Osamarafique998
commited on
Commit
•
45f50b1
1
Parent(s):
e5aba18
Delete
Browse files- trainer_state.json +0 -1816
trainer_state.json
DELETED
@@ -1,1816 +0,0 @@
|
|
1 |
-
{
|
2 |
-
"best_metric": null,
|
3 |
-
"best_model_checkpoint": null,
|
4 |
-
"epoch": 4.123711340206185,
|
5 |
-
"global_step": 300,
|
6 |
-
"is_hyper_param_search": false,
|
7 |
-
"is_local_process_zero": true,
|
8 |
-
"is_world_process_zero": true,
|
9 |
-
"log_history": [
|
10 |
-
{
|
11 |
-
"epoch": 0.01,
|
12 |
-
"learning_rate": 3.7037037037037036e-07,
|
13 |
-
"loss": 0.238,
|
14 |
-
"step": 1
|
15 |
-
},
|
16 |
-
{
|
17 |
-
"epoch": 0.03,
|
18 |
-
"learning_rate": 7.407407407407407e-07,
|
19 |
-
"loss": 0.2949,
|
20 |
-
"step": 2
|
21 |
-
},
|
22 |
-
{
|
23 |
-
"epoch": 0.04,
|
24 |
-
"learning_rate": 1.111111111111111e-06,
|
25 |
-
"loss": 0.2922,
|
26 |
-
"step": 3
|
27 |
-
},
|
28 |
-
{
|
29 |
-
"epoch": 0.05,
|
30 |
-
"learning_rate": 1.4814814814814815e-06,
|
31 |
-
"loss": 0.1945,
|
32 |
-
"step": 4
|
33 |
-
},
|
34 |
-
{
|
35 |
-
"epoch": 0.07,
|
36 |
-
"learning_rate": 1.8518518518518519e-06,
|
37 |
-
"loss": 0.1929,
|
38 |
-
"step": 5
|
39 |
-
},
|
40 |
-
{
|
41 |
-
"epoch": 0.08,
|
42 |
-
"learning_rate": 2.222222222222222e-06,
|
43 |
-
"loss": 0.1124,
|
44 |
-
"step": 6
|
45 |
-
},
|
46 |
-
{
|
47 |
-
"epoch": 0.1,
|
48 |
-
"learning_rate": 2.5925925925925925e-06,
|
49 |
-
"loss": 0.0661,
|
50 |
-
"step": 7
|
51 |
-
},
|
52 |
-
{
|
53 |
-
"epoch": 0.11,
|
54 |
-
"learning_rate": 2.962962962962963e-06,
|
55 |
-
"loss": 0.0261,
|
56 |
-
"step": 8
|
57 |
-
},
|
58 |
-
{
|
59 |
-
"epoch": 0.12,
|
60 |
-
"learning_rate": 3.3333333333333333e-06,
|
61 |
-
"loss": 0.008,
|
62 |
-
"step": 9
|
63 |
-
},
|
64 |
-
{
|
65 |
-
"epoch": 0.14,
|
66 |
-
"learning_rate": 3.7037037037037037e-06,
|
67 |
-
"loss": 0.0013,
|
68 |
-
"step": 10
|
69 |
-
},
|
70 |
-
{
|
71 |
-
"epoch": 0.15,
|
72 |
-
"learning_rate": 4.074074074074074e-06,
|
73 |
-
"loss": 0.0008,
|
74 |
-
"step": 11
|
75 |
-
},
|
76 |
-
{
|
77 |
-
"epoch": 0.16,
|
78 |
-
"learning_rate": 4.444444444444444e-06,
|
79 |
-
"loss": 0.0001,
|
80 |
-
"step": 12
|
81 |
-
},
|
82 |
-
{
|
83 |
-
"epoch": 0.18,
|
84 |
-
"learning_rate": 4.814814814814815e-06,
|
85 |
-
"loss": 0.0001,
|
86 |
-
"step": 13
|
87 |
-
},
|
88 |
-
{
|
89 |
-
"epoch": 0.19,
|
90 |
-
"learning_rate": 5.185185185185185e-06,
|
91 |
-
"loss": 0.0002,
|
92 |
-
"step": 14
|
93 |
-
},
|
94 |
-
{
|
95 |
-
"epoch": 0.21,
|
96 |
-
"learning_rate": 5.555555555555557e-06,
|
97 |
-
"loss": 0.0,
|
98 |
-
"step": 15
|
99 |
-
},
|
100 |
-
{
|
101 |
-
"epoch": 0.22,
|
102 |
-
"learning_rate": 5.925925925925926e-06,
|
103 |
-
"loss": 0.0,
|
104 |
-
"step": 16
|
105 |
-
},
|
106 |
-
{
|
107 |
-
"epoch": 0.23,
|
108 |
-
"learning_rate": 6.296296296296297e-06,
|
109 |
-
"loss": 0.0001,
|
110 |
-
"step": 17
|
111 |
-
},
|
112 |
-
{
|
113 |
-
"epoch": 0.25,
|
114 |
-
"learning_rate": 6.666666666666667e-06,
|
115 |
-
"loss": 0.0,
|
116 |
-
"step": 18
|
117 |
-
},
|
118 |
-
{
|
119 |
-
"epoch": 0.26,
|
120 |
-
"learning_rate": 7.0370370370370375e-06,
|
121 |
-
"loss": 0.0,
|
122 |
-
"step": 19
|
123 |
-
},
|
124 |
-
{
|
125 |
-
"epoch": 0.27,
|
126 |
-
"learning_rate": 7.4074074074074075e-06,
|
127 |
-
"loss": 0.0,
|
128 |
-
"step": 20
|
129 |
-
},
|
130 |
-
{
|
131 |
-
"epoch": 0.29,
|
132 |
-
"learning_rate": 7.77777777777778e-06,
|
133 |
-
"loss": 0.0,
|
134 |
-
"step": 21
|
135 |
-
},
|
136 |
-
{
|
137 |
-
"epoch": 0.3,
|
138 |
-
"learning_rate": 8.148148148148148e-06,
|
139 |
-
"loss": 0.0,
|
140 |
-
"step": 22
|
141 |
-
},
|
142 |
-
{
|
143 |
-
"epoch": 0.32,
|
144 |
-
"learning_rate": 8.518518518518519e-06,
|
145 |
-
"loss": 0.0,
|
146 |
-
"step": 23
|
147 |
-
},
|
148 |
-
{
|
149 |
-
"epoch": 0.33,
|
150 |
-
"learning_rate": 8.888888888888888e-06,
|
151 |
-
"loss": 0.0,
|
152 |
-
"step": 24
|
153 |
-
},
|
154 |
-
{
|
155 |
-
"epoch": 0.34,
|
156 |
-
"learning_rate": 9.25925925925926e-06,
|
157 |
-
"loss": 0.0,
|
158 |
-
"step": 25
|
159 |
-
},
|
160 |
-
{
|
161 |
-
"epoch": 0.36,
|
162 |
-
"learning_rate": 9.62962962962963e-06,
|
163 |
-
"loss": 0.0,
|
164 |
-
"step": 26
|
165 |
-
},
|
166 |
-
{
|
167 |
-
"epoch": 0.37,
|
168 |
-
"learning_rate": 1e-05,
|
169 |
-
"loss": 0.0,
|
170 |
-
"step": 27
|
171 |
-
},
|
172 |
-
{
|
173 |
-
"epoch": 0.38,
|
174 |
-
"learning_rate": 1.037037037037037e-05,
|
175 |
-
"loss": 0.0,
|
176 |
-
"step": 28
|
177 |
-
},
|
178 |
-
{
|
179 |
-
"epoch": 0.4,
|
180 |
-
"learning_rate": 1.0740740740740742e-05,
|
181 |
-
"loss": 0.0,
|
182 |
-
"step": 29
|
183 |
-
},
|
184 |
-
{
|
185 |
-
"epoch": 0.41,
|
186 |
-
"learning_rate": 1.1111111111111113e-05,
|
187 |
-
"loss": 0.0,
|
188 |
-
"step": 30
|
189 |
-
},
|
190 |
-
{
|
191 |
-
"epoch": 0.43,
|
192 |
-
"learning_rate": 1.1481481481481482e-05,
|
193 |
-
"loss": 0.0,
|
194 |
-
"step": 31
|
195 |
-
},
|
196 |
-
{
|
197 |
-
"epoch": 0.44,
|
198 |
-
"learning_rate": 1.1851851851851852e-05,
|
199 |
-
"loss": 0.0005,
|
200 |
-
"step": 32
|
201 |
-
},
|
202 |
-
{
|
203 |
-
"epoch": 0.45,
|
204 |
-
"learning_rate": 1.2222222222222224e-05,
|
205 |
-
"loss": 0.0,
|
206 |
-
"step": 33
|
207 |
-
},
|
208 |
-
{
|
209 |
-
"epoch": 0.47,
|
210 |
-
"learning_rate": 1.2592592592592593e-05,
|
211 |
-
"loss": 0.0783,
|
212 |
-
"step": 34
|
213 |
-
},
|
214 |
-
{
|
215 |
-
"epoch": 0.48,
|
216 |
-
"learning_rate": 1.2962962962962964e-05,
|
217 |
-
"loss": 0.0,
|
218 |
-
"step": 35
|
219 |
-
},
|
220 |
-
{
|
221 |
-
"epoch": 0.49,
|
222 |
-
"learning_rate": 1.3333333333333333e-05,
|
223 |
-
"loss": 0.0,
|
224 |
-
"step": 36
|
225 |
-
},
|
226 |
-
{
|
227 |
-
"epoch": 0.51,
|
228 |
-
"learning_rate": 1.3703703703703706e-05,
|
229 |
-
"loss": 0.0004,
|
230 |
-
"step": 37
