irenelizihui commited on
Commit
a89e2e8
1 Parent(s): 1ea5a62

Upload trainer_state.json

Browse files
Files changed (1) hide show
  1. trainer_state.json +2032 -0
trainer_state.json ADDED
@@ -0,0 +1,2032 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 0.9995835068721366,
5
+ "global_step": 168000,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.0,
12
+ "learning_rate": 1.9940500981733805e-05,
13
+ "loss": 1.1973,
14
+ "step": 500
15
+ },
16
+ {
17
+ "epoch": 0.01,
18
+ "learning_rate": 1.9881001963467605e-05,
19
+ "loss": 1.2168,
20
+ "step": 1000
21
+ },
22
+ {
23
+ "epoch": 0.01,
24
+ "learning_rate": 1.9821502945201405e-05,
25
+ "loss": 1.1454,
26
+ "step": 1500
27
+ },
28
+ {
29
+ "epoch": 0.01,
30
+ "learning_rate": 1.9762003926935208e-05,
31
+ "loss": 1.1647,
32
+ "step": 2000
33
+ },
34
+ {
35
+ "epoch": 0.01,
36
+ "learning_rate": 1.9702504908669008e-05,
37
+ "loss": 1.1795,
38
+ "step": 2500
39
+ },
40
+ {
41
+ "epoch": 0.02,
42
+ "learning_rate": 1.964300589040281e-05,
43
+ "loss": 1.1723,
44
+ "step": 3000
45
+ },
46
+ {
47
+ "epoch": 0.02,
48
+ "learning_rate": 1.958350687213661e-05,
49
+ "loss": 1.157,
50
+ "step": 3500
51
+ },
52
+ {
53
+ "epoch": 0.02,
54
+ "learning_rate": 1.952400785387041e-05,
55
+ "loss": 1.1699,
56
+ "step": 4000
57
+ },
58
+ {
59
+ "epoch": 0.03,
60
+ "learning_rate": 1.9464508835604215e-05,
61
+ "loss": 1.1593,
62
+ "step": 4500
63
+ },
64
+ {
65
+ "epoch": 0.03,
66
+ "learning_rate": 1.9405009817338018e-05,
67
+ "loss": 1.1421,
68
+ "step": 5000
69
+ },
70
+ {
71
+ "epoch": 0.03,
72
+ "learning_rate": 1.9345510799071814e-05,
73
+ "loss": 1.0648,
74
+ "step": 5500
75
+ },
76
+ {
77
+ "epoch": 0.04,
78
+ "learning_rate": 1.9286011780805618e-05,
79
+ "loss": 1.1345,
80
+ "step": 6000
81
+ },
82
+ {
83
+ "epoch": 0.04,
84
+ "learning_rate": 1.922651276253942e-05,
85
+ "loss": 1.1722,
86
+ "step": 6500
87
+ },
88
+ {
89
+ "epoch": 0.04,
90
+ "learning_rate": 1.916701374427322e-05,
91
+ "loss": 1.1588,
92
+ "step": 7000
93
+ },
94
+ {
95
+ "epoch": 0.04,
96
+ "learning_rate": 1.910751472600702e-05,
97
+ "loss": 1.1502,
98
+ "step": 7500
99
+ },
100
+ {
101
+ "epoch": 0.05,
102
+ "learning_rate": 1.9048015707740824e-05,
103
+ "loss": 1.1297,
104
+ "step": 8000
105
+ },
106
+ {
107
+ "epoch": 0.05,
108
+ "learning_rate": 1.8988516689474624e-05,
109
+ "loss": 1.1478,
110
+ "step": 8500
111
+ },
112
+ {
113
+ "epoch": 0.05,
114
+ "learning_rate": 1.8929017671208427e-05,
115
+ "loss": 1.1025,
116
+ "step": 9000
117
+ },
118
+ {
119
+ "epoch": 0.06,
120
+ "learning_rate": 1.8869518652942227e-05,
121
+ "loss": 1.126,
122
+ "step": 9500
123
+ },
124
+ {
125
+ "epoch": 0.06,
126
+ "learning_rate": 1.8810019634676027e-05,
127
+ "loss": 1.1438,
128
+ "step": 10000
129
+ },
130
+ {
131
+ "epoch": 0.06,
132
+ "learning_rate": 1.875052061640983e-05,
133
+ "loss": 1.1064,
134
+ "step": 10500
135
+ },
136
+ {
137
+ "epoch": 0.07,
138
+ "learning_rate": 1.8691021598143634e-05,
139
+ "loss": 1.1369,
140
+ "step": 11000
141
+ },
142
+ {
143
+ "epoch": 0.07,
144
+ "learning_rate": 1.8631522579877434e-05,
145
+ "loss": 1.1036,
146
+ "step": 11500
147
+ },
148
+ {
149
+ "epoch": 0.07,
150
+ "learning_rate": 1.8572023561611234e-05,
151
+ "loss": 1.0771,
152
+ "step": 12000
153
+ },
154
+ {
155
+ "epoch": 0.07,
156
+ "learning_rate": 1.8512524543345037e-05,
157
+ "loss": 1.0841,
158
+ "step": 12500
159
+ },
160
+ {
161
+ "epoch": 0.08,
162
+ "learning_rate": 1.8453025525078837e-05,
163
+ "loss": 1.0818,
164
+ "step": 13000
165
+ },
166
+ {
167
+ "epoch": 0.08,
168
+ "learning_rate": 1.839352650681264e-05,
169
+ "loss": 1.1462,
170
+ "step": 13500
171
+ },
172
+ {
173
+ "epoch": 0.08,
174
+ "learning_rate": 1.833402748854644e-05,
175
+ "loss": 1.1335,
176
+ "step": 14000
177
+ },
178
+ {
179
+ "epoch": 0.09,
180
+ "learning_rate": 1.827452847028024e-05,
181
+ "loss": 1.079,
182
+ "step": 14500
183
+ },
184
+ {
185
+ "epoch": 0.09,
186
+ "learning_rate": 1.8215029452014044e-05,
187
+ "loss": 1.1276,
188
+ "step": 15000
189
+ },
190
+ {
191
+ "epoch": 0.09,
192
+ "learning_rate": 1.8155530433747847e-05,
193
+ "loss": 1.1119,
194
+ "step": 15500
195
+ },
196
+ {
197
+ "epoch": 0.1,
198
+ "learning_rate": 1.8096031415481647e-05,
199
+ "loss": 1.1343,
200
+ "step": 16000
201
+ },
202
+ {
203
+ "epoch": 0.1,
204
+ "learning_rate": 1.8036532397215447e-05,
205
+ "loss": 1.0787,
206
+ "step": 16500
207
+ },
208
+ {
209
+ "epoch": 0.1,
210
+ "learning_rate": 1.797703337894925e-05,
211
+ "loss": 1.1122,
212
+ "step": 17000
213
+ },
214
+ {
215
+ "epoch": 0.1,
216
+ "learning_rate": 1.791753436068305e-05,
217
+ "loss": 1.0904,
218
+ "step": 17500
219
+ },
220
+ {
221
+ "epoch": 0.11,
222
+ "learning_rate": 1.785803534241685e-05,
223
+ "loss": 1.0882,
224
+ "step": 18000
225
+ },
226
+ {
227
+ "epoch": 0.11,
228
+ "learning_rate": 1.7798536324150653e-05,
229
+ "loss": 1.1311,
230
+ "step": 18500
231
+ },
232
+ {
233
+ "epoch": 0.11,
234
+ "learning_rate": 1.7739037305884453e-05,
235
+ "loss": 1.0936,
236
+ "step": 19000
237
+ },
238
+ {
239
+ "epoch": 0.12,
240
+ "learning_rate": 1.7679538287618257e-05,
241
+ "loss": 1.0959,
242
+ "step": 19500
243
+ },
244
+ {
245
+ "epoch": 0.12,
246
+ "learning_rate": 1.7620039269352056e-05,
247
+ "loss": 1.0879,
248
+ "step": 20000
249
+ },
250
+ {
251
+ "epoch": 0.12,
252
+ "learning_rate": 1.7560540251085856e-05,
253
+ "loss": 1.0763,
254
+ "step": 20500
255
+ },
256
+ {
257
+ "epoch": 0.12,
258
+ "learning_rate": 1.750104123281966e-05,
259
+ "loss": 1.0713,
