kevinoli commited on
Commit
8c97585
1 Parent(s): bc0b76b

Training in progress, step 24000, checkpoint

Browse files
checkpoint-24000/config.json ADDED
@@ -0,0 +1,31 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "openai/clip-vit-large-patch14-336",
3
+ "architectures": [
4
+ "CLIPModel"
5
+ ],
6
+ "initializer_factor": 1.0,
7
+ "logit_scale_init_value": 2.6592,
8
+ "model_type": "clip",
9
+ "projection_dim": 768,
10
+ "text_config": {
11
+ "dropout": 0.0,
12
+ "hidden_size": 768,
13
+ "intermediate_size": 3072,
14
+ "model_type": "clip_text_model",
15
+ "num_attention_heads": 12,
16
+ "projection_dim": 768
17
+ },
18
+ "torch_dtype": "float32",
19
+ "transformers_version": "4.45.0.dev0",
20
+ "vision_config": {
21
+ "dropout": 0.0,
22
+ "hidden_size": 1024,
23
+ "image_size": 336,
24
+ "intermediate_size": 4096,
25
+ "model_type": "clip_vision_model",
26
+ "num_attention_heads": 16,
27
+ "num_hidden_layers": 24,
28
+ "patch_size": 14,
29
+ "projection_dim": 768
30
+ }
31
+ }
checkpoint-24000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d43cfd0d5d02ed77923c0a899a940e40f42d6e8097963f5acfcf39c1d7f3a460
3
+ size 1711848436
checkpoint-24000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c546615f99105f9aed0353292fc0878e1d4dedbd8dc45ebd991c463370078e7b
3
+ size 3424043887
checkpoint-24000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:77edbdd8df51c0d06d947e3380871965f0ebcd4c2180d6de0878d158327a6de7
3
+ size 14503
checkpoint-24000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a4e50baa31c178c87d15140869d27fb47b4ca05884fb4600ad6d60518832e285
3
+ size 623
checkpoint-24000/trainer_state.json ADDED
@@ -0,0 +1,753 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 1.496930480003357,
3
+ "best_model_checkpoint": "./output/clip-finetuned-csu-p14-336-e5l510-l/checkpoint-24000",
4
+ "epoch": 4.421518054532056,
5
+ "eval_steps": 500,
6
+ "global_step": 24000,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.09211495946941783,
13
+ "grad_norm": 21.660480499267578,
14
+ "learning_rate": 4.907885040530583e-10,
15
+ "loss": 0.3974,
16
+ "step": 500
17
+ },
18
+ {
19
+ "epoch": 0.09211495946941783,
20
+ "eval_loss": 1.5028706789016724,
21
+ "eval_runtime": 74.0662,
22
+ "eval_samples_per_second": 16.296,
23
+ "eval_steps_per_second": 2.039,
24
+ "step": 500
25
+ },
26
+ {
27
+ "epoch": 0.18422991893883567,
28
+ "grad_norm": 60.26338577270508,
29
+ "learning_rate": 4.815770081061165e-10,
30
+ "loss": 0.462,
31
+ "step": 1000
32
+ },
33
+ {
34
+ "epoch": 0.18422991893883567,
35
+ "eval_loss": 1.5025668144226074,
36
+ "eval_runtime": 75.1835,
37
+ "eval_samples_per_second": 16.054,
38
+ "eval_steps_per_second": 2.008,
39
+ "step": 1000
40
+ },
41
+ {
42
+ "epoch": 0.2763448784082535,
43
+ "grad_norm": 1.4199143648147583,
44
+ "learning_rate": 4.723655121591747e-10,
45
+ "loss": 0.5249,
46
+ "step": 1500
47
+ },
48
+ {
49
+ "epoch": 0.2763448784082535,
50
+ "eval_loss": 1.5022265911102295,
51
+ "eval_runtime": 76.3442,
52
+ "eval_samples_per_second": 15.81,
53
+ "eval_steps_per_second": 1.978,
54
+ "step": 1500
55
+ },
56
+ {