|
231 |
-
},
|
232 |
-
{
|
233 |
-
"epoch": 0.52,
|
234 |
-
"learning_rate": 1.4074074074074075e-05,
|
235 |
-
"loss": 0.0005,
|
236 |
-
"step": 38
|
237 |
-
},
|
238 |
-
{
|
239 |
-
"epoch": 0.54,
|
240 |
-
"learning_rate": 1.4444444444444446e-05,
|
241 |
-
"loss": 0.0003,
|
242 |
-
"step": 39
|
243 |
-
},
|
244 |
-
{
|
245 |
-
"epoch": 0.55,
|
246 |
-
"learning_rate": 1.4814814814814815e-05,
|
247 |
-
"loss": 0.0006,
|
248 |
-
"step": 40
|
249 |
-
},
|
250 |
-
{
|
251 |
-
"epoch": 0.56,
|
252 |
-
"learning_rate": 1.5185185185185187e-05,
|
253 |
-
"loss": 0.0039,
|
254 |
-
"step": 41
|
255 |
-
},
|
256 |
-
{
|
257 |
-
"epoch": 0.58,
|
258 |
-
"learning_rate": 1.555555555555556e-05,
|
259 |
-
"loss": 0.0048,
|
260 |
-
"step": 42
|
261 |
-
},
|
262 |
-
{
|
263 |
-
"epoch": 0.59,
|
264 |
-
"learning_rate": 1.5925925925925926e-05,
|
265 |
-
"loss": 0.0,
|
266 |
-
"step": 43
|
267 |
-
},
|
268 |
-
{
|
269 |
-
"epoch": 0.6,
|
270 |
-
"learning_rate": 1.6296296296296297e-05,
|
271 |
-
"loss": 0.0004,
|
272 |
-
"step": 44
|
273 |
-
},
|
274 |
-
{
|
275 |
-
"epoch": 0.62,
|
276 |
-
"learning_rate": 1.6666666666666667e-05,
|
277 |
-
"loss": 0.0048,
|
278 |
-
"step": 45
|
279 |
-
},
|
280 |
-
{
|
281 |
-
"epoch": 0.63,
|
282 |
-
"learning_rate": 1.7037037037037038e-05,
|
283 |
-
"loss": 0.0,
|
284 |
-
"step": 46
|
285 |
-
},
|
286 |
-
{
|
287 |
-
"epoch": 0.65,
|
288 |
-
"learning_rate": 1.740740740740741e-05,
|
289 |
-
"loss": 0.0167,
|
290 |
-
"step": 47
|
291 |
-
},
|
292 |
-
{
|
293 |
-
"epoch": 0.66,
|
294 |
-
"learning_rate": 1.7777777777777777e-05,
|
295 |
-
"loss": 0.0228,
|
296 |
-
"step": 48
|
297 |
-
},
|
298 |
-
{
|
299 |
-
"epoch": 0.67,
|
300 |
-
"learning_rate": 1.814814814814815e-05,
|
301 |
-
"loss": 0.0002,
|
302 |
-
"step": 49
|
303 |
-
},
|
304 |
-
{
|
305 |
-
"epoch": 0.69,
|
306 |
-
"learning_rate": 1.851851851851852e-05,
|
307 |
-
"loss": 0.0008,
|
308 |
-
"step": 50
|
309 |
-
},
|
310 |
-
{
|
311 |
-
"epoch": 0.7,
|
312 |
-
"learning_rate": 1.888888888888889e-05,
|
313 |
-
"loss": 0.0009,
|
314 |
-
"step": 51
|
315 |
-
},
|
316 |
-
{
|
317 |
-
"epoch": 0.71,
|
318 |
-
"learning_rate": 1.925925925925926e-05,
|
319 |
-
"loss": 0.1318,
|
320 |
-
"step": 52
|
321 |
-
},
|
322 |
-
{
|
323 |
-
"epoch": 0.73,
|
324 |
-
"learning_rate": 1.962962962962963e-05,
|
325 |
-
"loss": 0.0004,
|
326 |
-
"step": 53
|
327 |
-
},
|
328 |
-
{
|
329 |
-
"epoch": 0.74,
|
330 |
-
"learning_rate": 2e-05,
|
331 |
-
"loss": 0.0016,
|
332 |
-
"step": 54
|
333 |
-
},
|
334 |
-
{
|
335 |
-
"epoch": 0.76,
|
336 |
-
"learning_rate": 1.9999983812448848e-05,
|
337 |
-
"loss": 0.0055,
|
338 |
-
"step": 55
|
339 |
-
},
|
340 |
-
{
|
341 |
-
"epoch": 0.77,
|
342 |
-
"learning_rate": 1.9999935249847798e-05,
|
343 |
-
"loss": 0.0008,
|
344 |
-
"step": 56
|
345 |
-
},
|
346 |
-
{
|
347 |
-
"epoch": 0.78,
|
348 |
-
"learning_rate": 1.9999854312354064e-05,
|
349 |
-
"loss": 0.0007,
|
350 |
-
"step": 57
|
351 |
-
},
|
352 |
-
{
|
353 |
-
"epoch": 0.8,
|
354 |
-
"learning_rate": 1.9999741000229695e-05,
|
355 |
-
"loss": 0.0017,
|
356 |
-
"step": 58
|
357 |
-
},
|
358 |
-
{
|
359 |
-
"epoch": 0.81,
|
360 |
-
"learning_rate": 1.999959531384153e-05,
|
361 |
-
"loss": 0.0004,
|
362 |
-
"step": 59
|
363 |
-
},
|
364 |
-
{
|
365 |
-
"epoch": 0.82,
|
366 |
-
"learning_rate": 1.9999417253661235e-05,
|
367 |
-
"loss": 0.0004,
|
368 |
-
"step": 60
|
369 |
-
},
|
370 |
-
{
|
371 |
-
"epoch": 0.84,
|
372 |
-
"learning_rate": 1.9999206820265283e-05,
|
373 |
-
"loss": 0.0005,
|
374 |
-
"step": 61
|
375 |
-
},
|
376 |
-
{
|
377 |
-
"epoch": 0.85,
|
378 |
-
"learning_rate": 1.9998964014334946e-05,
|
379 |
-
"loss": 0.0003,
|
380 |
-
"step": 62
|
381 |
-
},
|
382 |
-
{
|
383 |
-
"epoch": 0.87,
|
384 |
-
"learning_rate": 1.9998688836656322e-05,
|
385 |
-
"loss": 0.0003,
|
386 |
-
"step": 63
|
387 |
-
},
|
388 |
-
{
|
389 |
-
"epoch": 0.88,
|
390 |
-
"learning_rate": 1.9998381288120296e-05,
|
391 |
-
"loss": 0.0023,
|
392 |
-
"step": 64
|
393 |
-
},
|
394 |
-
{
|
395 |
-
"epoch": 0.89,
|
396 |
-
"learning_rate": 1.9998041369722556e-05,
|
397 |
-
"loss": 0.0001,
|
398 |
-
"step": 65
|
399 |
-
},
|
400 |
-
{
|
401 |
-
"epoch": 0.91,
|
402 |
-
"learning_rate": 1.9997669082563597e-05,
|
403 |
-
"loss": 0.0001,
|
404 |
-
"step": 66
|
405 |
-
},
|
406 |
-
{
|
407 |
-
"epoch": 0.92,
|
408 |
-
"learning_rate": 1.9997264427848702e-05,
|
409 |
-
"loss": 0.0001,
|
410 |
-
"step": 67
|
411 |
-
},
|
412 |
-
{
|
413 |
-
"epoch": 0.93,
|
414 |
-
"learning_rate": 1.999682740688794e-05,
|
415 |
-
"loss": 0.0006,
|
416 |
-
"step": 68
|
417 |
-
},
|
418 |
-
{
|
419 |
-
"epoch": 0.95,
|
420 |
-
"learning_rate": 1.9996358021096174e-05,
|
421 |
-
"loss": 0.0153,
|
422 |
-
"step": 69
|
423 |
-
},
|
424 |
-
{
|
425 |
-
"epoch": 0.96,
|
426 |
-
"learning_rate": 1.999585627199305e-05,
|
427 |
-
"loss": 0.0001,
|
428 |
-
"step": 70
|
429 |
-
},
|
430 |
-
{
|
431 |
-
"epoch": 0.98,
|
432 |
-
"learning_rate": 1.9995322161202975e-05,
|
433 |
-
"loss": 0.0001,
|
434 |
-
"step": 71
|
435 |
-
},
|
436 |
-
{
|
437 |
-
"epoch": 0.99,
|
438 |
-
"learning_rate": 1.9994755690455154e-05,
|
439 |
-
"loss": 0.0002,
|
440 |
-
"step": 72
|
441 |
-
},
|
442 |
-
{
|
443 |
-
"epoch": 1.0,
|
444 |
-
"learning_rate": 1.9994156861583522e-05,
|
445 |
-
"loss": 0.0005,
|
446 |
-
"step": 73
|
447 |
-
},
|
448 |
-
{
|
449 |
-
"epoch": 1.02,
|
450 |
-
"learning_rate": 1.9993525676526807e-05,
|
451 |
-
"loss": 0.0003,
|
452 |
-
"step": 74
|
453 |
-
},
|
454 |
-
{
|
455 |
-
"epoch": 1.03,
|
456 |
-
"learning_rate": 1.9992862137328475e-05,
|
457 |
-
"loss": 0.0004,
|
458 |
-
"step": 75
|
459 |
-
},
|
460 |
-
{
|
461 |
-
"epoch": 1.04,
|
462 |
-
"learning_rate": 1.9992166246136738e-05,
|
463 |
-
"loss": 0.0002,
