sert121 commited on
Commit
bd0b28f
1 Parent(s): a8b651f

Delete trainer_state.json

Browse files
Files changed (1) hide show
  1. trainer_state.json +0 -316
trainer_state.json DELETED
@@ -1,316 +0,0 @@
1
- {
2
- "best_metric": null,
3
- "best_model_checkpoint": null,
4
- "epoch": 11.428571428571429,
5
- "global_step": 500,
6
- "is_hyper_param_search": false,
7
- "is_local_process_zero": true,
8
- "is_world_process_zero": true,
9
- "log_history": [
10
- {
11
- "epoch": 0.23,
12
- "learning_rate": 0.0002,
13
- "loss": 0.9795,
14
- "step": 10
15
- },
16
- {
17
- "epoch": 0.46,
18
- "learning_rate": 0.0002,
19
- "loss": 0.2923,
20
- "step": 20
21
- },
22
- {
23
- "epoch": 0.69,
24
- "learning_rate": 0.0002,
25
- "loss": 0.1384,
26
- "step": 30
27
- },
28
- {
29
- "epoch": 0.91,
30
- "learning_rate": 0.0002,
31
- "loss": 0.0773,
32
- "step": 40
33
- },
34
- {
35
- "epoch": 1.14,
36
- "learning_rate": 0.0002,
37
- "loss": 0.0578,
38
- "step": 50
39
- },
40
- {
41
- "epoch": 1.37,
42
- "learning_rate": 0.0002,
43
- "loss": 0.0502,
44
- "step": 60
45
- },
46
- {
47
- "epoch": 1.6,
48
- "learning_rate": 0.0002,
49
- "loss": 0.0534,
50
- "step": 70
51
- },
52
- {
53
- "epoch": 1.83,
54
- "learning_rate": 0.0002,
55
- "loss": 0.0466,
56
- "step": 80
57
- },
58
- {
59
- "epoch": 2.06,
60
- "learning_rate": 0.0002,
61
- "loss": 0.0504,
62
- "step": 90
63
- },
64
- {
65
- "epoch": 2.29,
66
- "learning_rate": 0.0002,
67
- "loss": 0.0387,
68
- "step": 100
69
- },
70
- {
71
- "epoch": 2.51,
72
- "learning_rate": 0.0002,
73
- "loss": 0.0402,
74
- "step": 110
75
- },
76
- {
77
- "epoch": 2.74,
78
- "learning_rate": 0.0002,
79
- "loss": 0.0384,
80
- "step": 120
81
- },
82
- {
83
- "epoch": 2.97,
84
- "learning_rate": 0.0002,
85
- "loss": 0.0399,
86
- "step": 130
87
- },
88
- {
89
- "epoch": 3.2,
90
- "learning_rate": 0.0002,
91
- "loss": 0.0366,
92
- "step": 140
93
- },
94
- {
95
- "epoch": 3.43,
96
- "learning_rate": 0.0002,
97
- "loss": 0.0293,
98
- "step": 150
99
- },
100
- {
101
- "epoch": 3.66,
102
- "learning_rate": 0.0002,
103
- "loss": 0.0352,
104
- "step": 160
105
- },
106
- {
107
- "epoch": 3.89,
108
- "learning_rate": 0.0002,
109
- "loss": 0.033,
110
- "step": 170
111
- },
112
- {
113
- "epoch": 4.11,
114
- "learning_rate": 0.0002,
115
- "loss": 0.0333,
116
- "step": 180
117
- },
118
- {
119
- "epoch": 4.34,
120
- "learning_rate": 0.0002,
121
- "loss": 0.0269,
122
- "step": 190
123
- },
124
- {
125
- "epoch": 4.57,
126
- "learning_rate": 0.0002,
127
- "loss": 0.0298,
128
- "step": 200
129
- },
130
- {
131
- "epoch": 4.8,
132
- "learning_rate": 0.0002,
133
- "loss": 0.025,
134
- "step": 210
135
- },
136
- {
137
- "epoch": 5.03,
138
- "learning_rate": 0.0002,
139
- "loss": 0.0253,
140
- "step": 220
141
- },
142
- {
143
- "epoch": 5.26,
144
- "learning_rate": 0.0002,
145
- "loss": 0.0221,
146
- "step": 230
147
- },
148
- {
149
- "epoch": 5.49,
150
- "learning_rate": 0.0002,
151
- "loss": 0.0224,
152
- "step": 240
153
- },
154
- {
155
