Upload folder using huggingface_hub
Browse files- adapter_config.json +1 -1
- adapter_model.safetensors +1 -1
- optimizer.pt +1 -1
- rng_state.pth +1 -1
- scheduler.pt +1 -1
- trainer_state.json +454 -2368
- training_args.bin +1 -1
adapter_config.json
CHANGED
@@ -16,9 +16,9 @@
|
|
16 |
"rank_pattern": {},
|
17 |
"revision": null,
|
18 |
"target_modules": [
|
19 |
-
"query_key_value",
|
20 |
"dense",
|
21 |
"dense_4h_to_h",
|
|
|
22 |
"dense_h_to_4h"
|
23 |
],
|
24 |
"task_type": "CAUSAL_LM"
|
|
|
16 |
"rank_pattern": {},
|
17 |
"revision": null,
|
18 |
"target_modules": [
|
|
|
19 |
"dense",
|
20 |
"dense_4h_to_h",
|
21 |
+
"query_key_value",
|
22 |
"dense_h_to_4h"
|
23 |
],
|
24 |
"task_type": "CAUSAL_LM"
|
adapter_model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 134235712
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:704c0d950400c00eaccd27d6f25ce2aa2b85faabebd9bff34883708dfe1aa444
|
3 |
size 134235712
|
optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 268515002
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:dc482a7746d4f942c0f8185ae2425fc1fdad4f9a39b7195fea307109263757ce
|
3 |
size 268515002
|
rng_state.pth
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 14244
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:62678783b8594ecb9abae27e23178e9faee14a19260c64e1ad6f70c4ee4c2c06
|
3 |
size 14244
|
scheduler.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 1064
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:74485e67705dc36efbfb69b1e54f842e1ff07894d01bb0e36d6d2526a318b300
|
3 |
size 1064
|
trainer_state.json
CHANGED
@@ -1,2833 +1,919 @@
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
-
"epoch":
|
5 |
"eval_steps": 500,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
10 |
"log_history": [
|
11 |
{
|
12 |
-
"epoch": 0.
|
13 |
-
"learning_rate": 4.
|
14 |
-
"loss": 2.
|
15 |
"step": 2
|
16 |
},
|
17 |
{
|
18 |
-
"epoch": 0.
|
19 |
-
"learning_rate":
|
20 |
-
"loss": 2.
|
21 |
"step": 4
|
22 |
},
|
23 |
{
|
24 |
-
"epoch": 0.
|
25 |
-
"learning_rate":
|
26 |
-
"loss": 2.
|
27 |
"step": 6
|
28 |
},
|
29 |
{
|
30 |
-
"epoch": 0.
|
31 |
-
"learning_rate":
|
32 |
-
"loss": 2.
|
33 |
"step": 8
|
34 |
},
|
35 |
{
|
36 |
-
"epoch": 0.
|
37 |
-
"learning_rate":
|
38 |
-
"loss": 2.
|
39 |
"step": 10
|
40 |
},
|
41 |
{
|
42 |
-
"epoch": 0.
|
43 |
-
"learning_rate":
|
44 |
-
"loss": 2.
|
45 |
"step": 12
|
46 |
},
|
47 |
{
|
48 |
-
"epoch": 0.
|
49 |
-
"learning_rate":
|
50 |
-
"loss": 2.
|
51 |
"step": 14
|
52 |
},
|
53 |
{
|
54 |
-
"epoch": 0.
|
55 |
-
"learning_rate":
|
56 |
-
"loss": 2.
|
57 |
"step": 16
|
58 |
},
|
59 |
{
|
60 |
-
"epoch": 0.
|
61 |
-
"learning_rate":
|
62 |
-
"loss": 2.
|
63 |
"step": 18
|
64 |
},
|
65 |
{
|
66 |
-
"epoch": 0.
|
67 |
-
"learning_rate":
|
68 |
-
"loss":
|
69 |
"step": 20
|
70 |
},
|
71 |
{
|
72 |
-
"epoch": 0.
|
73 |
-
"learning_rate":
|
74 |
-
"loss": 2.