|
464 |
-
"step": 76
|
465 |
-
},
|
466 |
-
{
|
467 |
-
"epoch": 1.06,
|
468 |
-
"learning_rate": 1.9991438005204557e-05,
|
469 |
-
"loss": 0.0002,
|
470 |
-
"step": 77
|
471 |
-
},
|
472 |
-
{
|
473 |
-
"epoch": 1.07,
|
474 |
-
"learning_rate": 1.999067741688961e-05,
|
475 |
-
"loss": 0.0002,
|
476 |
-
"step": 78
|
477 |
-
},
|
478 |
-
{
|
479 |
-
"epoch": 1.09,
|
480 |
-
"learning_rate": 1.9989884483654315e-05,
|
481 |
-
"loss": 0.0007,
|
482 |
-
"step": 79
|
483 |
-
},
|
484 |
-
{
|
485 |
-
"epoch": 1.1,
|
486 |
-
"learning_rate": 1.9989059208065803e-05,
|
487 |
-
"loss": 0.0002,
|
488 |
-
"step": 80
|
489 |
-
},
|
490 |
-
{
|
491 |
-
"epoch": 1.11,
|
492 |
-
"learning_rate": 1.998820159279591e-05,
|
493 |
-
"loss": 0.0001,
|
494 |
-
"step": 81
|
495 |
-
},
|
496 |
-
{
|
497 |
-
"epoch": 1.13,
|
498 |
-
"learning_rate": 1.9987311640621173e-05,
|
499 |
-
"loss": 0.0,
|
500 |
-
"step": 82
|
501 |
-
},
|
502 |
-
{
|
503 |
-
"epoch": 1.14,
|
504 |
-
"learning_rate": 1.9986389354422823e-05,
|
505 |
-
"loss": 0.0001,
|
506 |
-
"step": 83
|
507 |
-
},
|
508 |
-
{
|
509 |
-
"epoch": 1.15,
|
510 |
-
"learning_rate": 1.998543473718677e-05,
|
511 |
-
"loss": 0.0486,
|
512 |
-
"step": 84
|
513 |
-
},
|
514 |
-
{
|
515 |
-
"epoch": 1.17,
|
516 |
-
"learning_rate": 1.9984447792003602e-05,
|
517 |
-
"loss": 0.0001,
|
518 |
-
"step": 85
|
519 |
-
},
|
520 |
-
{
|
521 |
-
"epoch": 1.18,
|
522 |
-
"learning_rate": 1.9983428522068556e-05,
|
523 |
-
"loss": 0.0,
|
524 |
-
"step": 86
|
525 |
-
},
|
526 |
-
{
|
527 |
-
"epoch": 1.2,
|
528 |
-
"learning_rate": 1.998237693068153e-05,
|
529 |
-
"loss": 0.0006,
|
530 |
-
"step": 87
|
531 |
-
},
|
532 |
-
{
|
533 |
-
"epoch": 1.21,
|
534 |
-
"learning_rate": 1.998129302124707e-05,
|
535 |
-
"loss": 0.0005,
|
536 |
-
"step": 88
|
537 |
-
},
|
538 |
-
{
|
539 |
-
"epoch": 1.22,
|
540 |
-
"learning_rate": 1.9980176797274335e-05,
|
541 |
-
"loss": 0.0,
|
542 |
-
"step": 89
|
543 |
-
},
|
544 |
-
{
|
545 |
-
"epoch": 1.24,
|
546 |
-
"learning_rate": 1.997902826237712e-05,
|
547 |
-
"loss": 0.0002,
|
548 |
-
"step": 90
|
549 |
-
},
|
550 |
-
{
|
551 |
-
"epoch": 1.25,
|
552 |
-
"learning_rate": 1.9977847420273807e-05,
|
553 |
-
"loss": 0.0001,
|
554 |
-
"step": 91
|
555 |
-
},
|
556 |
-
{
|
557 |
-
"epoch": 1.26,
|
558 |
-
"learning_rate": 1.9976634274787396e-05,
|
559 |
-
"loss": 0.0001,
|
560 |
-
"step": 92
|
561 |
-
},
|
562 |
-
{
|
563 |
-
"epoch": 1.28,
|
564 |
-
"learning_rate": 1.997538882984545e-05,
|
565 |
-
"loss": 0.0008,
|
566 |
-
"step": 93
|
567 |
-
},
|
568 |
-
{
|
569 |
-
"epoch": 1.29,
|
570 |
-
"learning_rate": 1.9974111089480112e-05,
|
571 |
-
"loss": 0.0001,
|
572 |
-
"step": 94
|
573 |
-
},
|
574 |
-
{
|
575 |
-
"epoch": 1.31,
|
576 |
-
"learning_rate": 1.997280105782808e-05,
|
577 |
-
"loss": 0.0007,
|
578 |
-
"step": 95
|
579 |
-
},
|
580 |
-
{
|
581 |
-
"epoch": 1.32,
|
582 |
-
"learning_rate": 1.9971458739130598e-05,
|
583 |
-
"loss": 0.0,
|
584 |
-
"step": 96
|
585 |
-
},
|
586 |
-
{
|
587 |
-
"epoch": 1.33,
|
588 |
-
"learning_rate": 1.997008413773343e-05,
|
589 |
-
"loss": 0.0,
|
590 |
-
"step": 97
|
591 |
-
},
|
592 |
-
{
|
593 |
-
"epoch": 1.35,
|
594 |
-
"learning_rate": 1.9968677258086868e-05,
|
595 |
-
"loss": 0.0,
|
596 |
-
"step": 98
|
597 |
-
},
|
598 |
-
{
|
599 |
-
"epoch": 1.36,
|
600 |
-
"learning_rate": 1.9967238104745695e-05,
|
601 |
-
"loss": 0.0,
|
602 |
-
"step": 99
|
603 |
-
},
|
604 |
-
{
|
605 |
-
"epoch": 1.37,
|
606 |
-
"learning_rate": 1.9965766682369188e-05,
|
607 |
-
"loss": 0.0,
|
608 |
-
"step": 100
|
609 |
-
},
|
610 |
-
{
|
611 |
-
"epoch": 1.39,
|
612 |
-
"learning_rate": 1.9964262995721088e-05,
|
613 |
-
"loss": 0.0011,
|
614 |
-
"step": 101
|
615 |
-
},
|
616 |
-
{
|
617 |
-
"epoch": 1.4,
|
618 |
-
"learning_rate": 1.99627270496696e-05,
|
619 |
-
"loss": 0.0,
|
620 |
-
"step": 102
|
621 |
-
},
|
622 |
-
{
|
623 |
-
"epoch": 1.42,
|
624 |
-
"learning_rate": 1.996115884918736e-05,
|
625 |
-
"loss": 0.0001,
|
626 |
-
"step": 103
|
627 |
-
},
|
628 |
-
{
|
629 |
-
"epoch": 1.43,
|
630 |
-
"learning_rate": 1.9959558399351443e-05,
|
631 |
-
"loss": 0.0023,
|
632 |
-
"step": 104
|
633 |
-
},
|
634 |
-
{
|
635 |
-
"epoch": 1.44,
|
636 |
-
"learning_rate": 1.995792570534331e-05,
|
637 |
-
"loss": 0.0091,
|
638 |
-
"step": 105
|
639 |
-
},
|
640 |
-
{
|
641 |
-
"epoch": 1.46,
|
642 |
-
"learning_rate": 1.9956260772448835e-05,
|
643 |
-
"loss": 0.04,
|
644 |
-
"step": 106
|
645 |
-
},
|
646 |
-
{
|
647 |
-
"epoch": 1.47,
|
648 |
-
"learning_rate": 1.9954563606058246e-05,
|
649 |
-
"loss": 0.0001,
|
650 |
-
"step": 107
|
651 |
-
},
|
652 |
-
{
|
653 |
-
"epoch": 1.48,
|
654 |
-
"learning_rate": 1.995283421166614e-05,
|
655 |
-
"loss": 0.0003,
|
656 |
-
"step": 108
|
657 |
-
},
|
658 |
-
{
|
659 |
-
"epoch": 1.5,
|
660 |
-
"learning_rate": 1.9951072594871452e-05,
|
661 |
-
"loss": 0.0007,
|
662 |
-
"step": 109
|
663 |
-
},
|
664 |
-
{
|
665 |
-
"epoch": 1.51,
|
666 |
-
"learning_rate": 1.994927876137743e-05,
|
667 |
-
"loss": 0.0454,
|
668 |
-
"step": 110
|
669 |
-
},
|
670 |
-
{
|
671 |
-
"epoch": 1.53,
|
672 |
-
"learning_rate": 1.994745271699163e-05,
|
673 |
-
"loss": 0.0023,
|
674 |
-
"step": 111
|
675 |
-
},
|
676 |
-
{
|
677 |
-
"epoch": 1.54,
|
678 |
-
"learning_rate": 1.9945594467625895e-05,
|
679 |
-
"loss": 0.0061,
|
680 |
-
"step": 112
|
681 |
-
},
|
682 |
-
{
|
683 |
-
"epoch": 1.55,
|
684 |
-
"learning_rate": 1.9943704019296316e-05,
|
685 |
-
"loss": 0.0008,
|
686 |
-
"step": 113
|
687 |
-
},
|
688 |
-
{
|
689 |
-
"epoch": 1.57,
|
690 |
-
"learning_rate": 1.9941781378123244e-05,
|
691 |
-
"loss": 0.0063,
|
692 |
-
"step": 114
|
693 |
-
},
|
694 |
-
{
|
695 |
-
"epoch": 1.58,
|
696 |
-
"learning_rate": 1.993982655033125e-05,