57
+ "epoch": 0.36845983787767134,
58
+ "grad_norm": 1.2705016136169434,
59
+ "learning_rate": 4.6315401621223287e-10,
60
+ "loss": 0.4642,
61
+ "step": 2000
62
+ },
63
+ {
64
+ "epoch": 0.36845983787767134,
65
+ "eval_loss": 1.5019283294677734,
66
+ "eval_runtime": 77.0512,
67
+ "eval_samples_per_second": 15.665,
68
+ "eval_steps_per_second": 1.96,
69
+ "step": 2000
70
+ },
71
+ {
72
+ "epoch": 0.46057479734708917,
73
+ "grad_norm": 335.6837463378906,
74
+ "learning_rate": 4.5394252026529114e-10,
75
+ "loss": 0.5056,
76
+ "step": 2500
77
+ },
78
+ {
79
+ "epoch": 0.46057479734708917,
80
+ "eval_loss": 1.5015931129455566,
81
+ "eval_runtime": 76.8527,
82
+ "eval_samples_per_second": 15.705,
83
+ "eval_steps_per_second": 1.965,
84
+ "step": 2500
85
+ },
86
+ {
87
+ "epoch": 0.552689756816507,
88
+ "grad_norm": 402.1738586425781,
89
+ "learning_rate": 4.4473102431834935e-10,
90
+ "loss": 0.4604,
91
+ "step": 3000
92
+ },
93
+ {
94
+ "epoch": 0.552689756816507,
95
+ "eval_loss": 1.501311182975769,
96
+ "eval_runtime": 76.7728,
97
+ "eval_samples_per_second": 15.722,
98
+ "eval_steps_per_second": 1.967,
99
+ "step": 3000
100
+ },
101
+ {
102
+ "epoch": 0.6448047162859248,
103
+ "grad_norm": 120.70919799804688,
104
+ "learning_rate": 4.3551952837140756e-10,
105
+ "loss": 0.4508,
106
+ "step": 3500
107
+ },
108
+ {
109
+ "epoch": 0.6448047162859248,
110
+ "eval_loss": 1.5010316371917725,
111
+ "eval_runtime": 77.0074,
112
+ "eval_samples_per_second": 15.674,
113
+ "eval_steps_per_second": 1.961,
114
+ "step": 3500
115
+ },
116
+ {
117
+ "epoch": 0.7369196757553427,
118
+ "grad_norm": 103.2845687866211,
119
+ "learning_rate": 4.2630803242446577e-10,
120
+ "loss": 0.4674,
121
+ "step": 4000
122
+ },
123
+ {
124
+ "epoch": 0.7369196757553427,
125
+ "eval_loss": 1.5007821321487427,
126
+ "eval_runtime": 76.7277,
127
+ "eval_samples_per_second": 15.731,
128
+ "eval_steps_per_second": 1.968,
129
+ "step": 4000
130
+ },
131
+ {
132
+ "epoch": 0.8290346352247605,
133
+ "grad_norm": 464.5989074707031,
134
+ "learning_rate": 4.17096536477524e-10,
135
+ "loss": 0.4435,
136
+ "step": 4500
137
+ },
138
+ {
139
+ "epoch": 0.8290346352247605,
140
+ "eval_loss": 1.500537633895874,
141
+ "eval_runtime": 76.8855,
142
+ "eval_samples_per_second": 15.699,
143
+ "eval_steps_per_second": 1.964,
144
+ "step": 4500
145
+ },
146
+ {
147
+ "epoch": 0.9211495946941783,
148
+ "grad_norm": 6.3550262451171875,
149
+ "learning_rate": 4.078850405305822e-10,
150
+ "loss": 0.4573,
151
+ "step": 5000
152
+ },
153
+ {
154
+ "epoch": 0.9211495946941783,
155
+ "eval_loss": 1.5003185272216797,
156
+ "eval_runtime": 77.2072,
157
+ "eval_samples_per_second": 15.633,
158
+ "eval_steps_per_second": 1.956,
159
+ "step": 5000
160
+ },
161
+ {
162
+ "epoch": 1.013264554163596,
163
+ "grad_norm": 0.6333007216453552,
164
+ "learning_rate": 3.9867354458364045e-10,
165
+ "loss": 0.4681,
166
+ "step": 5500
167
+ },
168
+ {
169
+ "epoch": 1.013264554163596,
170
+ "eval_loss": 1.5001204013824463,
171
+ "eval_runtime": 76.9383,
172
+ "eval_samples_per_second": 15.688,
173
+ "eval_steps_per_second": 1.963,