|
75 |
"step": 22
|
76 |
},
|
77 |
{
|
78 |
-
"epoch": 0.
|
79 |
-
"learning_rate":
|
80 |
-
"loss": 2.
|
81 |
"step": 24
|
82 |
},
|
83 |
{
|
84 |
-
"epoch": 0.
|
85 |
-
"learning_rate":
|
86 |
-
"loss": 2.
|
87 |
"step": 26
|
88 |
},
|
89 |
{
|
90 |
-
"epoch": 0.
|
91 |
-
"learning_rate":
|
92 |
-
"loss": 2.
|
93 |
"step": 28
|
94 |
},
|
95 |
{
|
96 |
-
"epoch": 0.
|
97 |
-
"learning_rate":
|
98 |
-
"loss": 2.
|
99 |
"step": 30
|
100 |
},
|
101 |
{
|
102 |
-
"epoch": 0.
|
103 |
-
"learning_rate":
|
104 |
-
"loss": 2.
|
105 |
"step": 32
|
106 |
},
|
107 |
{
|
108 |
-
"epoch": 0.
|
109 |
-
"learning_rate":
|
110 |
-
"loss": 2.
|
111 |
"step": 34
|
112 |
},
|
113 |
{
|
114 |
-
"epoch": 0.
|
115 |
-
"learning_rate":
|
116 |
-
"loss": 2.
|
117 |
"step": 36
|
118 |
},
|
119 |
{
|
120 |
-
"epoch": 0.
|
121 |
-
"learning_rate":
|
122 |
-
"loss": 2.
|
123 |
"step": 38
|
124 |
},
|
125 |
{
|
126 |
-
"epoch": 0.
|
127 |
-
"learning_rate":
|
128 |
-
"loss": 2.
|
129 |
"step": 40
|
130 |
},
|
131 |
{
|
132 |
-
"epoch": 0.
|
133 |
-
"learning_rate":
|
134 |
-
"loss": 2.
|
135 |
"step": 42
|
136 |
},
|
137 |
{
|
138 |
-
"epoch": 0.
|
139 |
-
"learning_rate": 0.
|
140 |
-
"loss": 2.
|
141 |
"step": 44
|
142 |
},
|
143 |
{
|
144 |
-
"epoch": 0.
|
145 |
-
"learning_rate": 0.
|
146 |
-
"loss":
|
147 |
"step": 46
|
148 |
},
|
149 |
{
|
150 |
-
"epoch": 0.
|
151 |
-
"learning_rate": 0.
|
152 |
-
"loss": 2.
|
153 |
"step": 48
|
154 |
},
|
155 |
{
|
156 |
-
"epoch": 0.
|
157 |
-
"learning_rate": 0.
|
158 |
-
"loss":
|
159 |
"step": 50
|
160 |
},
|
161 |
{
|
162 |
-
"epoch": 0.
|
163 |
-
"learning_rate": 0.
|
164 |
-
"loss": 2.
|
165 |
"step": 52
|
166 |
},
|
167 |
{
|
168 |
-
"epoch": 0.
|
169 |
-
"learning_rate": 0.
|
170 |
-
"loss": 2.
|
171 |
"step": 54
|
172 |
},
|
173 |
{
|
174 |
-
"epoch": 0.
|
175 |
-
"learning_rate": 0.
|
176 |
-
"loss": 2.
|
177 |
"step": 56
|
178 |
},
|
179 |
{
|
180 |
-
"epoch": 0.
|
181 |
-
"learning_rate": 0.
|
182 |
-
"loss": 2.
|
183 |
"step": 58
|
184 |
},
|
185 |
{
|
186 |
-
"epoch": 0.
|
187 |
-
"learning_rate": 0.
|
188 |
-
"loss": 2.
|
189 |
"step": 60
|
190 |
},
|
191 |
{
|
192 |
-
"epoch": 0.
|
193 |
-
"learning_rate": 0.
|
194 |
-
"loss": 2.
|
195 |
"step": 62
|
196 |
},
|
197 |
{
|
198 |
-
"epoch": 0.