|
697 |
-
"loss": 0.0001,
|
698 |
-
"step": 115
|
699 |
-
},
|
700 |
-
{
|
701 |
-
"epoch": 1.59,
|
702 |
-
"learning_rate": 1.9937839542249108e-05,
|
703 |
-
"loss": 0.0004,
|
704 |
-
"step": 116
|
705 |
-
},
|
706 |
-
{
|
707 |
-
"epoch": 1.61,
|
708 |
-
"learning_rate": 1.993582036030978e-05,
|
709 |
-
"loss": 0.0002,
|
710 |
-
"step": 117
|
711 |
-
},
|
712 |
-
{
|
713 |
-
"epoch": 1.62,
|
714 |
-
"learning_rate": 1.9933769011050382e-05,
|
715 |
-
"loss": 0.0001,
|
716 |
-
"step": 118
|
717 |
-
},
|
718 |
-
{
|
719 |
-
"epoch": 1.64,
|
720 |
-
"learning_rate": 1.9931685501112183e-05,
|
721 |
-
"loss": 0.0006,
|
722 |
-
"step": 119
|
723 |
-
},
|
724 |
-
{
|
725 |
-
"epoch": 1.65,
|
726 |
-
"learning_rate": 1.9929569837240567e-05,
|
727 |
-
"loss": 0.0013,
|
728 |
-
"step": 120
|
729 |
-
},
|
730 |
-
{
|
731 |
-
"epoch": 1.66,
|
732 |
-
"learning_rate": 1.9927422026285016e-05,
|
733 |
-
"loss": 0.0001,
|
734 |
-
"step": 121
|
735 |
-
},
|
736 |
-
{
|
737 |
-
"epoch": 1.68,
|
738 |
-
"learning_rate": 1.992524207519909e-05,
|
739 |
-
"loss": 0.0002,
|
740 |
-
"step": 122
|
741 |
-
},
|
742 |
-
{
|
743 |
-
"epoch": 1.69,
|
744 |
-
"learning_rate": 1.9923029991040405e-05,
|
745 |
-
"loss": 0.0001,
|
746 |
-
"step": 123
|
747 |
-
},
|
748 |
-
{
|
749 |
-
"epoch": 1.7,
|
750 |
-
"learning_rate": 1.9920785780970604e-05,
|
751 |
-
"loss": 0.0008,
|
752 |
-
"step": 124
|
753 |
-
},
|
754 |
-
{
|
755 |
-
"epoch": 1.72,
|
756 |
-
"learning_rate": 1.991850945225534e-05,
|
757 |
-
"loss": 0.003,
|
758 |
-
"step": 125
|
759 |
-
},
|
760 |
-
{
|
761 |
-
"epoch": 1.73,
|
762 |
-
"learning_rate": 1.9916201012264255e-05,
|
763 |
-
"loss": 0.0176,
|
764 |
-
"step": 126
|
765 |
-
},
|
766 |
-
{
|
767 |
-
"epoch": 1.75,
|
768 |
-
"learning_rate": 1.9913860468470938e-05,
|
769 |
-
"loss": 0.0,
|
770 |
-
"step": 127
|
771 |
-
},
|
772 |
-
{
|
773 |
-
"epoch": 1.76,
|
774 |
-
"learning_rate": 1.9911487828452932e-05,
|
775 |
-
"loss": 0.0079,
|
776 |
-
"step": 128
|
777 |
-
},
|
778 |
-
{
|
779 |
-
"epoch": 1.77,
|
780 |
-
"learning_rate": 1.9909083099891682e-05,
|
781 |
-
"loss": 0.0004,
|
782 |
-
"step": 129
|
783 |
-
},
|
784 |
-
{
|
785 |
-
"epoch": 1.79,
|
786 |
-
"learning_rate": 1.9906646290572514e-05,
|
787 |
-
"loss": 0.0004,
|
788 |
-
"step": 130
|
789 |
-
},
|
790 |
-
{
|
791 |
-
"epoch": 1.8,
|
792 |
-
"learning_rate": 1.9904177408384635e-05,
|
793 |
-
"loss": 0.0003,
|
794 |
-
"step": 131
|
795 |
-
},
|
796 |
-
{
|
797 |
-
"epoch": 1.81,
|
798 |
-
"learning_rate": 1.990167646132107e-05,
|
799 |
-
"loss": 0.0021,
|
800 |
-
"step": 132
|
801 |
-
},
|
802 |
-
{
|
803 |
-
"epoch": 1.83,
|
804 |
-
"learning_rate": 1.9899143457478658e-05,
|
805 |
-
"loss": 0.0132,
|
806 |
-
"step": 133
|
807 |
-
},
|
808 |
-
{
|
809 |
-
"epoch": 1.84,
|
810 |
-
"learning_rate": 1.9896578405058028e-05,
|
811 |
-
"loss": 0.0022,
|
812 |
-
"step": 134
|
813 |
-
},
|
814 |
-
{
|
815 |
-
"epoch": 1.86,
|
816 |
-
"learning_rate": 1.9893981312363563e-05,
|
817 |
-
"loss": 0.0002,
|
818 |
-
"step": 135
|
819 |
-
},
|
820 |
-
{
|
821 |
-
"epoch": 1.87,
|
822 |
-
"learning_rate": 1.9893981312363563e-05,
|
823 |
-
"loss": 0.0132,
|
824 |
-
"step": 136
|
825 |
-
},
|
826 |
-
{
|
827 |
-
"epoch": 1.88,
|
828 |
-
"learning_rate": 1.9891352187803375e-05,
|
829 |
-
"loss": 0.0008,
|
830 |
-
"step": 137
|
831 |
-
},
|
832 |
-
{
|
833 |
-
"epoch": 1.9,
|
834 |
-
"learning_rate": 1.988869103988929e-05,
|
835 |
-
"loss": 0.0077,
|
836 |
-
"step": 138
|
837 |
-
},
|
838 |
-
{
|
839 |
-
"epoch": 1.91,
|
840 |
-
"learning_rate": 1.9885997877236788e-05,
|
841 |
-
"loss": 0.0003,
|
842 |
-
"step": 139
|
843 |
-
},
|
844 |
-
{
|
845 |
-
"epoch": 1.92,
|
846 |
-
"learning_rate": 1.9883272708565022e-05,
|
847 |
-
"loss": 0.0019,
|
848 |
-
"step": 140
|
849 |
-
},
|
850 |
-
{
|
851 |
-
"epoch": 1.94,
|
852 |
-
"learning_rate": 1.988051554269675e-05,
|
853 |
-
"loss": 0.0115,
|
854 |
-
"step": 141
|
855 |
-
},
|
856 |
-
{
|
857 |
-
"epoch": 1.95,
|
858 |
-
"learning_rate": 1.9877726388558325e-05,
|
859 |
-
"loss": 0.0012,
|
860 |
-
"step": 142
|
861 |
-
},
|
862 |
-
{
|
863 |
-
"epoch": 1.97,
|
864 |
-
"learning_rate": 1.987490525517966e-05,
|
865 |
-
"loss": 0.0023,
|
866 |
-
"step": 143
|
867 |
-
},
|
868 |
-
{
|
869 |
-
"epoch": 1.98,
|
870 |
-
"learning_rate": 1.9872052151694203e-05,
|
871 |
-
"loss": 0.0006,
|
872 |
-
"step": 144
|
873 |
-
},
|
874 |
-
{
|
875 |
-
"epoch": 1.99,
|
876 |
-
"learning_rate": 1.9869167087338908e-05,
|
877 |
-
"loss": 0.0011,
|
878 |
-
"step": 145
|
879 |
-
},
|
880 |
-
{
|
881 |
-
"epoch": 2.01,
|
882 |
-
"learning_rate": 1.98662500714542e-05,
|
883 |
-
"loss": 0.0004,
|
884 |
-
"step": 146
|
885 |
-
},
|
886 |
-
{
|
887 |
-
"epoch": 2.02,
|
888 |
-
"learning_rate": 1.986330111348395e-05,
|
889 |
-
"loss": 0.0058,
|
890 |
-
"step": 147
|
891 |
-
},
|
892 |
-
{
|
893 |
-
"epoch": 2.03,
|
894 |
-
"learning_rate": 1.9860320222975435e-05,
|
895 |
-
"loss": 0.0004,
|
896 |
-
"step": 148
|
897 |
-
},
|
898 |
-
{
|
899 |
-
"epoch": 2.05,
|
900 |
-
"learning_rate": 1.985730740957932e-05,
|
901 |
-
"loss": 0.0001,
|
902 |
-
"step": 149
|
903 |
-
},
|
904 |
-
{
|
905 |
-
"epoch": 2.06,
|
906 |
-
"learning_rate": 1.985426268304962e-05,
|
907 |
-
"loss": 0.0153,
|
908 |
-
"step": 150
|
909 |
-
},
|
910 |
-
{
|
911 |
-
"epoch": 2.08,
|
912 |
-
"learning_rate": 1.9851186053243667e-05,
|
913 |
-
"loss": 0.0008,
|
914 |
-
"step": 151
|
915 |
-
},
|
916 |
-
{
|
917 |
-
"epoch": 2.09,
|
918 |
-
"learning_rate": 1.9848077530122083e-05,
|
919 |
-
"loss": 0.0,
|
920 |
-
"step": 152
|
921 |
-
},
|
922 |
-
{
|
923 |
-
"epoch": 2.1,
|
924 |
-
"learning_rate": 1.9844937123748744e-05,
|
925 |
-
"loss": 0.0026,
|
926 |
-
"step": 153
|
927 |
-
},
|