260
+ "step": 21000
261
+ },
262
+ {
263
+ "epoch": 0.13,
264
+ "learning_rate": 1.7441542214553463e-05,
265
+ "loss": 1.1204,
266
+ "step": 21500
267
+ },
268
+ {
269
+ "epoch": 0.13,
270
+ "learning_rate": 1.7382043196287263e-05,
271
+ "loss": 1.0808,
272
+ "step": 22000
273
+ },
274
+ {
275
+ "epoch": 0.13,
276
+ "learning_rate": 1.7322544178021063e-05,
277
+ "loss": 1.1301,
278
+ "step": 22500
279
+ },
280
+ {
281
+ "epoch": 0.14,
282
+ "learning_rate": 1.7263045159754866e-05,
283
+ "loss": 1.1141,
284
+ "step": 23000
285
+ },
286
+ {
287
+ "epoch": 0.14,
288
+ "learning_rate": 1.7203546141488666e-05,
289
+ "loss": 1.109,
290
+ "step": 23500
291
+ },
292
+ {
293
+ "epoch": 0.14,
294
+ "learning_rate": 1.714404712322247e-05,
295
+ "loss": 1.0691,
296
+ "step": 24000
297
+ },
298
+ {
299
+ "epoch": 0.15,
300
+ "learning_rate": 1.708454810495627e-05,
301
+ "loss": 1.0805,
302
+ "step": 24500
303
+ },
304
+ {
305
+ "epoch": 0.15,
306
+ "learning_rate": 1.702504908669007e-05,
307
+ "loss": 1.0795,
308
+ "step": 25000
309
+ },
310
+ {
311
+ "epoch": 0.15,
312
+ "learning_rate": 1.6965550068423873e-05,
313
+ "loss": 1.0431,
314
+ "step": 25500
315
+ },
316
+ {
317
+ "epoch": 0.15,
318
+ "learning_rate": 1.6906051050157676e-05,
319
+ "loss": 1.0553,
320
+ "step": 26000
321
+ },
322
+ {
323
+ "epoch": 0.16,
324
+ "learning_rate": 1.6846552031891476e-05,
325
+ "loss": 1.0786,
326
+ "step": 26500
327
+ },
328
+ {
329
+ "epoch": 0.16,
330
+ "learning_rate": 1.6787053013625276e-05,
331
+ "loss": 1.1008,
332
+ "step": 27000
333
+ },
334
+ {
335
+ "epoch": 0.16,
336
+ "learning_rate": 1.672755399535908e-05,
337
+ "loss": 1.0776,
338
+ "step": 27500
339
+ },
340
+ {
341
+ "epoch": 0.17,
342
+ "learning_rate": 1.666805497709288e-05,
343
+ "loss": 1.0993,
344
+ "step": 28000
345
+ },
346
+ {
347
+ "epoch": 0.17,
348
+ "learning_rate": 1.660855595882668e-05,
349
+ "loss": 1.1101,
350
+ "step": 28500
351
+ },
352
+ {
353
+ "epoch": 0.17,
354
+ "learning_rate": 1.6549056940560482e-05,
355
+ "loss": 1.0791,
356
+ "step": 29000
357
+ },
358
+ {
359
+ "epoch": 0.18,
360
+ "learning_rate": 1.6489557922294282e-05,
361
+ "loss": 1.0705,
362
+ "step": 29500
363
+ },
364
+ {
365
+ "epoch": 0.18,
366
+ "learning_rate": 1.6430058904028086e-05,
367
+ "loss": 1.0618,
368
+ "step": 30000
369
+ },
370
+ {
371
+ "epoch": 0.18,
372
+ "learning_rate": 1.6370559885761885e-05,
373
+ "loss": 1.0408,
374
+ "step": 30500
375
+ },
376
+ {
377
+ "epoch": 0.18,
378
+ "learning_rate": 1.631106086749569e-05,
379
+ "loss": 1.1367,
380
+ "step": 31000
381
+ },
382
+ {
383
+ "epoch": 0.19,
384
+ "learning_rate": 1.625156184922949e-05,
385
+ "loss": 1.0727,
386
+ "step": 31500
387
+ },
388
+ {
389
+ "epoch": 0.19,
390
+ "learning_rate": 1.6192062830963292e-05,
391
+ "loss": 1.0444,
392
+ "step": 32000
393
+ },
394
+ {
395
+ "epoch": 0.19,
396
+ "learning_rate": 1.6132563812697092e-05,
397
+ "loss": 1.0723,
398
+ "step": 32500
399
+ },
400
+ {
401
+ "epoch": 0.2,
402
+ "learning_rate": 1.6073064794430892e-05,
403
+ "loss": 1.0627,
404
+ "step": 33000
405
+ },
406
+ {
407
+ "epoch": 0.2,
408
+ "learning_rate": 1.6013565776164695e-05,
409
+ "loss": 1.0835,
410
+ "step": 33500
411
+ },
412
+ {
413
+ "epoch": 0.2,
414
+ "learning_rate": 1.5954066757898495e-05,
415
+ "loss": 1.1624,
416
+ "step": 34000
417
+ },
418
+ {
419
+ "epoch": 0.21,
420
+ "learning_rate": 1.58945677396323e-05,
421
+ "loss": 1.0383,
422
+ "step": 34500
423
+ },
424
+ {
425
+ "epoch": 0.21,
426
+ "learning_rate": 1.58350687213661e-05,
427
+ "loss": 1.0805,
428
+ "step": 35000
429
+ },
430
+ {
431
+ "epoch": 0.21,
432
+ "learning_rate": 1.57755697030999e-05,
433
+ "loss": 1.0426,
434
+ "step": 35500
435
+ },
436
+ {
437
+ "epoch": 0.21,
438
+ "learning_rate": 1.57160706848337e-05,
439
+ "loss": 1.1281,
440
+ "step": 36000
441
+ },
442
+ {
443
+ "epoch": 0.22,
444
+ "learning_rate": 1.5656571666567505e-05,
445
+ "loss": 1.0695,
446
+ "step": 36500
447
+ },
448
+ {
449
+ "epoch": 0.22,
450
+ "learning_rate": 1.5597072648301305e-05,
451
+ "loss": 1.0617,
452
+ "step": 37000
453
+ },
454
+ {
455
+ "epoch": 0.22,
456
+ "learning_rate": 1.5537573630035105e-05,
457
+ "loss": 1.0849,
458
+ "step": 37500
459
+ },
460
+ {
461
+ "epoch": 0.23,
462
+ "learning_rate": 1.5478074611768908e-05,
463
+ "loss": 1.0923,
464
+ "step": 38000
465
+ },
466
+ {
467
+ "epoch": 0.23,
468
+ "learning_rate": 1.5418575593502708e-05,
469
+ "loss": 1.1421,
470
+ "step": 38500
471
+ },
472
+ {
473
+ "epoch": 0.23,
474
+ "learning_rate": 1.5359076575236508e-05,
475
+ "loss": 1.0981,
476
+ "step": 39000
477
+ },
478
+ {
479
+ "epoch": 0.24,
480
+ "learning_rate": 1.529957755697031e-05,
481
+ "loss": 1.0991,
482
+ "step": 39500
483
+ },
484
+ {
485
+ "epoch": 0.24,
486
+ "learning_rate": 1.5240078538704113e-05,
487
+ "loss": 1.0568,
488
+ "step": 40000
489
+ },
490
+ {
491
+ "epoch": 0.24,
492
+ "learning_rate": 1.5180579520437915e-05,
493
+ "loss": 1.1249,
494
+ "step": 40500
495
+ },
496
+ {
497
+ "epoch": 0.24,
498
+ "learning_rate": 1.5121080502171714e-05,
499
+ "loss": 1.0764,
500
+ "step": 41000
501
+ },
502
+ {
503
+ "epoch": 0.25,
504
+ "learning_rate": 1.5061581483905516e-05,
505
+ "loss": 1.0582,
506
+ "step": 41500
507
+ },
508
+ {
509
+ "epoch": 0.25,
510
+ "learning_rate": 1.5002082465639318e-05,
511
+ "loss": 1.0795,
512
+ "step": 42000
513
+ },
514
+ {
515
+ "epoch": 0.25,
516
+ "learning_rate": 1.494258344737312e-05,
517
+ "loss": 1.0514,
518
+ "step": 42500
519
+ },
520
+ {
521
+ "epoch": 0.26,
