoch": 8.54,
-      "learning_rate": 0.0024660714285714286,
-      "loss": 7.0543,
-      "step": 299
-    },
-    {
-      "epoch": 8.57,
-      "learning_rate": 0.0024642857142857144,
-      "loss": 7.0131,
-      "step": 300
-    },
-    {
-      "epoch": 8.6,
-      "learning_rate": 0.0024625,
-      "loss": 7.0294,
-      "step": 301
-    },
-    {
-      "epoch": 8.63,
-      "learning_rate": 0.0024607142857142857,
-      "loss": 7.0273,
-      "step": 302
-    },
-    {
-      "epoch": 8.66,
-      "learning_rate": 0.0024589285714285715,
-      "loss": 7.0074,
-      "step": 303
-    },
-    {
-      "epoch": 8.69,
-      "learning_rate": 0.002457142857142857,
-      "loss": 6.9747,
-      "step": 304
-    },
-    {
-      "epoch": 8.71,
-      "learning_rate": 0.002455357142857143,
-      "loss": 7.0617,
-      "step": 305
-    },
-    {
-      "epoch": 8.74,
-      "learning_rate": 0.0024535714285714287,
-      "loss": 7.0907,
-      "step": 306
-    },
-    {
-      "epoch": 8.77,
-      "learning_rate": 0.002451785714285714,
-      "loss": 7.0037,
-      "step": 307
-    },
-    {
-      "epoch": 8.8,
-      "learning_rate": 0.00245,
-      "loss": 6.969,
-      "step": 308
-    },
-    {
-      "epoch": 8.83,
-      "learning_rate": 0.0024482142857142858,
-      "loss": 7.0575,
-      "step": 309
-    },
-    {
-      "epoch": 8.86,
-      "learning_rate": 0.0024464285714285716,
-      "loss": 6.9494,
-      "step": 310
-    },
-    {
-      "epoch": 8.89,
-      "learning_rate": 0.002444642857142857,
-      "loss": 6.969,
-      "step": 311
-    },
-    {
-      "epoch": 8.91,
-      "learning_rate": 0.002442857142857143,
-      "loss": 6.8827,
-      "step": 312
-    },
-    {
-      "epoch": 8.94,
-      "learning_rate": 0.0024410714285714287,
-      "loss": 6.9058,
-      "step": 313
-    },
-    {
-      "epoch": 8.97,
-      "learning_rate": 0.002439285714285714,
-      "loss": 6.8808,
-      "step": 314
-    },
-    {
-      "epoch": 9.0,
-      "learning_rate": 0.0024375,
-      "loss": 6.9516,
-      "step": 315
-    },
-    {
-      "epoch": 9.03,
-      "learning_rate": 0.002435714285714286,
-      "loss": 6.9132,
-      "step": 316
-    },
-    {
-      "epoch": 9.06,
-      "learning_rate": 0.0024339285714285717,
-      "loss": 6.9058,
-      "step": 317
-    },
-    {
-      "epoch": 9.09,
-      "learning_rate": 0.002432142857142857,
-      "loss": 6.9332,
-      "step": 318
-    },
-    {
-      "epoch": 9.11,
-      "learning_rate": 0.002430357142857143,
-      "loss": 6.9757,
-      "step": 319
-    },
-    {
-      "epoch": 9.14,
-      "learning_rate": 0.002428571428571429,
-      "loss": 6.8261,
-      "step": 320
-    },
-    {
-      "epoch": 9.17,
-      "learning_rate": 0.0024267857142857142,
-      "loss": 6.8571,
-      "step": 321
-    },
-    {
-      "epoch": 9.2,
-      "learning_rate": 0.002425,
-      "loss": 6.8435,
-      "step": 322
-    },
-    {
-      "epoch": 9.23,
-      "learning_rate": 0.002423214285714286,
-      "loss": 6.9033,
-      "step": 323
-    },
-    {
-      "epoch": 9.26,
-      "learning_rate": 0.0024214285714285713,
-      "loss": 6.8042,
-      "step": 324
-    },
-    {
-      "epoch": 9.29,