|
199 |
-
"learning_rate": 0.
|
200 |
-
"loss": 2.
|
201 |
"step": 64
|
202 |
},
|
203 |
{
|
204 |
-
"epoch": 0.
|
205 |
-
"learning_rate": 0.
|
206 |
-
"loss": 2.
|
207 |
"step": 66
|
208 |
},
|
209 |
{
|
210 |
-
"epoch": 0.
|
211 |
-
"learning_rate": 0.
|
212 |
-
"loss": 2.
|
213 |
"step": 68
|
214 |
},
|
215 |
{
|
216 |
-
"epoch": 0.
|
217 |
-
"learning_rate": 0.
|
218 |
-
"loss": 2.
|
219 |
"step": 70
|
220 |
},
|
221 |
{
|
222 |
-
"epoch": 0.
|
223 |
-
"learning_rate": 0.
|
224 |
-
"loss": 2.
|
225 |
"step": 72
|
226 |
},
|
227 |
{
|
228 |
-
"epoch": 0.
|
229 |
-
"learning_rate": 0.
|
230 |
-
"loss": 2.
|
231 |
"step": 74
|
232 |
},
|
233 |
{
|
234 |
-
"epoch": 0.
|
235 |
-
"learning_rate": 0.
|
236 |
-
"loss": 2.
|
237 |
"step": 76
|
238 |
},
|
239 |
{
|
240 |
-
"epoch": 0.
|
241 |
-
"learning_rate": 0.
|
242 |
-
"loss": 2.
|
243 |
"step": 78
|
244 |
},
|
245 |
{
|
246 |
-
"epoch": 0.
|
247 |
-
"learning_rate": 0.
|
248 |
-
"loss": 2.
|
249 |
"step": 80
|
250 |
},
|
251 |
{
|
252 |
-
"epoch": 0.
|
253 |
-
"learning_rate": 0.
|
254 |
-
"loss": 2.
|
255 |
"step": 82
|
256 |
},
|
257 |
{
|
258 |
-
"epoch": 0.
|
259 |
-
"learning_rate": 0.
|
260 |
-
"loss": 2.
|
261 |
"step": 84
|
262 |
},
|
263 |
{
|
264 |
-
"epoch": 0.
|
265 |
-
"learning_rate": 0.
|
266 |
-
"loss":
|
267 |
"step": 86
|
268 |
},
|
269 |
{
|
270 |
-
"epoch": 0.
|
271 |
-
"learning_rate": 0.
|
272 |
-
"loss": 2.
|
273 |
"step": 88
|
274 |
},
|
275 |
{
|
276 |
-
"epoch": 0.
|
277 |
-
"learning_rate": 0.
|
278 |
-
"loss": 2.
|
279 |
"step": 90
|
280 |
},
|
281 |
{
|
282 |
-
"epoch": 0.
|
283 |
-
"learning_rate": 0.
|
284 |
-
"loss":
|
285 |
"step": 92
|
286 |
},
|
287 |
{
|
288 |
-
"epoch": 0.
|
289 |
-
"learning_rate": 0.
|
290 |
-
"loss":
|
291 |
"step": 94
|
292 |
},
|
293 |
{
|
294 |
-
"epoch": 0.
|
295 |
-
"learning_rate": 0.
|
296 |
-
"loss": 2.
|
297 |
"step": 96
|
298 |
},
|
299 |
{
|
300 |
-
"epoch": 0.
|
301 |
-
"learning_rate": 0.
|
302 |
-
"loss":
|
303 |
"step": 98
|
304 |
},
|
305 |
{
|
306 |
-
"epoch": 0.
|
307 |
-
"learning_rate": 0.
|
308 |
-
"loss":
|
309 |
"step": 100
|
310 |
},
|
311 |
{
|
312 |
-
"epoch":
|
313 |
-
"learning_rate": 0.
|
314 |
-
"loss": 2.
|
315 |
"step": 102
|
316 |
},
|
317 |
{
|
318 |
-
"epoch":
|
319 |
-
"learning_rate": 0.