522
+ "learning_rate": 1.488308442910692e-05,
523
+ "loss": 1.0427,
524
+ "step": 43000
525
+ },
526
+ {
527
+ "epoch": 0.26,
528
+ "learning_rate": 1.4823585410840723e-05,
529
+ "loss": 1.0842,
530
+ "step": 43500
531
+ },
532
+ {
533
+ "epoch": 0.26,
534
+ "learning_rate": 1.4764086392574524e-05,
535
+ "loss": 1.0961,
536
+ "step": 44000
537
+ },
538
+ {
539
+ "epoch": 0.26,
540
+ "learning_rate": 1.4704587374308326e-05,
541
+ "loss": 1.1226,
542
+ "step": 44500
543
+ },
544
+ {
545
+ "epoch": 0.27,
546
+ "learning_rate": 1.4645088356042127e-05,
547
+ "loss": 1.0607,
548
+ "step": 45000
549
+ },
550
+ {
551
+ "epoch": 0.27,
552
+ "learning_rate": 1.4585589337775927e-05,
553
+ "loss": 1.0479,
554
+ "step": 45500
555
+ },
556
+ {
557
+ "epoch": 0.27,
558
+ "learning_rate": 1.4526090319509729e-05,
559
+ "loss": 1.0572,
560
+ "step": 46000
561
+ },
562
+ {
563
+ "epoch": 0.28,
564
+ "learning_rate": 1.446659130124353e-05,
565
+ "loss": 1.1465,
566
+ "step": 46500
567
+ },
568
+ {
569
+ "epoch": 0.28,
570
+ "learning_rate": 1.4407092282977332e-05,
571
+ "loss": 1.041,
572
+ "step": 47000
573
+ },
574
+ {
575
+ "epoch": 0.28,
576
+ "learning_rate": 1.4347593264711132e-05,
577
+ "loss": 1.0366,
578
+ "step": 47500
579
+ },
580
+ {
581
+ "epoch": 0.29,
582
+ "learning_rate": 1.4288094246444934e-05,
583
+ "loss": 1.0661,
584
+ "step": 48000
585
+ },
586
+ {
587
+ "epoch": 0.29,
588
+ "learning_rate": 1.4228595228178737e-05,
589
+ "loss": 1.0495,
590
+ "step": 48500
591
+ },
592
+ {
593
+ "epoch": 0.29,
594
+ "learning_rate": 1.4169096209912539e-05,
595
+ "loss": 1.0769,
596
+ "step": 49000
597
+ },
598
+ {
599
+ "epoch": 0.29,
600
+ "learning_rate": 1.4109597191646339e-05,
601
+ "loss": 1.0329,
602
+ "step": 49500
603
+ },
604
+ {
605
+ "epoch": 0.3,
606
+ "learning_rate": 1.405009817338014e-05,
607
+ "loss": 1.0702,
608
+ "step": 50000
609
+ },
610
+ {
611
+ "epoch": 0.3,
612
+ "learning_rate": 1.3990599155113942e-05,
613
+ "loss": 1.0354,
614
+ "step": 50500
615
+ },
616
+ {
617
+ "epoch": 0.3,
618
+ "learning_rate": 1.3931100136847744e-05,
619
+ "loss": 1.0817,
620
+ "step": 51000
621
+ },
622
+ {
623
+ "epoch": 0.31,
624
+ "learning_rate": 1.3871601118581544e-05,
625
+ "loss": 1.0362,
626
+ "step": 51500
627
+ },
628
+ {
629
+ "epoch": 0.31,
630
+ "learning_rate": 1.3812102100315345e-05,
631
+ "loss": 1.0761,
632
+ "step": 52000
633
+ },
634
+ {
635
+ "epoch": 0.31,
636
+ "learning_rate": 1.3752603082049147e-05,
637
+ "loss": 1.0295,
638
+ "step": 52500
639
+ },
640
+ {
641
+ "epoch": 0.32,
642
+ "learning_rate": 1.369310406378295e-05,
643
+ "loss": 1.0365,
644
+ "step": 53000
645
+ },
646
+ {
647
+ "epoch": 0.32,
648
+ "learning_rate": 1.3633605045516748e-05,
649
+ "loss": 1.0378,
650
+ "step": 53500
651
+ },
652
+ {
653
+ "epoch": 0.32,
654
+ "learning_rate": 1.3574106027250552e-05,
655
+ "loss": 1.0172,
656
+ "step": 54000
657
+ },
658
+ {
659
+ "epoch": 0.32,
660
+ "learning_rate": 1.3514607008984353e-05,
661
+ "loss": 1.0553,
662
+ "step": 54500
663
+ },
664
+ {
665
+ "epoch": 0.33,
666
+ "learning_rate": 1.3455107990718155e-05,
667
+ "loss": 1.0595,
668
+ "step": 55000
669
+ },
670
+ {
671
+ "epoch": 0.33,
672
+ "learning_rate": 1.3395608972451957e-05,
673
+ "loss": 1.0443,
674
+ "step": 55500
675
+ },
676
+ {
677
+ "epoch": 0.33,
678
+ "learning_rate": 1.3336109954185756e-05,
679
+ "loss": 1.0359,
680
+ "step": 56000
681
+ },
682
+ {
683
+ "epoch": 0.34,
684
+ "learning_rate": 1.3276610935919558e-05,
685
+ "loss": 1.0631,
686
+ "step": 56500
687
+ },
688
+ {
689
+ "epoch": 0.34,
690
+ "learning_rate": 1.321711191765336e-05,
691
+ "loss": 1.0293,
692
+ "step": 57000
693
+ },
694
+ {
695
+ "epoch": 0.34,
696
+ "learning_rate": 1.3157612899387161e-05,
697
+ "loss": 1.0878,
698
+ "step": 57500
699
+ },
700
+ {
701
+ "epoch": 0.35,
702
+ "learning_rate": 1.3098113881120961e-05,
703
+ "loss": 1.054,
704
+ "step": 58000
705
+ },
706
+ {
707
+ "epoch": 0.35,
708
+ "learning_rate": 1.3038614862854765e-05,
709
+ "loss": 1.0523,
710
+ "step": 58500
711
+ },
712
+ {
713
+ "epoch": 0.35,
714
+ "learning_rate": 1.2979115844588566e-05,
715
+ "loss": 1.0117,
716
+ "step": 59000
717
+ },
718
+ {
719
+ "epoch": 0.35,
720
+ "learning_rate": 1.2919616826322368e-05,
721
+ "loss": 1.0612,
722
+ "step": 59500
723
+ },
724
+ {
725
+ "epoch": 0.36,
726
+ "learning_rate": 1.2860117808056168e-05,
727
+ "loss": 1.0106,
728
+ "step": 60000
729
+ },
730
+ {
731
+ "epoch": 0.36,
732
+ "learning_rate": 1.280061878978997e-05,
733
+ "loss": 1.026,
734
+ "step": 60500
735
+ },
736
+ {
737
+ "epoch": 0.36,
738
+ "learning_rate": 1.2741119771523771e-05,
739
+ "loss": 1.0533,
740
+ "step": 61000
741
+ },
742
+ {
743
+ "epoch": 0.37,
744
+ "learning_rate": 1.2681620753257573e-05,
745
+ "loss": 1.0651,
746
+ "step": 61500
747
+ },
748
+ {
749
+ "epoch": 0.37,
750
+ "learning_rate": 1.2622121734991373e-05,
751
+ "loss": 1.0717,
752
+ "step": 62000
753
+ },
754
+ {
755
+ "epoch": 0.37,
756
+ "learning_rate": 1.2562622716725174e-05,
757
+ "loss": 1.0679,
758
+ "step": 62500
759
+ },
760
+ {
761
+ "epoch": 0.37,
762
+ "learning_rate": 1.2503123698458976e-05,
763
+ "loss": 1.0594,
764
+ "step": 63000
765
+ },
766
+ {
767
+ "epoch": 0.38,
768
+ "learning_rate": 1.2443624680192779e-05,
769
+ "loss": 1.0457,
770
+ "step": 63500
771
+ },
772
+ {
773
+ "epoch": 0.38,
774
+ "learning_rate": 1.2384125661926577e-05,
775
+ "loss": 1.0491,
776
+ "step": 64000
777
+ },
778
+ {
779
+ "epoch": 0.38,
780
+ "learning_rate": 1.232462664366038e-05,