928 |
-
{
|
929 |
-
"epoch": 2.12,
|
930 |
-
"learning_rate": 1.9841764844290744e-05,
|
931 |
-
"loss": 0.0002,
|
932 |
-
"step": 154
|
933 |
-
},
|
934 |
-
{
|
935 |
-
"epoch": 2.13,
|
936 |
-
"learning_rate": 1.9838560702018373e-05,
|
937 |
-
"loss": 0.0002,
|
938 |
-
"step": 155
|
939 |
-
},
|
940 |
-
{
|
941 |
-
"epoch": 2.14,
|
942 |
-
"learning_rate": 1.9835324707305076e-05,
|
943 |
-
"loss": 0.0108,
|
944 |
-
"step": 156
|
945 |
-
},
|
946 |
-
{
|
947 |
-
"epoch": 2.16,
|
948 |
-
"learning_rate": 1.983205687062742e-05,
|
949 |
-
"loss": 0.0001,
|
950 |
-
"step": 157
|
951 |
-
},
|
952 |
-
{
|
953 |
-
"epoch": 2.17,
|
954 |
-
"learning_rate": 1.982875720256505e-05,
|
955 |
-
"loss": 0.0001,
|
956 |
-
"step": 158
|
957 |
-
},
|
958 |
-
{
|
959 |
-
"epoch": 2.19,
|
960 |
-
"learning_rate": 1.982542571380069e-05,
|
961 |
-
"loss": 0.0011,
|
962 |
-
"step": 159
|
963 |
-
},
|
964 |
-
{
|
965 |
-
"epoch": 2.2,
|
966 |
-
"learning_rate": 1.9822062415120053e-05,
|
967 |
-
"loss": 0.0001,
|
968 |
-
"step": 160
|
969 |
-
},
|
970 |
-
{
|
971 |
-
"epoch": 2.21,
|
972 |
-
"learning_rate": 1.9818667317411865e-05,
|
973 |
-
"loss": 0.0001,
|
974 |
-
"step": 161
|
975 |
-
},
|
976 |
-
{
|
977 |
-
"epoch": 2.23,
|
978 |
-
"learning_rate": 1.9815240431667786e-05,
|
979 |
-
"loss": 0.0004,
|
980 |
-
"step": 162
|
981 |
-
},
|
982 |
-
{
|
983 |
-
"epoch": 2.24,
|
984 |
-
"learning_rate": 1.9811781768982392e-05,
|
985 |
-
"loss": 0.0001,
|
986 |
-
"step": 163
|
987 |
-
},
|
988 |
-
{
|
989 |
-
"epoch": 2.25,
|
990 |
-
"learning_rate": 1.980829134055314e-05,
|
991 |
-
"loss": 0.0006,
|
992 |
-
"step": 164
|
993 |
-
},
|
994 |
-
{
|
995 |
-
"epoch": 2.27,
|
996 |
-
"learning_rate": 1.980476915768033e-05,
|
997 |
-
"loss": 0.0,
|
998 |
-
"step": 165
|
999 |
-
},
|
1000 |
-
{
|
1001 |
-
"epoch": 2.28,
|
1002 |
-
"learning_rate": 1.9801215231767056e-05,
|
1003 |
-
"loss": 0.0001,
|
1004 |
-
"step": 166
|
1005 |
-
},
|
1006 |
-
{
|
1007 |
-
"epoch": 2.3,
|
1008 |
-
"learning_rate": 1.9797629574319202e-05,
|
1009 |
-
"loss": 0.0313,
|
1010 |
-
"step": 167
|
1011 |
-
},
|
1012 |
-
{
|
1013 |
-
"epoch": 2.31,
|
1014 |
-
"learning_rate": 1.9794012196945362e-05,
|
1015 |
-
"loss": 0.0,
|
1016 |
-
"step": 168
|
1017 |
-
},
|
1018 |
-
{
|
1019 |
-
"epoch": 2.32,
|
1020 |
-
"learning_rate": 1.9790363111356838e-05,
|
1021 |
-
"loss": 0.0,
|
1022 |
-
"step": 169
|
1023 |
-
},
|
1024 |
-
{
|
1025 |
-
"epoch": 2.34,
|
1026 |
-
"learning_rate": 1.9786682329367578e-05,
|
1027 |
-
"loss": 0.0,
|
1028 |
-
"step": 170
|
1029 |
-
},
|
1030 |
-
{
|
1031 |
-
"epoch": 2.35,
|
1032 |
-
"learning_rate": 1.978296986289415e-05,
|
1033 |
-
"loss": 0.0001,
|
1034 |
-
"step": 171
|
1035 |
-
},
|
1036 |
-
{
|
1037 |
-
"epoch": 2.36,
|
1038 |
-
"learning_rate": 1.977922572395571e-05,
|
1039 |
-
"loss": 0.0128,
|
1040 |
-
"step": 172
|
1041 |
-
},
|
1042 |
-
{
|
1043 |
-
"epoch": 2.38,
|
1044 |
-
"learning_rate": 1.9775449924673937e-05,
|
1045 |
-
"loss": 0.0003,
|
1046 |
-
"step": 173
|
1047 |
-
},
|
1048 |
-
{
|
1049 |
-
"epoch": 2.39,
|
1050 |
-
"learning_rate": 1.977164247727302e-05,
|
1051 |
-
"loss": 0.0003,
|
1052 |
-
"step": 174
|
1053 |
-
},
|
1054 |
-
{
|
1055 |
-
"epoch": 2.41,
|
1056 |
-
"learning_rate": 1.9767803394079618e-05,
|
1057 |
-
"loss": 0.0002,
|
1058 |
-
"step": 175
|
1059 |
-
},
|
1060 |
-
{
|
1061 |
-
"epoch": 2.42,
|
1062 |
-
"learning_rate": 1.9763932687522796e-05,
|
1063 |
-
"loss": 0.0045,
|
1064 |
-
"step": 176
|
1065 |
-
},
|
1066 |
-
{
|
1067 |
-
"epoch": 2.43,
|
1068 |
-
"learning_rate": 1.9760030370134004e-05,
|
1069 |
-
"loss": 0.0012,
|
1070 |
-
"step": 177
|
1071 |
-
},
|
1072 |
-
{
|
1073 |
-
"epoch": 2.45,
|
1074 |
-
"learning_rate": 1.975609645454704e-05,
|
1075 |
-
"loss": 0.0001,
|
1076 |
-
"step": 178
|
1077 |
-
},
|
1078 |
-
{
|
1079 |
-
"epoch": 2.46,
|
1080 |
-
"learning_rate": 1.975213095349799e-05,
|
1081 |
-
"loss": 0.0001,
|
1082 |
-
"step": 179
|
1083 |
-
},
|
1084 |
-
{
|
1085 |
-
"epoch": 2.47,
|
1086 |
-
"learning_rate": 1.974813387982521e-05,
|
1087 |
-
"loss": 0.0,
|
1088 |
-
"step": 180
|
1089 |
-
},
|
1090 |
-
{
|
1091 |
-
"epoch": 2.49,
|
1092 |
-
"learning_rate": 1.9744105246469264e-05,
|
1093 |
-
"loss": 0.0001,
|
1094 |
-
"step": 181
|
1095 |
-
},
|
1096 |
-
{
|
1097 |
-
"epoch": 2.5,
|
1098 |
-
"learning_rate": 1.974004506647289e-05,
|
1099 |
-
"loss": 0.0097,
|
1100 |
-
"step": 182
|
1101 |
-
},
|
1102 |
-
{
|
1103 |
-
"epoch": 2.52,
|
1104 |
-
"learning_rate": 1.973595335298097e-05,
|
1105 |
-
"loss": 0.0001,
|
1106 |
-
"step": 183
|
1107 |
-
},
|
1108 |
-
{
|
1109 |
-
"epoch": 2.53,
|
1110 |
-
"learning_rate": 1.9731830119240465e-05,
|
1111 |
-
"loss": 0.0008,
|
1112 |
-
"step": 184
|
1113 |
-
},
|
1114 |
-
{
|
1115 |
-
"epoch": 2.54,
|
1116 |
-
"learning_rate": 1.9727675378600382e-05,
|
1117 |
-
"loss": 0.0002,
|
1118 |
-
"step": 185
|
1119 |
-
},
|
1120 |
-
{
|
1121 |
-
"epoch": 2.56,
|
1122 |
-
"learning_rate": 1.9723489144511744e-05,
|
1123 |
-
"loss": 0.0002,
|
1124 |
-
"step": 186
|
1125 |
-
},
|
1126 |
-
{
|
1127 |
-
"epoch": 2.57,
|
1128 |
-
"learning_rate": 1.971927143052752e-05,
|
1129 |
-
"loss": 0.0002,
|
1130 |
-
"step": 187
|
1131 |
-
},
|
1132 |
-
{
|
1133 |
-
"epoch": 2.58,
|
1134 |
-
"learning_rate": 1.971502225030261e-05,
|
1135 |
-
"loss": 0.0023,
|
1136 |
-
"step": 188
|
1137 |
-
},
|
1138 |
-
{
|
1139 |
-
"epoch": 2.6,
|
1140 |
-
"learning_rate": 1.9710741617593768e-05,
|
1141 |
-
"loss": 0.0003,
|
1142 |
-
"step": 189
|
1143 |
-
},
|
1144 |
-
{
|
1145 |
-
"epoch": 2.61,
|
1146 |
-
"learning_rate": 1.9706429546259592e-05,
|
1147 |
-
"loss": 0.0008,
|
1148 |
-
"step": 190
|
1149 |
-
},
|
1150 |
-
{
|
1151 |
-
"epoch": 2.63,
|
1152 |
-
"learning_rate": 1.970208605026046e-05,