|
320 |
-
"loss": 2.
|
321 |
"step": 104
|
322 |
},
|
323 |
{
|
324 |
-
"epoch":
|
325 |
-
"learning_rate": 0.
|
326 |
-
"loss": 2.
|
327 |
"step": 106
|
328 |
},
|
329 |
{
|
330 |
-
"epoch":
|
331 |
-
"learning_rate": 0.
|
332 |
-
"loss": 2.
|
333 |
"step": 108
|
334 |
},
|
335 |
{
|
336 |
-
"epoch":
|
337 |
-
"learning_rate": 0.
|
338 |
-
"loss": 2.
|
339 |
"step": 110
|
340 |
},
|
341 |
{
|
342 |
-
"epoch":
|
343 |
-
"learning_rate": 0.
|
344 |
-
"loss": 2.
|
345 |
"step": 112
|
346 |
},
|
347 |
{
|
348 |
-
"epoch":
|
349 |
-
"learning_rate": 0.
|
350 |
-
"loss": 2.
|
351 |
"step": 114
|
352 |
},
|
353 |
{
|
354 |
-
"epoch":
|
355 |
-
"learning_rate": 0.
|
356 |
-
"loss": 2.
|
357 |
"step": 116
|
358 |
},
|
359 |
{
|
360 |
-
"epoch":
|
361 |
-
"learning_rate": 0.
|
362 |
-
"loss": 2.
|
363 |
"step": 118
|
364 |
},
|
365 |
{
|
366 |
-
"epoch":
|
367 |
-
"learning_rate": 0.
|
368 |
-
"loss": 2.
|
369 |
"step": 120
|
370 |
},
|
371 |
{
|
372 |
-
"epoch":
|
373 |
-
"learning_rate": 0.
|
374 |
-
"loss": 2.
|
375 |
"step": 122
|
376 |
},
|
377 |
{
|
378 |
-
"epoch":
|
379 |
-
"learning_rate": 0.
|
380 |
-
"loss":
|
381 |
"step": 124
|
382 |
},
|
383 |
{
|
384 |
-
"epoch":
|
385 |
-
"learning_rate": 0.
|
386 |
-
"loss": 2.
|
387 |
"step": 126
|
388 |
},
|
389 |
{
|
390 |
-
"epoch":
|
391 |
-
"learning_rate": 0.
|
392 |
-
"loss": 2.
|
393 |
"step": 128
|
394 |
},
|
395 |
{
|
396 |
-
"epoch":
|
397 |
-
"learning_rate": 0.
|
398 |
-
"loss": 2.
|
399 |
"step": 130
|
400 |
},
|
401 |
{
|
402 |
-
"epoch":
|
403 |
-
"learning_rate": 0.
|
404 |
-
"loss": 2.
|
405 |
"step": 132
|
406 |
},
|
407 |
{
|
408 |
-
"epoch":
|
409 |
-
"learning_rate": 0.
|
410 |
-
"loss":
|
411 |
"step": 134
|
412 |
},
|
413 |
{
|
414 |
-
"epoch":
|
415 |
-
"learning_rate": 0.
|
416 |
-
"loss":
|
417 |
"step": 136
|
418 |
},
|
419 |
{
|
420 |
-
"epoch":
|
421 |
-
"learning_rate": 0.
|
422 |
-
"loss":
|
423 |
"step": 138
|
424 |
},
|
425 |
{
|
426 |
-
"epoch":
|
427 |
-
"learning_rate": 0.
|
428 |
-
"loss": 2.
|
429 |
"step": 140
|
430 |
},
|
431 |
{
|
432 |
-
"epoch":
|
433 |
-
"learning_rate": 0.
|
434 |
-
"loss":
|
435 |
"step": 142
|
436 |
},
|
437 |
{
|
438 |
-
"epoch":
|
439 |
-
"learning_rate": 0.
|
440 |
-
"loss": 1.
|
441 |
"step": 144
|
442 |
},
|
443 |
{
|
444 |
-
"epoch":
|
445 |
-
"learning_rate": 0.
|
446 |
-
"loss": 1.