781
+ "loss": 1.0212,
782
+ "step": 64500
783
+ },
784
+ {
785
+ "epoch": 0.39,
786
+ "learning_rate": 1.2265127625394182e-05,
787
+ "loss": 1.0079,
788
+ "step": 65000
789
+ },
790
+ {
791
+ "epoch": 0.39,
792
+ "learning_rate": 1.2205628607127984e-05,
793
+ "loss": 1.0513,
794
+ "step": 65500
795
+ },
796
+ {
797
+ "epoch": 0.39,
798
+ "learning_rate": 1.2146129588861786e-05,
799
+ "loss": 1.0651,
800
+ "step": 66000
801
+ },
802
+ {
803
+ "epoch": 0.4,
804
+ "learning_rate": 1.2086630570595585e-05,
805
+ "loss": 1.1196,
806
+ "step": 66500
807
+ },
808
+ {
809
+ "epoch": 0.4,
810
+ "learning_rate": 1.2027131552329387e-05,
811
+ "loss": 1.0522,
812
+ "step": 67000
813
+ },
814
+ {
815
+ "epoch": 0.4,
816
+ "learning_rate": 1.1967632534063189e-05,
817
+ "loss": 1.0775,
818
+ "step": 67500
819
+ },
820
+ {
821
+ "epoch": 0.4,
822
+ "learning_rate": 1.1908133515796992e-05,
823
+ "loss": 1.0494,
824
+ "step": 68000
825
+ },
826
+ {
827
+ "epoch": 0.41,
828
+ "learning_rate": 1.184863449753079e-05,
829
+ "loss": 1.0312,
830
+ "step": 68500
831
+ },
832
+ {
833
+ "epoch": 0.41,
834
+ "learning_rate": 1.1789135479264594e-05,
835
+ "loss": 0.9866,
836
+ "step": 69000
837
+ },
838
+ {
839
+ "epoch": 0.41,
840
+ "learning_rate": 1.1729636460998395e-05,
841
+ "loss": 1.0072,
842
+ "step": 69500
843
+ },
844
+ {
845
+ "epoch": 0.42,
846
+ "learning_rate": 1.1670137442732197e-05,
847
+ "loss": 1.0145,
848
+ "step": 70000
849
+ },
850
+ {
851
+ "epoch": 0.42,
852
+ "learning_rate": 1.1610638424465997e-05,
853
+ "loss": 1.0334,
854
+ "step": 70500
855
+ },
856
+ {
857
+ "epoch": 0.42,
858
+ "learning_rate": 1.1551139406199798e-05,
859
+ "loss": 1.1029,
860
+ "step": 71000
861
+ },
862
+ {
863
+ "epoch": 0.43,
864
+ "learning_rate": 1.14916403879336e-05,
865
+ "loss": 1.0556,
866
+ "step": 71500
867
+ },
868
+ {
869
+ "epoch": 0.43,
870
+ "learning_rate": 1.1432141369667402e-05,
871
+ "loss": 1.0046,
872
+ "step": 72000
873
+ },
874
+ {
875
+ "epoch": 0.43,
876
+ "learning_rate": 1.1372642351401202e-05,
877
+ "loss": 1.0827,
878
+ "step": 72500
879
+ },
880
+ {
881
+ "epoch": 0.43,
882
+ "learning_rate": 1.1313143333135003e-05,
883
+ "loss": 1.0392,
884
+ "step": 73000
885
+ },
886
+ {
887
+ "epoch": 0.44,
888
+ "learning_rate": 1.1253644314868807e-05,
889
+ "loss": 1.0635,
890
+ "step": 73500
891
+ },
892
+ {
893
+ "epoch": 0.44,
894
+ "learning_rate": 1.1194145296602608e-05,
895
+ "loss": 1.0246,
896
+ "step": 74000
897
+ },
898
+ {
899
+ "epoch": 0.44,
900
+ "learning_rate": 1.1134646278336408e-05,
901
+ "loss": 1.0271,
902
+ "step": 74500
903
+ },
904
+ {
905
+ "epoch": 0.45,
906
+ "learning_rate": 1.107514726007021e-05,
907
+ "loss": 0.9673,
908
+ "step": 75000
909
+ },
910
+ {
911
+ "epoch": 0.45,
912
+ "learning_rate": 1.1015648241804011e-05,
913
+ "loss": 1.0277,
914
+ "step": 75500
915
+ },
916
+ {
917
+ "epoch": 0.45,
918
+ "learning_rate": 1.0956149223537813e-05,
919
+ "loss": 1.0755,
920
+ "step": 76000
921
+ },
922
+ {
923
+ "epoch": 0.46,
924
+ "learning_rate": 1.0896650205271615e-05,
925
+ "loss": 1.017,
926
+ "step": 76500
927
+ },
928
+ {
929
+ "epoch": 0.46,
930
+ "learning_rate": 1.0837151187005414e-05,
931
+ "loss": 1.012,
932
+ "step": 77000
933
+ },
934
+ {
935
+ "epoch": 0.46,
936
+ "learning_rate": 1.0777652168739216e-05,
937
+ "loss": 1.0078,
938
+ "step": 77500
939
+ },
940
+ {
941
+ "epoch": 0.46,
942
+ "learning_rate": 1.0718153150473018e-05,
943
+ "loss": 1.0136,
944
+ "step": 78000
945
+ },
946
+ {
947
+ "epoch": 0.47,
948
+ "learning_rate": 1.0658654132206821e-05,
949
+ "loss": 1.0571,
950
+ "step": 78500
951
+ },
952
+ {
953
+ "epoch": 0.47,
954
+ "learning_rate": 1.059915511394062e-05,
955
+ "loss": 1.0558,
956
+ "step": 79000
957
+ },
958
+ {
959
+ "epoch": 0.47,
960
+ "learning_rate": 1.0539656095674423e-05,
961
+ "loss": 0.9997,
962
+ "step": 79500
963
+ },
964
+ {
965
+ "epoch": 0.48,
966
+ "learning_rate": 1.0480157077408224e-05,
967
+ "loss": 1.0411,
968
+ "step": 80000
969
+ },
970
+ {
971
+ "epoch": 0.48,
972
+ "learning_rate": 1.0420658059142026e-05,
973
+ "loss": 1.0569,
974
+ "step": 80500
975
+ },
976
+ {
977
+ "epoch": 0.48,
978
+ "learning_rate": 1.0361159040875826e-05,
979
+ "loss": 1.0182,
980
+ "step": 81000
981
+ },
982
+ {
983
+ "epoch": 0.48,
984
+ "learning_rate": 1.0301660022609627e-05,
985
+ "loss": 1.028,
986
+ "step": 81500
987
+ },
988
+ {
989
+ "epoch": 0.49,
990
+ "learning_rate": 1.0242161004343429e-05,
991
+ "loss": 1.0548,
992
+ "step": 82000
993
+ },
994
+ {
995
+ "epoch": 0.49,
996
+ "learning_rate": 1.018266198607723e-05,
997
+ "loss": 1.0717,
998
+ "step": 82500
999
+ },
1000
+ {
1001
+ "epoch": 0.49,
1002
+ "learning_rate": 1.012316296781103e-05,
1003
+ "loss": 1.0518,
1004
+ "step": 83000
1005
+ },
1006
+ {
1007
+ "epoch": 0.5,
1008
+ "learning_rate": 1.0063663949544832e-05,
1009
+ "loss": 1.025,
1010
+ "step": 83500
1011
+ },
1012
+ {
1013
+ "epoch": 0.5,
1014
+ "learning_rate": 1.0004164931278636e-05,
1015
+ "loss": 1.0131,
1016
+ "step": 84000
1017
+ },
1018
+ {
1019
+ "epoch": 0.5,
1020
+ "learning_rate": 9.944665913012435e-06,
1021
+ "loss": 1.0403,
1022
+ "step": 84500
1023
+ },
1024
+ {
1025
+ "epoch": 0.51,
1026
+ "learning_rate": 9.885166894746237e-06,
1027
+ "loss": 0.9678,
1028
+ "step": 85000
1029
+ },
1030
+ {
1031
+ "epoch": 0.51,
1032
+ "learning_rate": 9.825667876480039e-06,
1033
+ "loss": 1.0123,
1034
+ "step": 85500
1035
+ },
1036
+ {
1037
+ "epoch": 0.51,
1038
+ "learning_rate": 9.76616885821384e-06,
1039
+ "loss": 1.0033,