|
1153 |
-
"loss": 0.0003,
|
1154 |
-
"step": 191
|
1155 |
-
},
|
1156 |
-
{
|
1157 |
-
"epoch": 2.64,
|
1158 |
-
"learning_rate": 1.969771114365848e-05,
|
1159 |
-
"loss": 0.0,
|
1160 |
-
"step": 192
|
1161 |
-
},
|
1162 |
-
{
|
1163 |
-
"epoch": 2.65,
|
1164 |
-
"learning_rate": 1.9693304840617456e-05,
|
1165 |
-
"loss": 0.0001,
|
1166 |
-
"step": 193
|
1167 |
-
},
|
1168 |
-
{
|
1169 |
-
"epoch": 2.67,
|
1170 |
-
"learning_rate": 1.9688867155402845e-05,
|
1171 |
-
"loss": 0.0,
|
1172 |
-
"step": 194
|
1173 |
-
},
|
1174 |
-
{
|
1175 |
-
"epoch": 2.68,
|
1176 |
-
"learning_rate": 1.9684398102381694e-05,
|
1177 |
-
"loss": 0.0057,
|
1178 |
-
"step": 195
|
1179 |
-
},
|
1180 |
-
{
|
1181 |
-
"epoch": 2.69,
|
1182 |
-
"learning_rate": 1.967989769602261e-05,
|
1183 |
-
"loss": 0.0001,
|
1184 |
-
"step": 196
|
1185 |
-
},
|
1186 |
-
{
|
1187 |
-
"epoch": 2.71,
|
1188 |
-
"learning_rate": 1.9675365950895703e-05,
|
1189 |
-
"loss": 0.0001,
|
1190 |
-
"step": 197
|
1191 |
-
},
|
1192 |
-
{
|
1193 |
-
"epoch": 2.72,
|
1194 |
-
"learning_rate": 1.9670802881672543e-05,
|
1195 |
-
"loss": 0.0017,
|
1196 |
-
"step": 198
|
1197 |
-
},
|
1198 |
-
{
|
1199 |
-
"epoch": 2.74,
|
1200 |
-
"learning_rate": 1.9666208503126115e-05,
|
1201 |
-
"loss": 0.0003,
|
1202 |
-
"step": 199
|
1203 |
-
},
|
1204 |
-
{
|
1205 |
-
"epoch": 2.75,
|
1206 |
-
"learning_rate": 1.9661582830130766e-05,
|
1207 |
-
"loss": 0.0009,
|
1208 |
-
"step": 200
|
1209 |
-
},
|
1210 |
-
{
|
1211 |
-
"epoch": 2.76,
|
1212 |
-
"learning_rate": 1.965692587766216e-05,
|
1213 |
-
"loss": 0.0003,
|
1214 |
-
"step": 201
|
1215 |
-
},
|
1216 |
-
{
|
1217 |
-
"epoch": 2.78,
|
1218 |
-
"learning_rate": 1.965223766079723e-05,
|
1219 |
-
"loss": 0.0264,
|
1220 |
-
"step": 202
|
1221 |
-
},
|
1222 |
-
{
|
1223 |
-
"epoch": 2.79,
|
1224 |
-
"learning_rate": 1.9647518194714125e-05,
|
1225 |
-
"loss": 0.0,
|
1226 |
-
"step": 203
|
1227 |
-
},
|
1228 |
-
{
|
1229 |
-
"epoch": 2.8,
|
1230 |
-
"learning_rate": 1.964276749469216e-05,
|
1231 |
-
"loss": 0.0001,
|
1232 |
-
"step": 204
|
1233 |
-
},
|
1234 |
-
{
|
1235 |
-
"epoch": 2.82,
|
1236 |
-
"learning_rate": 1.963798557611178e-05,
|
1237 |
-
"loss": 0.0,
|
1238 |
-
"step": 205
|
1239 |
-
},
|
1240 |
-
{
|
1241 |
-
"epoch": 2.83,
|
1242 |
-
"learning_rate": 1.9633172454454498e-05,
|
1243 |
-
"loss": 0.0,
|
1244 |
-
"step": 206
|
1245 |
-
},
|
1246 |
-
{
|
1247 |
-
"epoch": 2.85,
|
1248 |
-
"learning_rate": 1.9628328145302837e-05,
|
1249 |
-
"loss": 0.0003,
|
1250 |
-
"step": 207
|
1251 |
-
},
|
1252 |
-
{
|
1253 |
-
"epoch": 2.86,
|
1254 |
-
"learning_rate": 1.9623452664340305e-05,
|
1255 |
-
"loss": 0.0,
|
1256 |
-
"step": 208
|
1257 |
-
},
|
1258 |
-
{
|
1259 |
-
"epoch": 2.87,
|
1260 |
-
"learning_rate": 1.9618546027351315e-05,
|
1261 |
-
"loss": 0.0,
|
1262 |
-
"step": 209
|
1263 |
-
},
|
1264 |
-
{
|
1265 |
-
"epoch": 2.89,
|
1266 |
-
"learning_rate": 1.961360825022116e-05,
|
1267 |
-
"loss": 0.0001,
|
1268 |
-
"step": 210
|
1269 |
-
},
|
1270 |
-
{
|
1271 |
-
"epoch": 2.9,
|
1272 |
-
"learning_rate": 1.9608639348935938e-05,
|
1273 |
-
"loss": 0.0001,
|
1274 |
-
"step": 211
|
1275 |
-
},
|
1276 |
-
{
|
1277 |
-
"epoch": 2.91,
|
1278 |
-
"learning_rate": 1.9603639339582522e-05,
|
1279 |
-
"loss": 0.0001,
|
1280 |
-
"step": 212
|
1281 |
-
},
|
1282 |
-
{
|
1283 |
-
"epoch": 2.93,
|
1284 |
-
"learning_rate": 1.9598608238348493e-05,
|
1285 |
-
"loss": 0.0002,
|
1286 |
-
"step": 213
|
1287 |
-
},
|
1288 |
-
{
|
1289 |
-
"epoch": 2.94,
|
1290 |
-
"learning_rate": 1.9593546061522094e-05,
|
1291 |
-
"loss": 0.0054,
|
1292 |
-
"step": 214
|
1293 |
-
},
|
1294 |
-
{
|
1295 |
-
"epoch": 2.96,
|
1296 |
-
"learning_rate": 1.958845282549217e-05,
|
1297 |
-
"loss": 0.0001,
|
1298 |
-
"step": 215
|
1299 |
-
},
|
1300 |
-
{
|
1301 |
-
"epoch": 2.97,
|
1302 |
-
"learning_rate": 1.958332854674813e-05,
|
1303 |
-
"loss": 0.0001,
|
1304 |
-
"step": 216
|
1305 |
-
},
|
1306 |
-
{
|
1307 |
-
"epoch": 2.98,
|
1308 |
-
"learning_rate": 1.957817324187987e-05,
|
1309 |
-
"loss": 0.0001,
|
1310 |
-
"step": 217
|
1311 |
-
},
|
1312 |
-
{
|
1313 |
-
"epoch": 3.0,
|
1314 |
-
"learning_rate": 1.9572986927577756e-05,
|
1315 |
-
"loss": 0.0091,
|
1316 |
-
"step": 218
|
1317 |
-
},
|
1318 |
-
{
|
1319 |
-
"epoch": 3.01,
|
1320 |
-
"learning_rate": 1.956776962063252e-05,
|
1321 |
-
"loss": 0.0011,
|
1322 |
-
"step": 219
|
1323 |
-
},
|
1324 |
-
{
|
1325 |
-
"epoch": 3.02,
|
1326 |
-
"learning_rate": 1.9562521337935255e-05,
|
1327 |
-
"loss": 0.0005,
|
1328 |
-
"step": 220
|
1329 |
-
},
|
1330 |
-
{
|
1331 |
-
"epoch": 3.04,
|
1332 |
-
"learning_rate": 1.955724209647733e-05,
|
1333 |
-
"loss": 0.0005,
|
1334 |
-
"step": 221
|
1335 |
-
},
|
1336 |
-
{
|
1337 |
-
"epoch": 3.05,
|
1338 |
-
"learning_rate": 1.9551931913350338e-05,
|
1339 |
-
"loss": 0.0002,
|
1340 |
-
"step": 222
|
1341 |
-
},
|
1342 |
-
{
|
1343 |
-
"epoch": 3.07,
|
1344 |
-
"learning_rate": 1.9546590805746054e-05,
|
1345 |
-
"loss": 0.0001,
|
1346 |
-
"step": 223
|
1347 |
-
},
|
1348 |
-
{
|
1349 |
-
"epoch": 3.08,
|
1350 |
-
"learning_rate": 1.954121879095637e-05,
|
1351 |
-
"loss": 0.0001,
|
1352 |
-
"step": 224
|
1353 |
-
},
|
1354 |
-
{
|
1355 |
-
"epoch": 3.09,
|
1356 |
-
"learning_rate": 1.953581588637323e-05,
|
1357 |
-
"loss": 0.0033,
|
1358 |
-
"step": 225
|
1359 |
-
},
|
1360 |
-
{
|
1361 |
-
"epoch": 3.11,
|
1362 |
-
"learning_rate": 1.953038210948861e-05,
|
1363 |
-
"loss": 0.0002,
|
1364 |
-
"step": 226
|
1365 |
-
},
|
1366 |
-
{
|
1367 |
-
"epoch": 3.12,
|
1368 |
-
"learning_rate": 1.9524917477894406e-05,
|
1369 |
-
"loss": 0.0006,
|
1370 |
-
"step": 227
|
1371 |
-
},
|
1372 |
-
{
|
1373 |
-
"epoch": 3.13,
|
1374 |
-
"learning_rate": 1.951942200928242e-05,
|
1375 |
-