174
+ "step": 5500
175
+ },
176
+ {
177
+ "epoch": 1.105379513633014,
178
+ "grad_norm": 238.3988800048828,
179
+ "learning_rate": 3.894620486366986e-10,
180
+ "loss": 0.4862,
181
+ "step": 6000
182
+ },
183
+ {
184
+ "epoch": 1.105379513633014,
185
+ "eval_loss": 1.4999229907989502,
186
+ "eval_runtime": 76.9483,
187
+ "eval_samples_per_second": 15.686,
188
+ "eval_steps_per_second": 1.962,
189
+ "step": 6000
190
+ },
191
+ {
192
+ "epoch": 1.1974944731024317,
193
+ "grad_norm": 492.2626953125,
194
+ "learning_rate": 3.8025055268975687e-10,
195
+ "loss": 0.4231,
196
+ "step": 6500
197
+ },
198
+ {
199
+ "epoch": 1.1974944731024317,
200
+ "eval_loss": 1.4997295141220093,
201
+ "eval_runtime": 75.1791,
202
+ "eval_samples_per_second": 16.055,
203
+ "eval_steps_per_second": 2.009,
204
+ "step": 6500
205
+ },
206
+ {
207
+ "epoch": 1.2896094325718497,
208
+ "grad_norm": 3.0739829540252686,
209
+ "learning_rate": 3.7103905674281503e-10,
210
+ "loss": 0.3734,
211
+ "step": 7000
212
+ },
213
+ {
214
+ "epoch": 1.2896094325718497,
215
+ "eval_loss": 1.4995641708374023,
216
+ "eval_runtime": 76.6897,
217
+ "eval_samples_per_second": 15.739,
218
+ "eval_steps_per_second": 1.969,
219
+ "step": 7000
220
+ },
221
+ {
222
+ "epoch": 1.3817243920412676,
223
+ "grad_norm": 142.28684997558594,
224
+ "learning_rate": 3.618275607958733e-10,
225
+ "loss": 0.4606,
226
+ "step": 7500
227
+ },
228
+ {
229
+ "epoch": 1.3817243920412676,
230
+ "eval_loss": 1.4993607997894287,
231
+ "eval_runtime": 75.831,
232
+ "eval_samples_per_second": 15.917,
233
+ "eval_steps_per_second": 1.991,
234
+ "step": 7500
235
+ },
236
+ {
237
+ "epoch": 1.4738393515106853,
238
+ "grad_norm": 1.5455149412155151,
239
+ "learning_rate": 3.526160648489315e-10,
240
+ "loss": 0.467,
241
+ "step": 8000
242
+ },
243
+ {
244
+ "epoch": 1.4738393515106853,
245
+ "eval_loss": 1.4991577863693237,
246
+ "eval_runtime": 75.1339,
247
+ "eval_samples_per_second": 16.065,
248
+ "eval_steps_per_second": 2.01,
249
+ "step": 8000
250
+ },
251
+ {
252
+ "epoch": 1.565954310980103,
253
+ "grad_norm": 215.83163452148438,
254
+ "learning_rate": 3.434045689019897e-10,
255
+ "loss": 0.4202,
256
+ "step": 8500
257
+ },
258
+ {
259
+ "epoch": 1.565954310980103,
260
+ "eval_loss": 1.498979091644287,
261
+ "eval_runtime": 75.4004,
262
+ "eval_samples_per_second": 16.008,
263
+ "eval_steps_per_second": 2.003,
264
+ "step": 8500
265
+ },
266
+ {
267
+ "epoch": 1.658069270449521,
268
+ "grad_norm": 2.8240511417388916,
269
+ "learning_rate": 3.341930729550479e-10,
270
+ "loss": 0.4866,
271
+ "step": 9000
272
+ },
273
+ {
274
+ "epoch": 1.658069270449521,
275
+ "eval_loss": 1.4988232851028442,
276
+ "eval_runtime": 75.0613,
277
+ "eval_samples_per_second": 16.08,
278
+ "eval_steps_per_second": 2.012,
279
+ "step": 9000
280
+ },
281
+ {
282
+ "epoch": 1.750184229918939,
283
+ "grad_norm": 7.242213249206543,
284
+ "learning_rate": 3.2498157700810614e-10,
285
+ "loss": 0.4848,
286
+ "step": 9500
287
+ },
288
+ {
289
+ "epoch": 1.750184229918939,
290
+ "eval_loss": 1.4986662864685059,
291
+ "eval_runtime": 75.3419,