|
447 |
"step": 146
|
448 |
},
|
449 |
{
|
450 |
-
"epoch":
|
451 |
-
"learning_rate": 0.
|
452 |
-
"loss": 1.
|
453 |
"step": 148
|
454 |
},
|
455 |
{
|
456 |
-
"epoch":
|
457 |
-
"learning_rate": 0.
|
458 |
-
"loss":
|
459 |
"step": 150
|
460 |
},
|
461 |
{
|
462 |
-
"epoch":
|
463 |
-
"learning_rate": 0.
|
464 |
-
"loss":
|
465 |
"step": 152
|
466 |
},
|
467 |
{
|
468 |
-
"epoch":
|
469 |
-
"learning_rate": 0.
|
470 |
-
"loss": 2.
|
471 |
"step": 154
|
472 |
},
|
473 |
{
|
474 |
-
"epoch":
|
475 |
-
"learning_rate":
|
476 |
-
"loss": 2.
|
477 |
"step": 156
|
478 |
},
|
479 |
{
|
480 |
-
"epoch":
|
481 |
-
"learning_rate":
|
482 |
-
"loss": 2.
|
483 |
"step": 158
|
484 |
},
|
485 |
{
|
486 |
-
"epoch":
|
487 |
-
"learning_rate":
|
488 |
-
"loss": 2.
|
489 |
"step": 160
|
490 |
},
|
491 |
{
|
492 |
-
"epoch":
|
493 |
-
"learning_rate":
|
494 |
-
"loss": 2.
|
495 |
"step": 162
|
496 |
},
|
497 |
{
|
498 |
-
"epoch":
|
499 |
-
"learning_rate":
|
500 |
-
"loss": 2.
|
501 |
"step": 164
|
502 |
},
|
503 |
{
|
504 |
-
"epoch":
|
505 |
-
"learning_rate":
|
506 |
-
"loss": 2.
|
507 |
"step": 166
|
508 |
},
|
509 |
{
|
510 |
-
"epoch":
|
511 |
-
"learning_rate":
|
512 |
-
"loss": 2.
|
513 |
"step": 168
|
514 |
},
|
515 |
{
|
516 |
-
"epoch":
|
517 |
-
"learning_rate":
|
518 |
-
"loss":
|
519 |
"step": 170
|
520 |
},
|
521 |
{
|
522 |
-
"epoch":
|
523 |
-
"learning_rate":
|
524 |
-
"loss": 2.
|
525 |
"step": 172
|
526 |
},
|
527 |
{
|
528 |
-
"epoch":
|
529 |
-
"learning_rate":
|
530 |
-
"loss": 2.
|
531 |
"step": 174
|
532 |
},
|
533 |
{
|
534 |
-
"epoch":
|
535 |
-
"learning_rate":
|
536 |
-
"loss": 2.
|
537 |
"step": 176
|
538 |
},
|
539 |
{
|
540 |
-
"epoch":
|
541 |
-
"learning_rate":
|
542 |
-
"loss": 2.
|
543 |
"step": 178
|
544 |
},
|
545 |
{
|
546 |
-
"epoch":
|
547 |
-
"learning_rate":
|
548 |
-
"loss": 2.
|
549 |
"step": 180
|
550 |
},
|
551 |
{
|
552 |
-
"epoch":
|
553 |
-
"learning_rate":
|
554 |
-
"loss": 2.
|
555 |
"step": 182
|
556 |
},
|
557 |
{
|
558 |
-
"epoch":
|
559 |
-
"learning_rate":
|
560 |
-
"loss":
|
561 |
"step": 184
|
562 |
},
|
563 |
{
|
564 |
-
"epoch":
|
565 |
-
"learning_rate":
|
566 |
-
"loss":
|
567 |
"step": 186
|
568 |
},
|
569 |
{
|
570 |
-
"epoch":
|
571 |
-
"learning_rate":
|
572 |
-
"loss":
|
573 |
"step": 188
|
574 |
},
|
575 |
{
|
576 |
-
"epoch":
|
577 |
-
"learning_rate":
|
578 |
-
"loss":
|
579 |
"step": 190
|
580 |
},
|
581 |
{
|
582 |
-
"epoch":
|
583 |
-
"learning_rate":
|
584 |
-
"loss":
|
585 |
"step": 192
|
586 |
},
|
587 |
{
|
588 |
-
"epoch":
|
589 |
-
"learning_rate":
|
590 |
-
"loss": 1.