-      "learning_rate": 0.002419642857142857,
-      "loss": 6.8732,
-      "step": 325
-    },
-    {
-      "epoch": 9.31,
-      "learning_rate": 0.002417857142857143,
-      "loss": 6.752,
-      "step": 326
-    },
-    {
-      "epoch": 9.34,
-      "learning_rate": 0.002416071428571429,
-      "loss": 6.8016,
-      "step": 327
-    },
-    {
-      "epoch": 9.37,
-      "learning_rate": 0.0024142857142857143,
-      "loss": 6.8879,
-      "step": 328
-    },
-    {
-      "epoch": 9.4,
-      "learning_rate": 0.0024125,
-      "loss": 6.7643,
-      "step": 329
-    },
-    {
-      "epoch": 9.43,
-      "learning_rate": 0.002410714285714286,
-      "loss": 6.7084,
-      "step": 330
-    },
-    {
-      "epoch": 9.46,
-      "learning_rate": 0.0024089285714285714,
-      "loss": 6.8049,
-      "step": 331
-    },
-    {
-      "epoch": 9.49,
-      "learning_rate": 0.0024071428571428573,
-      "loss": 6.7925,
-      "step": 332
-    },
-    {
-      "epoch": 9.51,
-      "learning_rate": 0.002405357142857143,
-      "loss": 6.7289,
-      "step": 333
-    },
-    {
-      "epoch": 9.54,
-      "learning_rate": 0.0024035714285714285,
-      "loss": 6.7439,
-      "step": 334
-    },
-    {
-      "epoch": 9.57,
-      "learning_rate": 0.0024017857142857144,
-      "loss": 6.7119,
-      "step": 335
-    },
-    {
-      "epoch": 9.6,
-      "learning_rate": 0.0024000000000000002,
-      "loss": 6.7251,
-      "step": 336
-    },
-    {
-      "epoch": 9.63,
-      "learning_rate": 0.002398214285714286,
-      "loss": 6.6659,
-      "step": 337
-    },
-    {
-      "epoch": 9.66,
-      "learning_rate": 0.0023964285714285715,
-      "loss": 6.7422,
-      "step": 338
-    },
-    {
-      "epoch": 9.69,
-      "learning_rate": 0.0023946428571428573,
-      "loss": 6.7852,
-      "step": 339
-    },
-    {
-      "epoch": 9.71,
-      "learning_rate": 0.002392857142857143,
-      "loss": 6.6828,
-      "step": 340
-    },
-    {
-      "epoch": 9.74,
-      "learning_rate": 0.0023910714285714286,
-      "loss": 6.686,
-      "step": 341
-    },
-    {
-      "epoch": 9.77,
-      "learning_rate": 0.002389285714285714,
-      "loss": 6.7326,
-      "step": 342
-    },
-    {
-      "epoch": 9.8,
-      "learning_rate": 0.0023875,
-      "loss": 6.5601,
-      "step": 343
-    },
-    {
-      "epoch": 9.83,
-      "learning_rate": 0.0023857142857142857,
-      "loss": 6.6646,
-      "step": 344
-    },
-    {
-      "epoch": 9.86,
-      "learning_rate": 0.002383928571428571,
-      "loss": 6.5673,
-      "step": 345
-    },
-    {
-      "epoch": 9.89,
-      "learning_rate": 0.002382142857142857,
-      "loss": 6.6227,
-      "step": 346
-    },
-    {
-      "epoch": 9.91,
-      "learning_rate": 0.002380357142857143,
-      "loss": 6.5526,
-      "step": 347
-    },
-    {
-      "epoch": 9.94,
-      "learning_rate": 0.0023785714285714287,
-      "loss": 6.6842,
-      "step": 348
-    },
-    {
-      "epoch": 9.97,
-      "learning_rate": 0.002376785714285714,
-      "loss": 6.6211,
-      "step": 349
-    },
-    {
-      "epoch": 10.0,
-      "learning_rate": 0.002375,
-      "loss": 6.6952,
-      "step": 350
-    },
-    {
-      "epoch": 10.03,