"loss": 0.0003,
|
1376 |
-
"step": 228
|
1377 |
-
},
|
1378 |
-
{
|
1379 |
-
"epoch": 3.15,
|
1380 |
-
"learning_rate": 1.9513895721444286e-05,
|
1381 |
-
"loss": 0.0002,
|
1382 |
-
"step": 229
|
1383 |
-
},
|
1384 |
-
{
|
1385 |
-
"epoch": 3.16,
|
1386 |
-
"learning_rate": 1.9508338632271425e-05,
|
1387 |
-
"loss": 0.0081,
|
1388 |
-
"step": 230
|
1389 |
-
},
|
1390 |
-
{
|
1391 |
-
"epoch": 3.18,
|
1392 |
-
"learning_rate": 1.9502750759754963e-05,
|
1393 |
-
"loss": 0.0001,
|
1394 |
-
"step": 231
|
1395 |
-
},
|
1396 |
-
{
|
1397 |
-
"epoch": 3.19,
|
1398 |
-
"learning_rate": 1.9497132121985695e-05,
|
1399 |
-
"loss": 0.0002,
|
1400 |
-
"step": 232
|
1401 |
-
},
|
1402 |
-
{
|
1403 |
-
"epoch": 3.2,
|
1404 |
-
"learning_rate": 1.9491482737154023e-05,
|
1405 |
-
"loss": 0.0193,
|
1406 |
-
"step": 233
|
1407 |
-
},
|
1408 |
-
{
|
1409 |
-
"epoch": 3.22,
|
1410 |
-
"learning_rate": 1.948580262354988e-05,
|
1411 |
-
"loss": 0.0002,
|
1412 |
-
"step": 234
|
1413 |
-
},
|
1414 |
-
{
|
1415 |
-
"epoch": 3.23,
|
1416 |
-
"learning_rate": 1.9480091799562706e-05,
|
1417 |
-
"loss": 0.0033,
|
1418 |
-
"step": 235
|
1419 |
-
},
|
1420 |
-
{
|
1421 |
-
"epoch": 3.24,
|
1422 |
-
"learning_rate": 1.9474350283681337e-05,
|
1423 |
-
"loss": 0.0035,
|
1424 |
-
"step": 236
|
1425 |
-
},
|
1426 |
-
{
|
1427 |
-
"epoch": 3.26,
|
1428 |
-
"learning_rate": 1.9468578094493998e-05,
|
1429 |
-
"loss": 0.0127,
|
1430 |
-
"step": 237
|
1431 |
-
},
|
1432 |
-
{
|
1433 |
-
"epoch": 3.27,
|
1434 |
-
"learning_rate": 1.9462775250688208e-05,
|
1435 |
-
"loss": 0.0,
|
1436 |
-
"step": 238
|
1437 |
-
},
|
1438 |
-
{
|
1439 |
-
"epoch": 3.29,
|
1440 |
-
"learning_rate": 1.9456941771050733e-05,
|
1441 |
-
"loss": 0.0002,
|
1442 |
-
"step": 239
|
1443 |
-
},
|
1444 |
-
{
|
1445 |
-
"epoch": 3.3,
|
1446 |
-
"learning_rate": 1.9451077674467526e-05,
|
1447 |
-
"loss": 0.0014,
|
1448 |
-
"step": 240
|
1449 |
-
},
|
1450 |
-
{
|
1451 |
-
"epoch": 3.31,
|
1452 |
-
"learning_rate": 1.9445182979923657e-05,
|
1453 |
-
"loss": 0.0035,
|
1454 |
-
"step": 241
|
1455 |
-
},
|
1456 |
-
{
|
1457 |
-
"epoch": 3.33,
|
1458 |
-
"learning_rate": 1.9439257706503257e-05,
|
1459 |
-
"loss": 0.0002,
|
1460 |
-
"step": 242
|
1461 |
-
},
|
1462 |
-
{
|
1463 |
-
"epoch": 3.34,
|
1464 |
-
"learning_rate": 1.9433301873389463e-05,
|
1465 |
-
"loss": 0.0003,
|
1466 |
-
"step": 243
|
1467 |
-
},
|
1468 |
-
{
|
1469 |
-
"epoch": 3.35,
|
1470 |
-
"learning_rate": 1.9427315499864345e-05,
|
1471 |
-
"loss": 0.0009,
|
1472 |
-
"step": 244
|
1473 |
-
},
|
1474 |
-
{
|
1475 |
-
"epoch": 3.37,
|
1476 |
-
"learning_rate": 1.9421298605308847e-05,
|
1477 |
-
"loss": 0.0022,
|
1478 |
-
"step": 245
|
1479 |
-
},
|
1480 |
-
{
|
1481 |
-
"epoch": 3.38,
|
1482 |
-
"learning_rate": 1.941525120920273e-05,
|
1483 |
-
"loss": 0.0226,
|
1484 |
-
"step": 246
|
1485 |
-
},
|
1486 |
-
{
|
1487 |
-
"epoch": 3.4,
|
1488 |
-
"learning_rate": 1.94091733311245e-05,
|
1489 |
-
"loss": 0.0057,
|
1490 |
-
"step": 247
|
1491 |
-
},
|
1492 |
-
{
|
1493 |
-
"epoch": 3.41,
|
1494 |
-
"learning_rate": 1.9403064990751347e-05,
|
1495 |
-
"loss": 0.0025,
|
1496 |
-
"step": 248
|
1497 |
-
},
|
1498 |
-
{
|
1499 |
-
"epoch": 3.42,
|
1500 |
-
"learning_rate": 1.9396926207859085e-05,
|
1501 |
-
"loss": 0.0088,
|
1502 |
-
"step": 249
|
1503 |
-
},
|
1504 |
-
{
|
1505 |
-
"epoch": 3.44,
|
1506 |
-
"learning_rate": 1.939075700232209e-05,
|
1507 |
-
"loss": 0.0001,
|
1508 |
-
"step": 250
|
1509 |
-
},
|
1510 |
-
{
|
1511 |
-
"epoch": 3.45,
|
1512 |
-
"learning_rate": 1.938455739411323e-05,
|
1513 |
-
"loss": 0.0324,
|
1514 |
-
"step": 251
|
1515 |
-
},
|
1516 |
-
{
|
1517 |
-
"epoch": 3.46,
|
1518 |
-
"learning_rate": 1.937832740330379e-05,
|
1519 |
-
"loss": 0.019,
|
1520 |
-
"step": 252
|
1521 |
-
},
|
1522 |
-
{
|
1523 |
-
"epoch": 3.48,
|
1524 |
-
"learning_rate": 1.937206705006344e-05,
|
1525 |
-
"loss": 0.0001,
|
1526 |
-
"step": 253
|
1527 |
-
},
|
1528 |
-
{
|
1529 |
-
"epoch": 3.49,
|
1530 |
-
"learning_rate": 1.936577635466013e-05,
|
1531 |
-
"loss": 0.0364,
|
1532 |
-
"step": 254
|
1533 |
-
},
|
1534 |
-
{
|
1535 |
-
"epoch": 3.51,
|
1536 |
-
"learning_rate": 1.9359455337460054e-05,
|
1537 |
-
"loss": 0.0002,
|
1538 |
-
"step": 255
|
1539 |
-
},
|
1540 |
-
{
|
1541 |
-
"epoch": 3.52,
|
1542 |
-
"learning_rate": 1.9353104018927568e-05,
|
1543 |
-
"loss": 0.0009,
|
1544 |
-
"step": 256
|
1545 |
-
},
|
1546 |
-
{
|
1547 |
-
"epoch": 3.53,
|
1548 |
-
"learning_rate": 1.9346722419625136e-05,
|
1549 |
-
"loss": 0.0014,
|
1550 |
-
"step": 257
|
1551 |
-
},
|
1552 |
-
{
|
1553 |
-
"epoch": 3.55,
|
1554 |
-
"learning_rate": 1.934031056021325e-05,
|
1555 |
-
"loss": 0.02,
|
1556 |
-
"step": 258
|
1557 |
-
},
|
1558 |
-
{
|
1559 |
-
"epoch": 3.56,
|
1560 |
-
"learning_rate": 1.933386846145036e-05,
|
1561 |
-
"loss": 0.0063,
|
1562 |
-
"step": 259
|
1563 |
-
},
|
1564 |
-
{
|
1565 |
-
"epoch": 3.57,
|
1566 |
-
"learning_rate": 1.9327396144192837e-05,
|
1567 |
-
"loss": 0.0015,
|
1568 |
-
"step": 260
|
1569 |
-
},
|
1570 |
-
{
|
1571 |
-
"epoch": 3.59,
|
1572 |
-
"learning_rate": 1.9320893629394876e-05,
|
1573 |
-
"loss": 0.0016,
|
1574 |
-
"step": 261
|
1575 |
-
},
|
1576 |
-
{
|
1577 |
-
"epoch": 3.6,
|
1578 |
-
"learning_rate": 1.9314360938108427e-05,
|
1579 |
-
"loss": 0.0008,
|
1580 |
-
"step": 262
|
1581 |
-
},
|
1582 |
-
{
|
1583 |
-
"epoch": 3.62,
|
1584 |
-
"learning_rate": 1.9307798091483146e-05,
|
1585 |
-
"loss": 0.0272,
|
1586 |
-
"step": 263
|
1587 |
-
},
|
1588 |
-
{
|
1589 |
-
"epoch": 3.63,
|
1590 |
-
"learning_rate": 1.9301205110766323e-05,
|
1591 |
-
"loss": 0.0018,
|
1592 |
-
"step": 264
|
1593 |
-
},
|
1594 |
-
{
|
1595 |
-
"epoch": 3.64,
|
1596 |
-
"learning_rate": 1.9294582017302797e-05,
|
1597 |