|
591 |
"step": 194
|
592 |
},
|
593 |
{
|
594 |
-
"epoch":
|
595 |
-
"learning_rate":
|
596 |
-
"loss":
|
597 |
"step": 196
|
598 |
},
|
599 |
{
|
600 |
-
"epoch":
|
601 |
-
"learning_rate":
|
602 |
-
"loss":
|
603 |
"step": 198
|
604 |
},
|
605 |
{
|
606 |
-
"epoch":
|
607 |
-
"learning_rate":
|
608 |
-
"loss": 1.
|
609 |
"step": 200
|
610 |
},
|
611 |
{
|
612 |
-
"epoch":
|
613 |
-
"learning_rate":
|
614 |
-
"loss":
|
615 |
"step": 202
|
616 |
},
|
617 |
{
|
618 |
-
"epoch":
|
619 |
-
"learning_rate":
|
620 |
-
"loss": 2.
|
621 |
"step": 204
|
622 |
},
|
623 |
{
|
624 |
-
"epoch":
|
625 |
-
"learning_rate":
|
626 |
-
"loss": 2.
|
627 |
"step": 206
|
628 |
},
|
629 |
{
|
630 |
-
"epoch":
|
631 |
-
"learning_rate":
|
632 |
-
"loss": 2.
|
633 |
"step": 208
|
634 |
},
|
635 |
{
|
636 |
-
"epoch":
|
637 |
-
"learning_rate":
|
638 |
-
"loss": 2.
|
639 |
"step": 210
|
640 |
},
|
641 |
{
|
642 |
-
"epoch":
|
643 |
-
"learning_rate":
|
644 |
-
"loss": 2.
|
645 |
"step": 212
|
646 |
},
|
647 |
{
|
648 |
-
"epoch":
|
649 |
-
"learning_rate":
|
650 |
-
"loss": 2.
|
651 |
"step": 214
|
652 |
},
|
653 |
{
|
654 |
-
"epoch":
|
655 |
-
"learning_rate":
|
656 |
-
"loss": 2.
|
657 |
"step": 216
|
658 |
},
|
659 |
{
|
660 |
-
"epoch":
|
661 |
-
"learning_rate":
|
662 |
-
"loss":
|
663 |
"step": 218
|
664 |
},
|
665 |
{
|
666 |
-
"epoch":
|
667 |
-
"learning_rate":
|
668 |
-
"loss":
|
669 |
"step": 220
|
670 |
},
|
671 |
{
|
672 |
-
"epoch":
|
673 |
-
"learning_rate":
|
674 |
-
"loss":
|
675 |
"step": 222
|
676 |
},
|
677 |
{
|
678 |
-
"epoch":
|
679 |
-
"learning_rate":
|
680 |
-
"loss":
|
681 |
"step": 224
|
682 |
},
|
683 |
{
|
684 |
-
"epoch":
|
685 |
-
"learning_rate":
|
686 |
-
"loss":
|
687 |
"step": 226
|
688 |
},
|
689 |
{
|
690 |
-
"epoch":
|
691 |
-
"learning_rate":
|
692 |
-
"loss": 2.