-      "learning_rate": 0.002373214285714286,
-      "loss": 6.5324,
-      "step": 351
-    },
-    {
-      "epoch": 10.06,
-      "learning_rate": 0.002371428571428571,
-      "loss": 6.5792,
-      "step": 352
-    },
-    {
-      "epoch": 10.09,
-      "learning_rate": 0.002369642857142857,
-      "loss": 6.5276,
-      "step": 353
-    },
-    {
-      "epoch": 10.11,
-      "learning_rate": 0.002367857142857143,
-      "loss": 6.5634,
-      "step": 354
-    },
-    {
-      "epoch": 10.14,
-      "learning_rate": 0.0023660714285714288,
-      "loss": 6.5385,
-      "step": 355
-    },
-    {
-      "epoch": 10.17,
-      "learning_rate": 0.002364285714285714,
-      "loss": 6.4516,
-      "step": 356
-    },
-    {
-      "epoch": 10.2,
-      "learning_rate": 0.0023625,
-      "loss": 6.5641,
-      "step": 357
-    },
-    {
-      "epoch": 10.23,
-      "learning_rate": 0.002360714285714286,
-      "loss": 6.5001,
-      "step": 358
-    },
-    {
-      "epoch": 10.26,
-      "learning_rate": 0.0023589285714285713,
-      "loss": 6.4846,
-      "step": 359
-    },
-    {
-      "epoch": 10.29,
-      "learning_rate": 0.002357142857142857,
-      "loss": 6.4638,
-      "step": 360
-    },
-    {
-      "epoch": 10.31,
-      "learning_rate": 0.002355357142857143,
-      "loss": 6.5217,
-      "step": 361
-    },
-    {
-      "epoch": 10.34,
-      "learning_rate": 0.0023535714285714284,
-      "loss": 6.5444,
-      "step": 362
-    },
-    {
-      "epoch": 10.37,
-      "learning_rate": 0.0023517857142857142,
-      "loss": 6.496,
-      "step": 363
-    },
-    {
-      "epoch": 10.4,
-      "learning_rate": 0.00235,
-      "loss": 6.5345,
-      "step": 364
-    },
-    {
-      "epoch": 10.43,
-      "learning_rate": 0.002348214285714286,
-      "loss": 6.4732,
-      "step": 365
-    },
-    {
-      "epoch": 10.46,
-      "learning_rate": 0.0023464285714285714,
-      "loss": 6.4765,
-      "step": 366
-    },
-    {
-      "epoch": 10.49,
-      "learning_rate": 0.002344642857142857,
-      "loss": 6.3881,
-      "step": 367
-    },
-    {
-      "epoch": 10.51,
-      "learning_rate": 0.002342857142857143,
-      "loss": 6.4908,
-      "step": 368
-    },
-    {
-      "epoch": 10.54,
-      "learning_rate": 0.0023410714285714285,
-      "loss": 6.4593,
-      "step": 369
-    },
-    {
-      "epoch": 10.57,
-      "learning_rate": 0.0023392857142857143,
-      "loss": 6.5006,
-      "step": 370
-    },
-    {
-      "epoch": 10.6,
-      "learning_rate": 0.0023375,
-      "loss": 6.4495,
-      "step": 371
-    },
-    {
-      "epoch": 10.63,
-      "learning_rate": 0.0023357142857142856,
-      "loss": 6.3569,
-      "step": 372
-    },
-    {
-      "epoch": 10.66,
-      "learning_rate": 0.0023339285714285714,
-      "loss": 6.3592,
-      "step": 373
-    },
-    {
-      "epoch": 10.69,
-      "learning_rate": 0.0023321428571428573,
-      "loss": 6.3258,
-      "step": 374
-    },
-    {
-      "epoch": 10.71,
-      "learning_rate": 0.002330357142857143,
-      "loss": 6.3216,
-      "step": 375
-    },
-    {
-      "epoch": 10.74,
-      "learning_rate": 0.0023285714285714285,
-      "loss": 6.4878,
-      "step": 376
-    },
-    {
-      "epoch": 10.77,