1040
+ "step": 86000
1041
+ },
1042
+ {
1043
+ "epoch": 0.51,
1044
+ "learning_rate": 9.706669839947642e-06,
1045
+ "loss": 1.0143,
1046
+ "step": 86500
1047
+ },
1048
+ {
1049
+ "epoch": 0.52,
1050
+ "learning_rate": 9.647170821681444e-06,
1051
+ "loss": 1.0296,
1052
+ "step": 87000
1053
+ },
1054
+ {
1055
+ "epoch": 0.52,
1056
+ "learning_rate": 9.587671803415245e-06,
1057
+ "loss": 1.0152,
1058
+ "step": 87500
1059
+ },
1060
+ {
1061
+ "epoch": 0.52,
1062
+ "learning_rate": 9.528172785149045e-06,
1063
+ "loss": 1.0504,
1064
+ "step": 88000
1065
+ },
1066
+ {
1067
+ "epoch": 0.53,
1068
+ "learning_rate": 9.468673766882848e-06,
1069
+ "loss": 1.0308,
1070
+ "step": 88500
1071
+ },
1072
+ {
1073
+ "epoch": 0.53,
1074
+ "learning_rate": 9.409174748616648e-06,
1075
+ "loss": 1.0092,
1076
+ "step": 89000
1077
+ },
1078
+ {
1079
+ "epoch": 0.53,
1080
+ "learning_rate": 9.34967573035045e-06,
1081
+ "loss": 1.0669,
1082
+ "step": 89500
1083
+ },
1084
+ {
1085
+ "epoch": 0.54,
1086
+ "learning_rate": 9.290176712084252e-06,
1087
+ "loss": 1.0227,
1088
+ "step": 90000
1089
+ },
1090
+ {
1091
+ "epoch": 0.54,
1092
+ "learning_rate": 9.230677693818053e-06,
1093
+ "loss": 1.0142,
1094
+ "step": 90500
1095
+ },
1096
+ {
1097
+ "epoch": 0.54,
1098
+ "learning_rate": 9.171178675551853e-06,
1099
+ "loss": 1.073,
1100
+ "step": 91000
1101
+ },
1102
+ {
1103
+ "epoch": 0.54,
1104
+ "learning_rate": 9.111679657285657e-06,
1105
+ "loss": 1.0564,
1106
+ "step": 91500
1107
+ },
1108
+ {
1109
+ "epoch": 0.55,
1110
+ "learning_rate": 9.052180639019456e-06,
1111
+ "loss": 1.0278,
1112
+ "step": 92000
1113
+ },
1114
+ {
1115
+ "epoch": 0.55,
1116
+ "learning_rate": 8.992681620753258e-06,
1117
+ "loss": 1.0213,
1118
+ "step": 92500
1119
+ },
1120
+ {
1121
+ "epoch": 0.55,
1122
+ "learning_rate": 8.93318260248706e-06,
1123
+ "loss": 1.0354,
1124
+ "step": 93000
1125
+ },
1126
+ {
1127
+ "epoch": 0.56,
1128
+ "learning_rate": 8.873683584220861e-06,
1129
+ "loss": 1.0315,
1130
+ "step": 93500
1131
+ },
1132
+ {
1133
+ "epoch": 0.56,
1134
+ "learning_rate": 8.814184565954663e-06,
1135
+ "loss": 1.0585,
1136
+ "step": 94000
1137
+ },
1138
+ {
1139
+ "epoch": 0.56,
1140
+ "learning_rate": 8.754685547688465e-06,
1141
+ "loss": 1.0345,
1142
+ "step": 94500
1143
+ },
1144
+ {
1145
+ "epoch": 0.57,
1146
+ "learning_rate": 8.695186529422264e-06,
1147
+ "loss": 1.0612,
1148
+ "step": 95000
1149
+ },
1150
+ {
1151
+ "epoch": 0.57,
1152
+ "learning_rate": 8.635687511156066e-06,
1153
+ "loss": 1.0734,
1154
+ "step": 95500
1155
+ },
1156
+ {
1157
+ "epoch": 0.57,
1158
+ "learning_rate": 8.576188492889868e-06,
1159
+ "loss": 1.0822,
1160
+ "step": 96000
1161
+ },
1162
+ {
1163
+ "epoch": 0.57,
1164
+ "learning_rate": 8.51668947462367e-06,
1165
+ "loss": 1.0095,
1166
+ "step": 96500
1167
+ },
1168
+ {
1169
+ "epoch": 0.58,
1170
+ "learning_rate": 8.457190456357471e-06,
1171
+ "loss": 1.0104,
1172
+ "step": 97000
1173
+ },
1174
+ {
1175
+ "epoch": 0.58,
1176
+ "learning_rate": 8.397691438091273e-06,
1177
+ "loss": 1.0637,
1178
+ "step": 97500
1179
+ },
1180
+ {
1181
+ "epoch": 0.58,
1182
+ "learning_rate": 8.338192419825074e-06,
1183
+ "loss": 1.0427,
1184
+ "step": 98000
1185
+ },
1186
+ {
1187
+ "epoch": 0.59,
1188
+ "learning_rate": 8.278693401558874e-06,
1189
+ "loss": 1.0073,
1190
+ "step": 98500
1191
+ },
1192
+ {
1193
+ "epoch": 0.59,
1194
+ "learning_rate": 8.219194383292677e-06,
1195
+ "loss": 0.9647,
1196
+ "step": 99000
1197
+ },
1198
+ {
1199
+ "epoch": 0.59,
1200
+ "learning_rate": 8.159695365026477e-06,
1201
+ "loss": 1.0151,
1202
+ "step": 99500
1203
+ },
1204
+ {
1205
+ "epoch": 0.59,
1206
+ "learning_rate": 8.100196346760279e-06,
1207
+ "loss": 1.0446,
1208
+ "step": 100000
1209
+ },
1210
+ {
1211
+ "epoch": 0.6,
1212
+ "learning_rate": 8.04069732849408e-06,
1213
+ "loss": 1.0598,
1214
+ "step": 100500
1215
+ },
1216
+ {
1217
+ "epoch": 0.6,
1218
+ "learning_rate": 7.981198310227882e-06,
1219
+ "loss": 1.0044,
1220
+ "step": 101000
1221
+ },
1222
+ {
1223
+ "epoch": 0.6,
1224
+ "learning_rate": 7.921699291961684e-06,
1225
+ "loss": 1.0272,
1226
+ "step": 101500
1227
+ },
1228
+ {
1229
+ "epoch": 0.61,
1230
+ "learning_rate": 7.862200273695486e-06,
1231
+ "loss": 1.009,
1232
+ "step": 102000
1233
+ },
1234
+ {
1235
+ "epoch": 0.61,
1236
+ "learning_rate": 7.802701255429285e-06,
1237
+ "loss": 1.0505,
1238
+ "step": 102500
1239
+ },
1240
+ {
1241
+ "epoch": 0.61,
1242
+ "learning_rate": 7.743202237163087e-06,
1243
+ "loss": 1.0559,
1244
+ "step": 103000
1245
+ },
1246
+ {
1247
+ "epoch": 0.62,
1248
+ "learning_rate": 7.683703218896889e-06,
1249
+ "loss": 1.0412,
1250
+ "step": 103500
1251
+ },
1252
+ {
1253
+ "epoch": 0.62,
1254
+ "learning_rate": 7.62420420063069e-06,
1255
+ "loss": 1.0651,
1256
+ "step": 104000
1257
+ },
1258
+ {
1259
+ "epoch": 0.62,
1260
+ "learning_rate": 7.564705182364491e-06,
1261
+ "loss": 0.9968,
1262
+ "step": 104500
1263
+ },
1264
+ {
1265
+ "epoch": 0.62,
1266
+ "learning_rate": 7.505206164098294e-06,
1267
+ "loss": 1.0395,
1268
+ "step": 105000
1269
+ },
1270
+ {
1271
+ "epoch": 0.63,
1272
+ "learning_rate": 7.445707145832094e-06,
1273
+ "loss": 1.0378,
1274
+ "step": 105500
1275
+ },
1276
+ {
1277
+ "epoch": 0.63,
1278
+ "learning_rate": 7.386208127565896e-06,
1279
+ "loss": 0.99,
1280
+ "step": 106000
1281
+ },
1282
+ {
1283
+ "epoch": 0.63,
1284
+ "learning_rate": 7.326709109299697e-06,
1285
+ "loss": 0.9792,
1286
+ "step": 106500
1287
+ },
1288
+ {
1289
+ "epoch": 0.64,
1290