|
693 |
"step": 228
|
694 |
},
|
695 |
{
|
696 |
-
"epoch":
|
697 |
-
"learning_rate":
|
698 |
-
"loss":
|
699 |
"step": 230
|
700 |
},
|
701 |
{
|
702 |
-
"epoch":
|
703 |
-
"learning_rate":
|
704 |
-
"loss":
|
705 |
"step": 232
|
706 |
},
|
707 |
{
|
708 |
-
"epoch":
|
709 |
-
"learning_rate":
|
710 |
-
"loss":
|
711 |
"step": 234
|
712 |
},
|
713 |
{
|
714 |
-
"epoch":
|
715 |
-
"learning_rate":
|
716 |
-
"loss":
|
717 |
"step": 236
|
718 |
},
|
719 |
{
|
720 |
-
"epoch":
|
721 |
-
"learning_rate":
|
722 |
-
"loss":
|
723 |
"step": 238
|
724 |
},
|
725 |
{
|
726 |
-
"epoch":
|
727 |
-
"learning_rate":
|
728 |
-
"loss":
|
729 |
"step": 240
|
730 |
},
|
731 |
{
|
732 |
-
"epoch":
|
733 |
-
"learning_rate":
|
734 |
-
"loss":
|
735 |
"step": 242
|
736 |
},
|
737 |
{
|
738 |
-
"epoch":
|
739 |
-
"learning_rate":
|
740 |
-
"loss": 1.
|
741 |
"step": 244
|
742 |
},
|
743 |
{
|
744 |
-
"epoch":
|
745 |
-
"learning_rate":
|
746 |
-
"loss":
|
747 |
"step": 246
|
748 |
},
|
749 |
{
|
750 |
-
"epoch":
|
751 |
-
"learning_rate":
|
752 |
-
"loss": 1.
|
753 |
"step": 248
|
754 |
},
|
755 |
{
|
756 |
-
"epoch":
|
757 |
-
"learning_rate":
|
758 |
-
"loss": 1.
|
759 |
"step": 250
|
760 |
},
|
761 |
{
|
762 |
-
"epoch":
|
763 |
-
"learning_rate":
|
764 |
-
"loss":
|
765 |
"step": 252
|
766 |
},
|
767 |
{
|
768 |
-
"epoch":
|
769 |
-
"learning_rate":
|
770 |
-
"loss": 2.
|
771 |
"step": 254
|
772 |
},
|
773 |
{
|
774 |
-
"epoch":
|
775 |
-
"learning_rate":
|
776 |
-
"loss": 2.
|
777 |
"step": 256
|
778 |
},
|
779 |
{
|
780 |
-
"epoch":
|
781 |
-
"learning_rate":
|
782 |
-
"loss": 2.
|
783 |
"step": 258
|
784 |
},
|
785 |
{
|
786 |
-
"epoch":
|
787 |
-
"learning_rate":
|
788 |
-
"loss": 2.
|
789 |
"step": 260
|
790 |
},
|
791 |
{
|
792 |
-
"epoch":
|
793 |
-
"learning_rate":
|
794 |
-
"loss": 2.
|
795 |
"step": 262
|
796 |
},
|
797 |
{
|
798 |
-
"epoch":
|
799 |
-
"learning_rate":
|
800 |
-
"loss": 2.
|
801 |
"step": 264
|
802 |
},
|
803 |
{
|
804 |
-
"epoch":
|
805 |
-
"learning_rate":
|
806 |
-
"loss": 2.
|
807 |
"step": 266
|
808 |
},
|
809 |
{
|
810 |
-
"epoch":
|
811 |
-
"learning_rate":
|
812 |
-
"loss": 2.
|
813 |
"step": 268
|
814 |
},
|
815 |
{
|
816 |
-
"epoch":
|
817 |
-
"learning_rate":
|
818 |
-
"loss": 2.
|
819 |
"step": 270
|
820 |
},
|
821 |
{
|
822 |
-
"epoch":
|
823 |
-
"learning_rate":
|
824 |
-
"loss":
|
825 |
"step": 272
|
826 |
},
|
827 |
{
|
828 |
-
"epoch":
|
829 |
-
"learning_rate":
|
830 |
-
"loss": 2.
|
831 |
"step": 274
|
832 |
},
|
833 |
{
|
834 |
-
"epoch":
|
835 |
-
"learning_rate":
|
836 |
-
"loss":
|
837 |
"step": 276
|
838 |
},
|
839 |
{
|
840 |
-
"epoch":
|
841 |
-
"learning_rate":
|
842 |
-
"loss": 2.