-      "learning_rate": 0.0023267857142857144,
-      "loss": 6.3412,
-      "step": 377
-    },
-    {
-      "epoch": 10.8,
-      "learning_rate": 0.0023250000000000002,
-      "loss": 6.3925,
-      "step": 378
-    },
-    {
-      "epoch": 10.83,
-      "learning_rate": 0.0023232142857142857,
-      "loss": 6.275,
-      "step": 379
-    },
-    {
-      "epoch": 10.86,
-      "learning_rate": 0.0023214285714285715,
-      "loss": 6.3575,
-      "step": 380
-    },
-    {
-      "epoch": 10.89,
-      "learning_rate": 0.0023196428571428574,
-      "loss": 6.3259,
-      "step": 381
-    },
-    {
-      "epoch": 10.91,
-      "learning_rate": 0.002317857142857143,
-      "loss": 6.315,
-      "step": 382
-    },
-    {
-      "epoch": 10.94,
-      "learning_rate": 0.0023160714285714286,
-      "loss": 6.277,
-      "step": 383
-    },
-    {
-      "epoch": 10.97,
-      "learning_rate": 0.0023142857142857145,
-      "loss": 6.3259,
-      "step": 384
-    },
-    {
-      "epoch": 11.0,
-      "learning_rate": 0.0023125000000000003,
-      "loss": 6.3747,
-      "step": 385
-    },
-    {
-      "epoch": 11.03,
-      "learning_rate": 0.0023107142857142857,
-      "loss": 6.3646,
-      "step": 386
-    },
-    {
-      "epoch": 11.06,
-      "learning_rate": 0.0023089285714285716,
-      "loss": 6.3687,
-      "step": 387
-    },
-    {
-      "epoch": 11.09,
-      "learning_rate": 0.0023071428571428574,
-      "loss": 6.3374,
-      "step": 388
-    },
-    {
-      "epoch": 11.11,
-      "learning_rate": 0.002305357142857143,
-      "loss": 6.3129,
-      "step": 389
-    },
-    {
-      "epoch": 11.14,
-      "learning_rate": 0.0023035714285714287,
-      "loss": 6.3425,
-      "step": 390
-    },
-    {
-      "epoch": 11.17,
-      "learning_rate": 0.0023017857142857145,
-      "loss": 6.2122,
-      "step": 391
-    },
-    {
-      "epoch": 11.2,
-      "learning_rate": 0.0023000000000000004,
-      "loss": 6.2768,
-      "step": 392
-    },
-    {
-      "epoch": 11.23,
-      "learning_rate": 0.002298214285714286,
-      "loss": 6.2853,
-      "step": 393
-    },
-    {
-      "epoch": 11.26,
-      "learning_rate": 0.0022964285714285712,
-      "loss": 6.3215,
-      "step": 394
-    },
-    {
-      "epoch": 11.29,
-      "learning_rate": 0.002294642857142857,
-      "loss": 6.3244,
-      "step": 395
-    },
-    {
-      "epoch": 11.31,
-      "learning_rate": 0.002292857142857143,
-      "loss": 6.2399,
-      "step": 396
-    },
-    {
-      "epoch": 11.34,
-      "learning_rate": 0.0022910714285714283,
-      "loss": 6.2457,
-      "step": 397
-    },
-    {
-      "epoch": 11.37,
-      "learning_rate": 0.002289285714285714,
-      "loss": 6.2018,
-      "step": 398
-    },
-    {
-      "epoch": 11.4,
-      "learning_rate": 0.0022875,
-      "loss": 6.2101,
-      "step": 399
-    },
-    {
-      "epoch": 11.43,
-      "learning_rate": 0.0022857142857142855,
-      "loss": 6.2257,
-      "step": 400
     }
   ],
   "logging_steps": 1,
   "max_steps": 1680,
   "num_train_epochs": 48,
   "save_steps": 100,
-  "total_flos": 2.3080655169481728e+17,
+  "total_flos": 1.1559978611371008e+17,
   "trial_name": null,
   "trial_params": null
 }