+ "learning_rate": 7.267210091033498e-06,
1291
+ "loss": 1.0482,
1292
+ "step": 107000
1293
+ },
1294
+ {
1295
+ "epoch": 0.64,
1296
+ "learning_rate": 7.207711072767299e-06,
1297
+ "loss": 1.0143,
1298
+ "step": 107500
1299
+ },
1300
+ {
1301
+ "epoch": 0.64,
1302
+ "learning_rate": 7.148212054501102e-06,
1303
+ "loss": 1.0321,
1304
+ "step": 108000
1305
+ },
1306
+ {
1307
+ "epoch": 0.65,
1308
+ "learning_rate": 7.088713036234903e-06,
1309
+ "loss": 1.0312,
1310
+ "step": 108500
1311
+ },
1312
+ {
1313
+ "epoch": 0.65,
1314
+ "learning_rate": 7.029214017968704e-06,
1315
+ "loss": 1.022,
1316
+ "step": 109000
1317
+ },
1318
+ {
1319
+ "epoch": 0.65,
1320
+ "learning_rate": 6.969714999702506e-06,
1321
+ "loss": 1.0529,
1322
+ "step": 109500
1323
+ },
1324
+ {
1325
+ "epoch": 0.65,
1326
+ "learning_rate": 6.9102159814363064e-06,
1327
+ "loss": 1.0269,
1328
+ "step": 110000
1329
+ },
1330
+ {
1331
+ "epoch": 0.66,
1332
+ "learning_rate": 6.850716963170109e-06,
1333
+ "loss": 0.9678,
1334
+ "step": 110500
1335
+ },
1336
+ {
1337
+ "epoch": 0.66,
1338
+ "learning_rate": 6.79121794490391e-06,
1339
+ "loss": 1.0315,
1340
+ "step": 111000
1341
+ },
1342
+ {
1343
+ "epoch": 0.66,
1344
+ "learning_rate": 6.731718926637711e-06,
1345
+ "loss": 0.9849,
1346
+ "step": 111500
1347
+ },
1348
+ {
1349
+ "epoch": 0.67,
1350
+ "learning_rate": 6.672219908371512e-06,
1351
+ "loss": 1.0397,
1352
+ "step": 112000
1353
+ },
1354
+ {
1355
+ "epoch": 0.67,
1356
+ "learning_rate": 6.6127208901053146e-06,
1357
+ "loss": 1.0511,
1358
+ "step": 112500
1359
+ },
1360
+ {
1361
+ "epoch": 0.67,
1362
+ "learning_rate": 6.553221871839115e-06,
1363
+ "loss": 1.0381,
1364
+ "step": 113000
1365
+ },
1366
+ {
1367
+ "epoch": 0.68,
1368
+ "learning_rate": 6.493722853572917e-06,
1369
+ "loss": 1.0584,
1370
+ "step": 113500
1371
+ },
1372
+ {
1373
+ "epoch": 0.68,
1374
+ "learning_rate": 6.434223835306718e-06,
1375
+ "loss": 0.9933,
1376
+ "step": 114000
1377
+ },
1378
+ {
1379
+ "epoch": 0.68,
1380
+ "learning_rate": 6.374724817040519e-06,
1381
+ "loss": 1.0439,
1382
+ "step": 114500
1383
+ },
1384
+ {
1385
+ "epoch": 0.68,
1386
+ "learning_rate": 6.31522579877432e-06,
1387
+ "loss": 1.047,
1388
+ "step": 115000
1389
+ },
1390
+ {
1391
+ "epoch": 0.69,
1392
+ "learning_rate": 6.255726780508123e-06,
1393
+ "loss": 1.0349,
1394
+ "step": 115500
1395
+ },
1396
+ {
1397
+ "epoch": 0.69,
1398
+ "learning_rate": 6.196227762241923e-06,
1399
+ "loss": 0.9923,
1400
+ "step": 116000
1401
+ },
1402
+ {
1403
+ "epoch": 0.69,
1404
+ "learning_rate": 6.136728743975725e-06,
1405
+ "loss": 0.9953,
1406
+ "step": 116500
1407
+ },
1408
+ {
1409
+ "epoch": 0.7,
1410
+ "learning_rate": 6.077229725709526e-06,
1411
+ "loss": 1.0083,
1412
+ "step": 117000
1413
+ },
1414
+ {
1415
+ "epoch": 0.7,
1416
+ "learning_rate": 6.017730707443327e-06,
1417
+ "loss": 1.0258,
1418
+ "step": 117500
1419
+ },
1420
+ {
1421
+ "epoch": 0.7,
1422
+ "learning_rate": 5.958231689177128e-06,
1423
+ "loss": 0.9729,
1424
+ "step": 118000
1425
+ },
1426
+ {
1427
+ "epoch": 0.71,
1428
+ "learning_rate": 5.898732670910931e-06,
1429
+ "loss": 1.0351,
1430
+ "step": 118500
1431
+ },
1432
+ {
1433
+ "epoch": 0.71,
1434
+ "learning_rate": 5.8392336526447314e-06,
1435
+ "loss": 1.0298,
1436
+ "step": 119000
1437
+ },
1438
+ {
1439
+ "epoch": 0.71,
1440
+ "learning_rate": 5.779734634378533e-06,
1441
+ "loss": 1.0067,
1442
+ "step": 119500
1443
+ },
1444
+ {
1445
+ "epoch": 0.71,
1446
+ "learning_rate": 5.7202356161123355e-06,
1447
+ "loss": 1.0235,
1448
+ "step": 120000
1449
+ },
1450
+ {
1451
+ "epoch": 0.72,
1452
+ "learning_rate": 5.660736597846136e-06,
1453
+ "loss": 1.0018,
1454
+ "step": 120500
1455
+ },
1456
+ {
1457
+ "epoch": 0.72,
1458
+ "learning_rate": 5.601237579579938e-06,
1459
+ "loss": 1.0096,
1460
+ "step": 121000
1461
+ },
1462
+ {
1463
+ "epoch": 0.72,
1464
+ "learning_rate": 5.541738561313739e-06,
1465
+ "loss": 0.9732,
1466
+ "step": 121500
1467
+ },
1468
+ {
1469
+ "epoch": 0.73,
1470
+ "learning_rate": 5.48223954304754e-06,
1471
+ "loss": 1.0359,
1472
+ "step": 122000
1473
+ },
1474
+ {
1475
+ "epoch": 0.73,
1476
+ "learning_rate": 5.422740524781341e-06,
1477
+ "loss": 0.9913,
1478
+ "step": 122500
1479
+ },
1480
+ {
1481
+ "epoch": 0.73,
1482
+ "learning_rate": 5.363241506515144e-06,
1483
+ "loss": 1.0132,
1484
+ "step": 123000
1485
+ },
1486
+ {
1487
+ "epoch": 0.73,
1488
+ "learning_rate": 5.303742488248944e-06,
1489
+ "loss": 1.0526,
1490
+ "step": 123500
1491
+ },
1492
+ {
1493
+ "epoch": 0.74,
1494
+ "learning_rate": 5.244243469982746e-06,
1495
+ "loss": 0.9742,
1496
+ "step": 124000
1497
+ },
1498
+ {
1499
+ "epoch": 0.74,
1500
+ "learning_rate": 5.184744451716547e-06,
1501
+ "loss": 0.9988,
1502
+ "step": 124500
1503
+ },
1504
+ {
1505
+ "epoch": 0.74,
1506
+ "learning_rate": 5.125245433450348e-06,
1507
+ "loss": 0.9938,
1508
+ "step": 125000
1509
+ },
1510
+ {
1511
+ "epoch": 0.75,
1512
+ "learning_rate": 5.065746415184149e-06,
1513
+ "loss": 1.022,
1514
+ "step": 125500
1515
+ },
1516
+ {
1517
+ "epoch": 0.75,
1518
+ "learning_rate": 5.006247396917952e-06,
1519
+ "loss": 1.0214,
1520
+ "step": 126000
1521
+ },
1522
+ {
1523
+ "epoch": 0.75,
1524
+ "learning_rate": 4.946748378651752e-06,
1525
+ "loss": 0.9758,
1526
+ "step": 126500
1527
+ },
1528
+ {
1529
+ "epoch": 0.76,
1530
+ "learning_rate": 4.887249360385554e-06,
1531
+ "loss": 1.0235,
1532
+ "step": 127000
1533
+ },
1534
+ {
1535
+ "epoch": 0.76,
1536
+ "learning_rate": 4.827750342119356e-06,
1537
+ "loss": 1.0387,