292
+ "eval_samples_per_second": 16.02,
293
+ "eval_steps_per_second": 2.004,
294
+ "step": 9500
295
+ },
296
+ {
297
+ "epoch": 1.8422991893883567,
298
+ "grad_norm": 908.7202758789062,
299
+ "learning_rate": 3.1577008106116435e-10,
300
+ "loss": 0.4385,
301
+ "step": 10000
302
+ },
303
+ {
304
+ "epoch": 1.8422991893883567,
305
+ "eval_loss": 1.4985153675079346,
306
+ "eval_runtime": 75.0638,
307
+ "eval_samples_per_second": 16.08,
308
+ "eval_steps_per_second": 2.012,
309
+ "step": 10000
310
+ },
311
+ {
312
+ "epoch": 1.9344141488577744,
313
+ "grad_norm": 4.991658687591553,
314
+ "learning_rate": 3.065585851142226e-10,
315
+ "loss": 0.4781,
316
+ "step": 10500
317
+ },
318
+ {
319
+ "epoch": 1.9344141488577744,
320
+ "eval_loss": 1.498375654220581,
321
+ "eval_runtime": 75.0746,
322
+ "eval_samples_per_second": 16.077,
323
+ "eval_steps_per_second": 2.011,
324
+ "step": 10500
325
+ },
326
+ {
327
+ "epoch": 2.026529108327192,
328
+ "grad_norm": 0.4935278594493866,
329
+ "learning_rate": 2.9734708916728077e-10,
330
+ "loss": 0.4287,
331
+ "step": 11000
332
+ },
333
+ {
334
+ "epoch": 2.026529108327192,
335
+ "eval_loss": 1.4982625246047974,
336
+ "eval_runtime": 75.5519,
337
+ "eval_samples_per_second": 15.976,
338
+ "eval_steps_per_second": 1.999,
339
+ "step": 11000
340
+ },
341
+ {
342
+ "epoch": 2.1186440677966103,
343
+ "grad_norm": 441.09906005859375,
344
+ "learning_rate": 2.8813559322033903e-10,
345
+ "loss": 0.4392,
346
+ "step": 11500
347
+ },
348
+ {
349
+ "epoch": 2.1186440677966103,
350
+ "eval_loss": 1.4981573820114136,
351
+ "eval_runtime": 75.0431,
352
+ "eval_samples_per_second": 16.084,
353
+ "eval_steps_per_second": 2.012,
354
+ "step": 11500
355
+ },
356
+ {
357
+ "epoch": 2.210759027266028,
358
+ "grad_norm": 1.5746198892593384,
359
+ "learning_rate": 2.789240972733972e-10,
360
+ "loss": 0.4447,
361
+ "step": 12000
362
+ },
363
+ {
364
+ "epoch": 2.210759027266028,
365
+ "eval_loss": 1.498041033744812,
366
+ "eval_runtime": 75.1277,
367
+ "eval_samples_per_second": 16.066,
368
+ "eval_steps_per_second": 2.01,
369
+ "step": 12000
370
+ },
371
+ {
372
+ "epoch": 2.3028739867354457,
373
+ "grad_norm": 604.6005249023438,
374
+ "learning_rate": 2.6971260132645545e-10,
375
+ "loss": 0.4449,
376
+ "step": 12500
377
+ },
378
+ {
379
+ "epoch": 2.3028739867354457,
380
+ "eval_loss": 1.497951865196228,
381
+ "eval_runtime": 75.2241,
382
+ "eval_samples_per_second": 16.045,
383
+ "eval_steps_per_second": 2.007,
384
+ "step": 12500
385
+ },
386
+ {
387
+ "epoch": 2.3949889462048635,
388
+ "grad_norm": 5.409970760345459,
389
+ "learning_rate": 2.6050110537951366e-10,
390
+ "loss": 0.4748,
391
+ "step": 13000
392
+ },
393
+ {
394
+ "epoch": 2.3949889462048635,
395
+ "eval_loss": 1.4978498220443726,
396
+ "eval_runtime": 75.3614,
397
+ "eval_samples_per_second": 16.016,
398
+ "eval_steps_per_second": 2.004,
399
+ "step": 13000
400
+ },
401
+ {
402
+ "epoch": 2.4871039056742816,
403
+ "grad_norm": 17.807907104492188,
404
+ "learning_rate": 2.5128960943257187e-10,
405
+ "loss": 0.4935,
406
+ "step": 13500
407
+ },
408
+ {