- "epoch": 5.71,
156
- "learning_rate": 0.0002,
157
- "loss": 0.0202,
158
- "step": 250
159
- },
160
- {
161
- "epoch": 5.94,
162
- "learning_rate": 0.0002,
163
- "loss": 0.0291,
164
- "step": 260
165
- },
166
- {
167
- "epoch": 6.17,
168
- "learning_rate": 0.0002,
169
- "loss": 0.0231,
170
- "step": 270
171
- },
172
- {
173
- "epoch": 6.4,
174
- "learning_rate": 0.0002,
175
- "loss": 0.0196,
176
- "step": 280
177
- },
178
- {
179
- "epoch": 6.63,
180
- "learning_rate": 0.0002,
181
- "loss": 0.0223,
182
- "step": 290
183
- },
184
- {
185
- "epoch": 6.86,
186
- "learning_rate": 0.0002,
187
- "loss": 0.0198,
188
- "step": 300
189
- },
190
- {
191
- "epoch": 7.09,
192
- "learning_rate": 0.0002,
193
- "loss": 0.0191,
194
- "step": 310
195
- },
196
- {
197
- "epoch": 7.31,
198
- "learning_rate": 0.0002,
199
- "loss": 0.0192,
200
- "step": 320
201
- },
202
- {
203
- "epoch": 7.54,
204
- "learning_rate": 0.0002,
205
- "loss": 0.0183,
206
- "step": 330
207
- },
208
- {
209
- "epoch": 7.77,
210
- "learning_rate": 0.0002,
211
- "loss": 0.0207,
212
- "step": 340
213
- },
214
- {
215
- "epoch": 8.0,
216
- "learning_rate": 0.0002,
217
- "loss": 0.02,
218
- "step": 350
219
- },
220
- {
221
- "epoch": 8.23,
222
- "learning_rate": 0.0002,
223
- "loss": 0.018,
224
- "step": 360
225
- },
226
- {
227
- "epoch": 8.46,
228
- "learning_rate": 0.0002,
229
- "loss": 0.0162,
230
- "step": 370
231
- },
232
- {
233
- "epoch": 8.69,
234
- "learning_rate": 0.0002,
235
- "loss": 0.0185,
236
- "step": 380
237
- },
238
- {
239
- "epoch": 8.91,
240
- "learning_rate": 0.0002,
241
- "loss": 0.0194,
242
- "step": 390
243
- },
244
- {
245
- "epoch": 9.14,
246
- "learning_rate": 0.0002,
247
- "loss": 0.015,
248
- "step": 400
249
- },
250
- {
251
- "epoch": 9.37,
252
- "learning_rate": 0.0002,
253
- "loss": 0.0176,
254
- "step": 410
255
- },
256
- {
257
- "epoch": 9.6,
258
- "learning_rate": 0.0002,
259
- "loss": 0.0165,
260
- "step": 420
261
- },
262
- {
263
- "epoch": 9.83,
264
- "learning_rate": 0.0002,
265
- "loss": 0.0163,
266
- "step": 430
267
- },
268
- {
269
- "epoch": 10.06,
270
- "learning_rate": 0.0002,
271
- "loss": 0.0165,
272
- "step": 440
273
- },
274
- {
275
- "epoch": 10.29,
276
- "learning_rate": 0.0002,
277
- "loss": 0.0163,
278
- "step": 450
279
- },
280
- {
281
- "epoch": 10.51,
282
- "learning_rate": 0.0002,
283
- "loss": 0.0164,
284
- "step": 460
285
- },
286
- {
287
- "epoch": 10.74,
288
- "learning_rate": 0.0002,
289
- "loss": 0.0157,
290
- "step": 470
291
- },
292
- {
293
- "epoch": 10.97,
294
- "learning_rate": 0.0002,
295
- "loss": 0.016,
296
- "step": 480
297
- },
298
- {
299
- "epoch": 11.2,
300
- "learning_rate": 0.0002,
301
- "loss": 0.0171,
302
- "step": 490
303
- },
304
- {
305
- "epoch": 11.43,
306
- "learning_rate": 0.0002,
307
- "loss": 0.0148,
308
- "step": 500
309
- }
310
- ],
311
- "max_steps": 500,
312
- "num_train_epochs": 12,
313
- "total_flos": 8.4635406041088e+16,
314
- "trial_name": null,
315
- "trial_params": null
316
- }