1538
+ "step": 127500
1539
+ },
1540
+ {
1541
+ "epoch": 0.76,
1542
+ "learning_rate": 4.768251323853157e-06,
1543
+ "loss": 1.0319,
1544
+ "step": 128000
1545
+ },
1546
+ {
1547
+ "epoch": 0.76,
1548
+ "learning_rate": 4.708752305586958e-06,
1549
+ "loss": 0.9961,
1550
+ "step": 128500
1551
+ },
1552
+ {
1553
+ "epoch": 0.77,
1554
+ "learning_rate": 4.64925328732076e-06,
1555
+ "loss": 1.0318,
1556
+ "step": 129000
1557
+ },
1558
+ {
1559
+ "epoch": 0.77,
1560
+ "learning_rate": 4.589754269054561e-06,
1561
+ "loss": 1.1137,
1562
+ "step": 129500
1563
+ },
1564
+ {
1565
+ "epoch": 0.77,
1566
+ "learning_rate": 4.530255250788362e-06,
1567
+ "loss": 1.0398,
1568
+ "step": 130000
1569
+ },
1570
+ {
1571
+ "epoch": 0.78,
1572
+ "learning_rate": 4.470756232522164e-06,
1573
+ "loss": 0.9347,
1574
+ "step": 130500
1575
+ },
1576
+ {
1577
+ "epoch": 0.78,
1578
+ "learning_rate": 4.411257214255965e-06,
1579
+ "loss": 1.0618,
1580
+ "step": 131000
1581
+ },
1582
+ {
1583
+ "epoch": 0.78,
1584
+ "learning_rate": 4.351758195989766e-06,
1585
+ "loss": 0.9826,
1586
+ "step": 131500
1587
+ },
1588
+ {
1589
+ "epoch": 0.79,
1590
+ "learning_rate": 4.292259177723568e-06,
1591
+ "loss": 1.0422,
1592
+ "step": 132000
1593
+ },
1594
+ {
1595
+ "epoch": 0.79,
1596
+ "learning_rate": 4.232760159457369e-06,
1597
+ "loss": 0.9928,
1598
+ "step": 132500
1599
+ },
1600
+ {
1601
+ "epoch": 0.79,
1602
+ "learning_rate": 4.17326114119117e-06,
1603
+ "loss": 1.0487,
1604
+ "step": 133000
1605
+ },
1606
+ {
1607
+ "epoch": 0.79,
1608
+ "learning_rate": 4.113762122924972e-06,
1609
+ "loss": 1.0379,
1610
+ "step": 133500
1611
+ },
1612
+ {
1613
+ "epoch": 0.8,
1614
+ "learning_rate": 4.054263104658773e-06,
1615
+ "loss": 1.0225,
1616
+ "step": 134000
1617
+ },
1618
+ {
1619
+ "epoch": 0.8,
1620
+ "learning_rate": 3.994764086392575e-06,
1621
+ "loss": 0.951,
1622
+ "step": 134500
1623
+ },
1624
+ {
1625
+ "epoch": 0.8,
1626
+ "learning_rate": 3.935265068126376e-06,
1627
+ "loss": 1.0393,
1628
+ "step": 135000
1629
+ },
1630
+ {
1631
+ "epoch": 0.81,
1632
+ "learning_rate": 3.875766049860177e-06,
1633
+ "loss": 0.9851,
1634
+ "step": 135500
1635
+ },
1636
+ {
1637
+ "epoch": 0.81,
1638
+ "learning_rate": 3.816267031593979e-06,
1639
+ "loss": 0.9716,
1640
+ "step": 136000
1641
+ },
1642
+ {
1643
+ "epoch": 0.81,
1644
+ "learning_rate": 3.7567680133277807e-06,
1645
+ "loss": 0.9974,
1646
+ "step": 136500
1647
+ },
1648
+ {
1649
+ "epoch": 0.82,
1650
+ "learning_rate": 3.697268995061582e-06,
1651
+ "loss": 1.0052,
1652
+ "step": 137000
1653
+ },
1654
+ {
1655
+ "epoch": 0.82,
1656
+ "learning_rate": 3.6377699767953835e-06,
1657
+ "loss": 0.9949,
1658
+ "step": 137500
1659
+ },
1660
+ {
1661
+ "epoch": 0.82,
1662
+ "learning_rate": 3.5782709585291847e-06,
1663
+ "loss": 1.0279,
1664
+ "step": 138000
1665
+ },
1666
+ {
1667
+ "epoch": 0.82,
1668
+ "learning_rate": 3.518771940262986e-06,
1669
+ "loss": 0.9636,
1670
+ "step": 138500
1671
+ },
1672
+ {
1673
+ "epoch": 0.83,
1674
+ "learning_rate": 3.4592729219967875e-06,
1675
+ "loss": 1.0322,
1676
+ "step": 139000
1677
+ },
1678
+ {
1679
+ "epoch": 0.83,
1680
+ "learning_rate": 3.3997739037305887e-06,
1681
+ "loss": 1.0153,
1682
+ "step": 139500
1683
+ },
1684
+ {
1685
+ "epoch": 0.83,
1686
+ "learning_rate": 3.3402748854643903e-06,
1687
+ "loss": 0.9982,
1688
+ "step": 140000
1689
+ },
1690
+ {
1691
+ "epoch": 0.84,
1692
+ "learning_rate": 3.2807758671981915e-06,
1693
+ "loss": 1.022,
1694
+ "step": 140500
1695
+ },
1696
+ {
1697
+ "epoch": 0.84,
1698
+ "learning_rate": 3.2212768489319927e-06,
1699
+ "loss": 1.0208,
1700
+ "step": 141000
1701
+ },
1702
+ {
1703
+ "epoch": 0.84,
1704
+ "learning_rate": 3.1617778306657944e-06,
1705
+ "loss": 0.9761,
1706
+ "step": 141500
1707
+ },
1708
+ {
1709
+ "epoch": 0.84,
1710
+ "learning_rate": 3.1022788123995956e-06,
1711
+ "loss": 1.0125,
1712
+ "step": 142000
1713
+ },
1714
+ {
1715
+ "epoch": 0.85,
1716
+ "learning_rate": 3.0427797941333968e-06,
1717
+ "loss": 1.0531,
1718
+ "step": 142500
1719
+ },
1720
+ {
1721
+ "epoch": 0.85,
1722
+ "learning_rate": 2.9832807758671984e-06,
1723
+ "loss": 1.0232,
1724
+ "step": 143000
1725
+ },
1726
+ {
1727
+ "epoch": 0.85,
1728
+ "learning_rate": 2.9237817576009996e-06,
1729
+ "loss": 1.067,
1730
+ "step": 143500
1731
+ },
1732
+ {
1733
+ "epoch": 0.86,
1734
+ "learning_rate": 2.8642827393348012e-06,
1735
+ "loss": 0.9986,
1736
+ "step": 144000
1737
+ },
1738
+ {
1739
+ "epoch": 0.86,
1740
+ "learning_rate": 2.8047837210686024e-06,
1741
+ "loss": 1.0009,
1742
+ "step": 144500
1743
+ },
1744
+ {
1745
+ "epoch": 0.86,
1746
+ "learning_rate": 2.7452847028024036e-06,
1747
+ "loss": 1.0113,
1748
+ "step": 145000
1749
+ },
1750
+ {
1751
+ "epoch": 0.87,
1752
+ "learning_rate": 2.6857856845362052e-06,
1753
+ "loss": 0.9991,
1754
+ "step": 145500
1755
+ },
1756
+ {
1757
+ "epoch": 0.87,
1758
+ "learning_rate": 2.6262866662700064e-06,
1759
+ "loss": 1.0153,
1760
+ "step": 146000
1761
+ },
1762
+ {
1763
+ "epoch": 0.87,
1764
+ "learning_rate": 2.5667876480038085e-06,
1765
+ "loss": 1.0382,
1766
+ "step": 146500
1767
+ },
1768
+ {
1769
+ "epoch": 0.87,
1770
+ "learning_rate": 2.5072886297376097e-06,
1771
+ "loss": 0.993,
1772
+ "step": 147000
1773
+ },
1774
+ {
1775
+ "epoch": 0.88,
1776
+ "learning_rate": 2.447789611471411e-06,
1777
+ "loss": 0.9704,
1778
+ "step": 147500
1779
+ },
1780
+ {
1781
+ "epoch": 0.88,
1782
+ "learning_rate": 2.388290593205212e-06,
1783
+ "loss": 0.9935,
1784
+ "step": 148000
1785
+ },
1786