409
+ "epoch": 2.4871039056742816,
410
+ "eval_loss": 1.4977525472640991,
411
+ "eval_runtime": 74.8827,
412
+ "eval_samples_per_second": 16.119,
413
+ "eval_steps_per_second": 2.016,
414
+ "step": 13500
415
+ },
416
+ {
417
+ "epoch": 2.5792188651436994,
418
+ "grad_norm": 190.154541015625,
419
+ "learning_rate": 2.420781134856301e-10,
420
+ "loss": 0.4113,
421
+ "step": 14000
422
+ },
423
+ {
424
+ "epoch": 2.5792188651436994,
425
+ "eval_loss": 1.4976625442504883,
426
+ "eval_runtime": 76.3316,
427
+ "eval_samples_per_second": 15.813,
428
+ "eval_steps_per_second": 1.978,
429
+ "step": 14000
430
+ },
431
+ {
432
+ "epoch": 2.671333824613117,
433
+ "grad_norm": 3.5760951042175293,
434
+ "learning_rate": 2.328666175386883e-10,
435
+ "loss": 0.3934,
436
+ "step": 14500
437
+ },
438
+ {
439
+ "epoch": 2.671333824613117,
440
+ "eval_loss": 1.4975850582122803,
441
+ "eval_runtime": 76.0918,
442
+ "eval_samples_per_second": 15.862,
443
+ "eval_steps_per_second": 1.984,
444
+ "step": 14500
445
+ },
446
+ {
447
+ "epoch": 2.7634487840825352,
448
+ "grad_norm": 195.72146606445312,
449
+ "learning_rate": 2.236551215917465e-10,
450
+ "loss": 0.4901,
451
+ "step": 15000
452
+ },
453
+ {
454
+ "epoch": 2.7634487840825352,
455
+ "eval_loss": 1.4975062608718872,
456
+ "eval_runtime": 74.9535,
457
+ "eval_samples_per_second": 16.103,
458
+ "eval_steps_per_second": 2.015,
459
+ "step": 15000
460
+ },
461
+ {
462
+ "epoch": 2.855563743551953,
463
+ "grad_norm": 0.14779114723205566,
464
+ "learning_rate": 2.1444362564480471e-10,
465
+ "loss": 0.4601,
466
+ "step": 15500
467
+ },
468
+ {
469
+ "epoch": 2.855563743551953,
470
+ "eval_loss": 1.497436761856079,
471
+ "eval_runtime": 75.9666,
472
+ "eval_samples_per_second": 15.889,
473
+ "eval_steps_per_second": 1.988,
474
+ "step": 15500
475
+ },
476
+ {
477
+ "epoch": 2.9476787030213707,
478
+ "grad_norm": 205.7298583984375,
479
+ "learning_rate": 2.0523212969786292e-10,
480
+ "loss": 0.4224,
481
+ "step": 16000
482
+ },
483
+ {
484
+ "epoch": 2.9476787030213707,
485
+ "eval_loss": 1.4973747730255127,
486
+ "eval_runtime": 76.7648,
487
+ "eval_samples_per_second": 15.723,
488
+ "eval_steps_per_second": 1.967,
489
+ "step": 16000
490
+ },
491
+ {
492
+ "epoch": 3.0397936624907884,
493
+ "grad_norm": 9.98619270324707,
494
+ "learning_rate": 1.9602063375092116e-10,
495
+ "loss": 0.4381,
496
+ "step": 16500
497
+ },
498
+ {
499
+ "epoch": 3.0397936624907884,
500
+ "eval_loss": 1.497320294380188,
501
+ "eval_runtime": 75.2158,
502
+ "eval_samples_per_second": 16.047,
503
+ "eval_steps_per_second": 2.008,
504
+ "step": 16500
505
+ },
506
+ {
507
+ "epoch": 3.131908621960206,
508
+ "grad_norm": 5.3734331130981445,
509
+ "learning_rate": 1.8680913780397937e-10,
510
+ "loss": 0.4942,
511
+ "step": 17000
512
+ },
513
+ {
514
+ "epoch": 3.131908621960206,
515
+ "eval_loss": 1.4972634315490723,
516
+ "eval_runtime": 75.2534,
517
+ "eval_samples_per_second": 16.039,
518
+ "eval_steps_per_second": 2.007,
519
+ "step": 17000
520
+ },
521
+ {
522
+ "epoch": 3.2240235814296243,
523
+ "grad_norm": 117.80599212646484,
524