-
"loss": 0.0168,
|
1598 |
-
"step": 265
|
1599 |
-
},
|
1600 |
-
{
|
1601 |
-
"epoch": 3.66,
|
1602 |
-
"learning_rate": 1.9287928832534896e-05,
|
1603 |
-
"loss": 0.0009,
|
1604 |
-
"step": 266
|
1605 |
-
},
|
1606 |
-
{
|
1607 |
-
"epoch": 3.67,
|
1608 |
-
"learning_rate": 1.9281245578002382e-05,
|
1609 |
-
"loss": 0.0006,
|
1610 |
-
"step": 267
|
1611 |
-
},
|
1612 |
-
{
|
1613 |
-
"epoch": 3.68,
|
1614 |
-
"learning_rate": 1.9274532275342355e-05,
|
1615 |
-
"loss": 0.0005,
|
1616 |
-
"step": 268
|
1617 |
-
},
|
1618 |
-
{
|
1619 |
-
"epoch": 3.7,
|
1620 |
-
"learning_rate": 1.9267788946289202e-05,
|
1621 |
-
"loss": 0.0008,
|
1622 |
-
"step": 269
|
1623 |
-
},
|
1624 |
-
{
|
1625 |
-
"epoch": 3.71,
|
1626 |
-
"learning_rate": 1.9261015612674518e-05,
|
1627 |
-
"loss": 0.0012,
|
1628 |
-
"step": 270
|
1629 |
-
},
|
1630 |
-
{
|
1631 |
-
"epoch": 3.73,
|
1632 |
-
"learning_rate": 1.9254212296427043e-05,
|
1633 |
-
"loss": 0.0006,
|
1634 |
-
"step": 271
|
1635 |
-
},
|
1636 |
-
{
|
1637 |
-
"epoch": 3.74,
|
1638 |
-
"learning_rate": 1.9247379019572584e-05,
|
1639 |
-
"loss": 0.0008,
|
1640 |
-
"step": 272
|
1641 |
-
},
|
1642 |
-
{
|
1643 |
-
"epoch": 3.75,
|
1644 |
-
"learning_rate": 1.924051580423394e-05,
|
1645 |
-
"loss": 0.0004,
|
1646 |
-
"step": 273
|
1647 |
-
},
|
1648 |
-
{
|
1649 |
-
"epoch": 3.77,
|
1650 |
-
"learning_rate": 1.923362267263084e-05,
|
1651 |
-
"loss": 0.0003,
|
1652 |
-
"step": 274
|
1653 |
-
},
|
1654 |
-
{
|
1655 |
-
"epoch": 3.78,
|
1656 |
-
"learning_rate": 1.9226699647079874e-05,
|
1657 |
-
"loss": 0.0013,
|
1658 |
-
"step": 275
|
1659 |
-
},
|
1660 |
-
{
|
1661 |
-
"epoch": 3.79,
|
1662 |
-
"learning_rate": 1.9219746749994404e-05,
|
1663 |
-
"loss": 0.0209,
|
1664 |
-
"step": 276
|
1665 |
-
},
|
1666 |
-
{
|
1667 |
-
"epoch": 3.81,
|
1668 |
-
"learning_rate": 1.921276400388451e-05,
|
1669 |
-
"loss": 0.0002,
|
1670 |
-
"step": 277
|
1671 |
-
},
|
1672 |
-
{
|
1673 |
-
"epoch": 3.82,
|
1674 |
-
"learning_rate": 1.92057514313569e-05,
|
1675 |
-
"loss": 0.0073,
|
1676 |
-
"step": 278
|
1677 |
-
},
|
1678 |
-
{
|
1679 |
-
"epoch": 3.84,
|
1680 |
-
"learning_rate": 1.919870905511485e-05,
|
1681 |
-
"loss": 0.0056,
|
1682 |
-
"step": 279
|
1683 |
-
},
|
1684 |
-
{
|
1685 |
-
"epoch": 3.85,
|
1686 |
-
"learning_rate": 1.9191636897958123e-05,
|
1687 |
-
"loss": 0.0057,
|
1688 |
-
"step": 280
|
1689 |
-
},
|
1690 |
-
{
|
1691 |
-
"epoch": 3.86,
|
1692 |
-
"learning_rate": 1.91845349827829e-05,
|
1693 |
-
"loss": 0.0146,
|
1694 |
-
"step": 281
|
1695 |
-
},
|
1696 |
-
{
|
1697 |
-
"epoch": 3.88,
|
1698 |
-
"learning_rate": 1.9177403332581715e-05,
|
1699 |
-
"loss": 0.0,
|
1700 |
-
"step": 282
|
1701 |
-
},
|
1702 |
-
{
|
1703 |
-
"epoch": 3.89,
|
1704 |
-
"learning_rate": 1.9170241970443344e-05,
|
1705 |
-
"loss": 0.0003,
|
1706 |
-
"step": 283
|
1707 |
-
},
|
1708 |
-
{
|
1709 |
-
"epoch": 3.9,
|
1710 |
-
"learning_rate": 1.9163050919552777e-05,
|
1711 |
-
"loss": 0.0017,
|
1712 |
-
"step": 284
|
1713 |
-
},
|
1714 |
-
{
|
1715 |
-
"epoch": 3.92,
|
1716 |
-
"learning_rate": 1.915583020319111e-05,
|
1717 |
-
"loss": 0.0015,
|
1718 |
-
"step": 285
|
1719 |
-
},
|
1720 |
-
{
|
1721 |
-
"epoch": 3.93,
|
1722 |
-
"learning_rate": 1.9148579844735497e-05,
|
1723 |
-
"loss": 0.0082,
|
1724 |
-
"step": 286
|
1725 |
-
},
|
1726 |
-
{
|
1727 |
-
"epoch": 3.95,
|
1728 |
-
"learning_rate": 1.9141299867659036e-05,
|
1729 |
-
"loss": 0.0002,
|
1730 |
-
"step": 287
|
1731 |
-
},
|
1732 |
-
{
|
1733 |
-
"epoch": 3.96,
|
1734 |
-
"learning_rate": 1.9133990295530737e-05,
|
1735 |
-
"loss": 0.0021,
|
1736 |
-
"step": 288
|
1737 |
-
},
|
1738 |
-
{
|
1739 |
-
"epoch": 3.97,
|
1740 |
-
"learning_rate": 1.9126651152015404e-05,
|
1741 |
-
"loss": 0.0001,
|
1742 |
-
"step": 289
|
1743 |
-
},
|
1744 |
-
{
|
1745 |
-
"epoch": 3.99,
|
1746 |
-
"learning_rate": 1.9119282460873597e-05,
|
1747 |
-
"loss": 0.0001,
|
1748 |
-
"step": 290
|
1749 |
-
},
|
1750 |
-
{
|
1751 |
-
"epoch": 4.0,
|
1752 |
-
"learning_rate": 1.9111884245961524e-05,
|
1753 |
-
"loss": 0.0002,
|
1754 |
-
"step": 291
|
1755 |
-
},
|
1756 |
-
{
|
1757 |
-
"epoch": 4.01,
|
1758 |
-
"learning_rate": 1.9104456531230986e-05,
|
1759 |
-
"loss": 0.0002,
|
1760 |
-
"step": 292
|
1761 |
-
},
|
1762 |
-
{
|
1763 |
-
"epoch": 4.03,
|
1764 |
-
"learning_rate": 1.9096999340729284e-05,
|
1765 |
-
"loss": 0.0002,
|
1766 |
-
"step": 293
|
1767 |
-
},
|
1768 |
-
{
|
1769 |
-
"epoch": 4.04,
|
1770 |
-
"learning_rate": 1.9089512698599148e-05,
|
1771 |
-
"loss": 0.0001,
|
1772 |
-
"step": 294
|
1773 |
-
},
|
1774 |
-
{
|
1775 |
-
"epoch": 4.05,
|
1776 |
-
"learning_rate": 1.9081996629078655e-05,
|
1777 |
-
"loss": 0.0219,
|
1778 |
-
"step": 295
|
1779 |
-
},
|
1780 |
-
{
|
1781 |
-
"epoch": 4.07,
|
1782 |
-
"learning_rate": 1.9074451156501163e-05,
|
1783 |
-
"loss": 0.0001,
|
1784 |
-
"step": 296
|
1785 |
-
},
|
1786 |
-
{
|
1787 |
-
"epoch": 4.08,
|
1788 |
-
"learning_rate": 1.9066876305295218e-05,
|
1789 |
-
"loss": 0.0003,
|
1790 |
-
"step": 297
|
1791 |
-
},
|
1792 |
-
{
|
1793 |
-
"epoch": 4.1,
|
1794 |
-
"learning_rate": 1.905927209998447e-05,
|
1795 |
-
"loss": 0.0,
|
1796 |
-
"step": 298
|
1797 |
-
},
|
1798 |
-
{
|
1799 |
-
"epoch": 4.11,
|
1800 |
-
"learning_rate": 1.905163856518762e-05,
|
1801 |
-
"loss": 0.0019,
|
1802 |
-
"step": 299
|
1803 |
-
},
|
1804 |
-
{
|
1805 |
-
"epoch": 4.12,
|
1806 |
-
"learning_rate": 1.9043975725618304e-05,
|
1807 |
-
"loss": 0.0068,
|
1808 |
-
"step": 300
|
1809 |
-
}
|
1810 |
-
],
|
1811 |
-
"max_steps": 1800,
|
1812 |
-
"num_train_epochs": 25,
|
1813 |
-
"total_flos": 9.93619813072896e+16,
|
1814 |
-
"trial_name": null,
|
1815 |
-
"trial_params": null
|
1816 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|