+ {
1787
+ "epoch": 0.88,
1788
+ "learning_rate": 2.3287915749390137e-06,
1789
+ "loss": 1.0239,
1790
+ "step": 148500
1791
+ },
1792
+ {
1793
+ "epoch": 0.89,
1794
+ "learning_rate": 2.269292556672815e-06,
1795
+ "loss": 0.9927,
1796
+ "step": 149000
1797
+ },
1798
+ {
1799
+ "epoch": 0.89,
1800
+ "learning_rate": 2.2097935384066165e-06,
1801
+ "loss": 1.0262,
1802
+ "step": 149500
1803
+ },
1804
+ {
1805
+ "epoch": 0.89,
1806
+ "learning_rate": 2.150294520140418e-06,
1807
+ "loss": 0.9762,
1808
+ "step": 150000
1809
+ },
1810
+ {
1811
+ "epoch": 0.9,
1812
+ "learning_rate": 2.0907955018742194e-06,
1813
+ "loss": 1.0298,
1814
+ "step": 150500
1815
+ },
1816
+ {
1817
+ "epoch": 0.9,
1818
+ "learning_rate": 2.0312964836080206e-06,
1819
+ "loss": 0.9869,
1820
+ "step": 151000
1821
+ },
1822
+ {
1823
+ "epoch": 0.9,
1824
+ "learning_rate": 1.971797465341822e-06,
1825
+ "loss": 1.0192,
1826
+ "step": 151500
1827
+ },
1828
+ {
1829
+ "epoch": 0.9,
1830
+ "learning_rate": 1.9122984470756234e-06,
1831
+ "loss": 1.0034,
1832
+ "step": 152000
1833
+ },
1834
+ {
1835
+ "epoch": 0.91,
1836
+ "learning_rate": 1.8527994288094248e-06,
1837
+ "loss": 1.0179,
1838
+ "step": 152500
1839
+ },
1840
+ {
1841
+ "epoch": 0.91,
1842
+ "learning_rate": 1.7933004105432262e-06,
1843
+ "loss": 1.0189,
1844
+ "step": 153000
1845
+ },
1846
+ {
1847
+ "epoch": 0.91,
1848
+ "learning_rate": 1.7338013922770274e-06,
1849
+ "loss": 0.9545,
1850
+ "step": 153500
1851
+ },
1852
+ {
1853
+ "epoch": 0.92,
1854
+ "learning_rate": 1.6743023740108288e-06,
1855
+ "loss": 1.0462,
1856
+ "step": 154000
1857
+ },
1858
+ {
1859
+ "epoch": 0.92,
1860
+ "learning_rate": 1.6148033557446302e-06,
1861
+ "loss": 1.0083,
1862
+ "step": 154500
1863
+ },
1864
+ {
1865
+ "epoch": 0.92,
1866
+ "learning_rate": 1.5553043374784319e-06,
1867
+ "loss": 0.9755,
1868
+ "step": 155000
1869
+ },
1870
+ {
1871
+ "epoch": 0.93,
1872
+ "learning_rate": 1.4958053192122333e-06,
1873
+ "loss": 1.0234,
1874
+ "step": 155500
1875
+ },
1876
+ {
1877
+ "epoch": 0.93,
1878
+ "learning_rate": 1.4363063009460345e-06,
1879
+ "loss": 1.0116,
1880
+ "step": 156000
1881
+ },
1882
+ {
1883
+ "epoch": 0.93,
1884
+ "learning_rate": 1.3768072826798359e-06,
1885
+ "loss": 1.0502,
1886
+ "step": 156500
1887
+ },
1888
+ {
1889
+ "epoch": 0.93,
1890
+ "learning_rate": 1.3173082644136373e-06,
1891
+ "loss": 0.9593,
1892
+ "step": 157000
1893
+ },
1894
+ {
1895
+ "epoch": 0.94,
1896
+ "learning_rate": 1.2578092461474387e-06,
1897
+ "loss": 1.0195,
1898
+ "step": 157500
1899
+ },
1900
+ {
1901
+ "epoch": 0.94,
1902
+ "learning_rate": 1.19831022788124e-06,
1903
+ "loss": 1.0212,
1904
+ "step": 158000
1905
+ },
1906
+ {
1907
+ "epoch": 0.94,
1908
+ "learning_rate": 1.1388112096150413e-06,
1909
+ "loss": 0.9523,
1910
+ "step": 158500
1911
+ },
1912
+ {
1913
+ "epoch": 0.95,
1914
+ "learning_rate": 1.079312191348843e-06,
1915
+ "loss": 0.9902,
1916
+ "step": 159000
1917
+ },
1918
+ {
1919
+ "epoch": 0.95,
1920
+ "learning_rate": 1.0198131730826442e-06,
1921
+ "loss": 1.0248,
1922
+ "step": 159500
1923
+ },
1924
+ {
1925
+ "epoch": 0.95,
1926
+ "learning_rate": 9.603141548164456e-07,
1927
+ "loss": 1.0062,
1928
+ "step": 160000
1929
+ },
1930
+ {
1931
+ "epoch": 0.95,
1932
+ "learning_rate": 9.00815136550247e-07,
1933
+ "loss": 1.0035,
1934
+ "step": 160500
1935
+ },
1936
+ {
1937
+ "epoch": 0.96,
1938
+ "learning_rate": 8.413161182840483e-07,
1939
+ "loss": 0.9755,
1940
+ "step": 161000
1941
+ },
1942
+ {
1943
+ "epoch": 0.96,
1944
+ "learning_rate": 7.818171000178498e-07,
1945
+ "loss": 1.0131,
1946
+ "step": 161500
1947
+ },
1948
+ {
1949
+ "epoch": 0.96,
1950
+ "learning_rate": 7.223180817516512e-07,
1951
+ "loss": 1.0353,
1952
+ "step": 162000
1953
+ },
1954
+ {
1955
+ "epoch": 0.97,
1956
+ "learning_rate": 6.628190634854525e-07,
1957
+ "loss": 1.0138,
1958
+ "step": 162500
1959
+ },
1960
+ {
1961
+ "epoch": 0.97,
1962
+ "learning_rate": 6.033200452192539e-07,
1963
+ "loss": 0.9738,
1964
+ "step": 163000
1965
+ },
1966
+ {
1967
+ "epoch": 0.97,
1968
+ "learning_rate": 5.438210269530553e-07,
1969
+ "loss": 0.9878,
1970
+ "step": 163500
1971
+ },
1972
+ {
1973
+ "epoch": 0.98,
1974
+ "learning_rate": 4.843220086868568e-07,
1975
+ "loss": 0.9787,
1976
+ "step": 164000
1977
+ },
1978
+ {
1979
+ "epoch": 0.98,
1980
+ "learning_rate": 4.2482299042065806e-07,
1981
+ "loss": 0.9903,
1982
+ "step": 164500
1983
+ },
1984
+ {
1985
+ "epoch": 0.98,
1986
+ "learning_rate": 3.653239721544595e-07,
1987
+ "loss": 1.032,
1988
+ "step": 165000
1989
+ },
1990
+ {
1991
+ "epoch": 0.98,
1992
+ "learning_rate": 3.0582495388826084e-07,
1993
+ "loss": 1.0436,
1994
+ "step": 165500
1995
+ },
1996
+ {
1997
+ "epoch": 0.99,
1998
+ "learning_rate": 2.4632593562206225e-07,
1999
+ "loss": 0.9956,
2000
+ "step": 166000
2001
+ },
2002
+ {
2003
+ "epoch": 0.99,
2004
+ "learning_rate": 1.8682691735586366e-07,
2005
+ "loss": 1.0102,
2006
+ "step": 166500
2007
+ },
2008
+ {
2009
+ "epoch": 0.99,
2010
+ "learning_rate": 1.2732789908966505e-07,
2011
+ "loss": 0.9634,
2012
+ "step": 167000
2013
+ },
2014
+ {
2015
+ "epoch": 1.0,
2016
+ "learning_rate": 6.782888082346642e-08,
2017
+ "loss": 1.004,
2018
+ "step": 167500
2019
+ },
2020
+ {
2021
+ "epoch": 1.0,
2022
+ "learning_rate": 8.329862557267807e-09,
2023
+ "loss": 0.9777,
2024
+ "step": 168000
2025
+ }
2026
+ ],
2027
+ "max_steps": 168070,
2028
+ "num_train_epochs": 1,
2029
+ "total_flos": 6437535729057792.0,
2030
+ "trial_name": null,
2031
+ "trial_params": null
2032
+ }