+ "learning_rate": 1.7759764185703758e-10,
525
+ "loss": 0.3885,
526
+ "step": 17500
527
+ },
528
+ {
529
+ "epoch": 3.2240235814296243,
530
+ "eval_loss": 1.4972138404846191,
531
+ "eval_runtime": 75.5586,
532
+ "eval_samples_per_second": 15.974,
533
+ "eval_steps_per_second": 1.998,
534
+ "step": 17500
535
+ },
536
+ {
537
+ "epoch": 3.316138540899042,
538
+ "grad_norm": 81.40491485595703,
539
+ "learning_rate": 1.683861459100958e-10,
540
+ "loss": 0.4407,
541
+ "step": 18000
542
+ },
543
+ {
544
+ "epoch": 3.316138540899042,
545
+ "eval_loss": 1.4971661567687988,
546
+ "eval_runtime": 75.6731,
547
+ "eval_samples_per_second": 15.95,
548
+ "eval_steps_per_second": 1.995,
549
+ "step": 18000
550
+ },
551
+ {
552
+ "epoch": 3.4082535003684598,
553
+ "grad_norm": 16.30617904663086,
554
+ "learning_rate": 1.59174649963154e-10,
555
+ "loss": 0.4635,
556
+ "step": 18500
557
+ },
558
+ {
559
+ "epoch": 3.4082535003684598,
560
+ "eval_loss": 1.4971280097961426,
561
+ "eval_runtime": 75.728,
562
+ "eval_samples_per_second": 15.939,
563
+ "eval_steps_per_second": 1.994,
564
+ "step": 18500
565
+ },
566
+ {
567
+ "epoch": 3.500368459837878,
568
+ "grad_norm": 40.21035385131836,
569
+ "learning_rate": 1.4996315401621224e-10,
570
+ "loss": 0.4228,
571
+ "step": 19000
572
+ },
573
+ {
574
+ "epoch": 3.500368459837878,
575
+ "eval_loss": 1.4970934391021729,
576
+ "eval_runtime": 76.4366,
577
+ "eval_samples_per_second": 15.791,
578
+ "eval_steps_per_second": 1.975,
579
+ "step": 19000
580
+ },
581
+ {
582
+ "epoch": 3.5924834193072956,
583
+ "grad_norm": 7.447707176208496,
584
+ "learning_rate": 1.4075165806927045e-10,
585
+ "loss": 0.464,
586
+ "step": 19500
587
+ },
588
+ {
589
+ "epoch": 3.5924834193072956,
590
+ "eval_loss": 1.4970619678497314,
591
+ "eval_runtime": 75.781,
592
+ "eval_samples_per_second": 15.927,
593
+ "eval_steps_per_second": 1.993,
594
+ "step": 19500
595
+ },
596
+ {
597
+ "epoch": 3.6845983787767134,
598
+ "grad_norm": 0.005426365882158279,
599
+ "learning_rate": 1.3154016212232866e-10,
600
+ "loss": 0.4746,
601
+ "step": 20000
602
+ },
603
+ {
604
+ "epoch": 3.6845983787767134,
605
+ "eval_loss": 1.4970370531082153,
606
+ "eval_runtime": 75.7317,
607
+ "eval_samples_per_second": 15.938,
608
+ "eval_steps_per_second": 1.994,
609
+ "step": 20000
610
+ },
611
+ {
612
+ "epoch": 3.776713338246131,
613
+ "grad_norm": 485.1770935058594,
614
+ "learning_rate": 1.223286661753869e-10,
615
+ "loss": 0.4447,
616
+ "step": 20500
617
+ },
618
+ {
619
+ "epoch": 3.776713338246131,
620
+ "eval_loss": 1.497014045715332,
621
+ "eval_runtime": 76.0421,
622
+ "eval_samples_per_second": 15.873,
623
+ "eval_steps_per_second": 1.986,
624
+ "step": 20500
625
+ },
626
+ {
627
+ "epoch": 3.868828297715549,
628
+ "grad_norm": 11.55502986907959,
629
+ "learning_rate": 1.1311717022844511e-10,
630
+ "loss": 0.4663,
631
+ "step": 21000
632
+ },
633
+ {
634
+ "epoch": 3.868828297715549,
635
+ "eval_loss": 1.4969953298568726,
636
+ "eval_runtime": 76.2357,
637
+ "eval_samples_per_second": 15.832,
638
+ "eval_steps_per_second": 1.981,
639
+ "step": 21000
640
+ },
641
+ {
642
+ "epoch": 3.960943257184967,
643
+ "grad_norm": 21.07769012451172,
644
+ "learning_rate": 1.0390567428150332e-10,
645
+ "loss": 0.4246,
646
+ "step": 21500
647
+ },
648
+ {
649
+ "epoch": 3.960943257184967,
650
+ "eval_loss": 1.4969803094863892,
651
+ "eval_runtime": 76.1279,
652
+ "eval_samples_per_second": 15.855,
653
+ "eval_steps_per_second": 1.984,
654
+ "step": 21500
655
+ },
656
+ {
657
+ "epoch": 4.053058216654384,
658
+ "grad_norm": 284.8838806152344,
659
+ "learning_rate": 9.469417833456154e-11,
660
+ "loss": 0.4592,
661
+ "step": 22000
662
+ },
663
+ {
664
+ "epoch": 4.053058216654384,
665
+ "eval_loss": 1.4969651699066162,
666
+ "eval_runtime": 75.8298,
667
+ "eval_samples_per_second": 15.917,
668
+ "eval_steps_per_second": 1.991,
669
+ "step": 22000
670
+ },
671
+ {
672
+ "epoch": 4.145173176123802,
673
+ "grad_norm": 561.0142211914062,
674
+ "learning_rate": 8.548268238761975e-11,
675
+ "loss": 0.4181,
676
+ "step": 22500
677
+ },
678
+ {
679
+ "epoch": 4.145173176123802,
680
+ "eval_loss": 1.4969545602798462,
681
+ "eval_runtime": 75.8888,
682
+ "eval_samples_per_second": 15.905,
683
+ "eval_steps_per_second": 1.99,
684
+ "step": 22500
685
+ },
686
+ {
687
+ "epoch": 4.237288135593221,
688
+ "grad_norm": 10.654740333557129,
689
+ "learning_rate": 7.627118644067798e-11,
690
+ "loss": 0.4913,
691
+ "step": 23000
692
+ },
693
+ {
694
+ "epoch": 4.237288135593221,
695
+ "eval_loss": 1.4969446659088135,
696
+ "eval_runtime": 75.7275,
697
+ "eval_samples_per_second": 15.939,
698
+ "eval_steps_per_second": 1.994,
699
+ "step": 23000
700
+ },
701
+ {
702
+ "epoch": 4.329403095062638,
703
+ "grad_norm": 160.60205078125,
704
+ "learning_rate": 6.705969049373619e-11,
705
+ "loss": 0.4491,
706
+ "step": 23500
707
+ },
708
+ {
709
+ "epoch": 4.329403095062638,
710
+ "eval_loss": 1.4969370365142822,
711
+ "eval_runtime": 76.3184,
712
+ "eval_samples_per_second": 15.815,
713
+ "eval_steps_per_second": 1.979,
714
+ "step": 23500
715
+ },
716
+ {
717
+ "epoch": 4.421518054532056,
718
+ "grad_norm": 374.78900146484375,
719
+ "learning_rate": 5.78481945467944e-11,
720
+ "loss": 0.5054,
721
+ "step": 24000
722
+ },
723
+ {
724
+ "epoch": 4.421518054532056,
725
+ "eval_loss": 1.496930480003357,
726
+ "eval_runtime": 76.1638,
727
+ "eval_samples_per_second": 15.847,
728
+ "eval_steps_per_second": 1.983,
729
+ "step": 24000
730
+ }
731
+ ],
732
+ "logging_steps": 500,
733
+ "max_steps": 27140,
734
+ "num_input_tokens_seen": 0,
735
+ "num_train_epochs": 5,
736
+ "save_steps": 500,
737
+ "stateful_callbacks": {
738
+ "TrainerControl": {
739
+ "args": {
740
+ "should_epoch_stop": false,
741
+ "should_evaluate": false,
742
+ "should_log": false,
743
+ "should_save": true,
744
+ "should_training_stop": false
745
+ },
746
+ "attributes": {}
747
+ }
748
+ },
749
+ "total_flos": 8634919946942520.0,
750
+ "train_batch_size": 2,
751
+ "trial_name": null,
752
+ "trial_params": null
753
+ }
checkpoint-24000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4216477e635a737e95915c1d07b3b9cd2ee3a4f3b4c352bfa7225c18c894aa58
3
+ size 4847