|
843 |
"step": 278
|
844 |
},
|
845 |
{
|
846 |
-
"epoch":
|
847 |
-
"learning_rate":
|
848 |
-
"loss":
|
849 |
"step": 280
|
850 |
},
|
851 |
{
|
852 |
-
"epoch":
|
853 |
-
"learning_rate":
|
854 |
-
"loss":
|
855 |
"step": 282
|
856 |
},
|
857 |
{
|
858 |
-
"epoch":
|
859 |
-
"learning_rate":
|
860 |
-
"loss": 1.
|
861 |
"step": 284
|
862 |
},
|
863 |
{
|
864 |
-
"epoch":
|
865 |
-
"learning_rate":
|
866 |
-
"loss":
|
867 |
"step": 286
|
868 |
},
|
869 |
{
|
870 |
-
"epoch":
|
871 |
-
"learning_rate":
|
872 |
-
"loss":
|
873 |
"step": 288
|
874 |
},
|
875 |
{
|
876 |
-
"epoch":
|
877 |
-
"learning_rate":
|
878 |
-
"loss":
|
879 |
"step": 290
|
880 |
},
|
881 |
{
|
882 |
-
"epoch":
|
883 |
-
"learning_rate":
|
884 |
-
"loss":
|
885 |
"step": 292
|
886 |
},
|
887 |
{
|
888 |
-
"epoch":
|
889 |
-
"learning_rate":
|
890 |
-
"loss": 1.
|
891 |
"step": 294
|
892 |
},
|
893 |
{
|
894 |
-
"epoch":
|
895 |
-
"learning_rate":
|
896 |
-
"loss": 1.
|
897 |
"step": 296
|
898 |
},
|
899 |
{
|
900 |
-
"epoch":
|
901 |
-
"learning_rate":
|
902 |
-
"loss": 1.
|
903 |
"step": 298
|
904 |
},
|
905 |
{
|
906 |
-
"epoch":
|
907 |
-
"learning_rate": 0.
|
908 |
-
"loss": 1.
|
909 |
"step": 300
|
910 |
-
},
|
911 |
-
{
|
912 |
-
"epoch": 0.32,
|
913 |
-
"learning_rate": 0.00019689599440180153,
|
914 |
-
"loss": 2.2535,
|
915 |
-
"step": 302
|
916 |
-
},
|
917 |
-
{
|
918 |
-
"epoch": 0.32,
|
919 |
-
"learning_rate": 0.00019683881879496107,
|
920 |
-
"loss": 2.2597,
|
921 |
-
"step": 304
|
922 |
-
},
|
923 |
-
{
|
924 |
-
"epoch": 0.33,
|
925 |
-
"learning_rate": 0.00019678112985184308,
|
926 |
-
"loss": 2.3117,
|
927 |
-
"step": 306
|
928 |
-
},
|
929 |
-
{
|
930 |
-
"epoch": 0.33,
|
931 |
-
"learning_rate": 0.00019672292787825292,
|
932 |
-
"loss": 2.1624,
|
933 |
-
"step": 308
|
934 |
-
},
|
935 |
-
{
|
936 |
-
"epoch": 0.33,
|
937 |
-
"learning_rate": 0.00019666421318271547,
|
938 |
-
"loss": 2.1857,
|
939 |
-
"step": 310
|
940 |
-
},
|
941 |
-
{
|
942 |
-
"epoch": 0.33,
|
943 |
-
"learning_rate": 0.0001966049860764735,
|
944 |
-
"loss": 2.2101,
|
945 |
-
"step": 312
|
946 |
-
},
|
947 |
-
{
|
948 |
-
"epoch": 0.33,
|
949 |
-
"learning_rate": 0.00019654524687348607,
|
950 |
-
"loss": 2.1749,
|
951 |
-
"step": 314
|
952 |
-
},
|
953 |
-
{
|
954 |
-
"epoch": 0.34,
|
955 |
-
"learning_rate": 0.00019648499589042676,
|
956 |
-
"loss": 2.1557,
|
957 |
-
"step": 316
|
958 |
-
},
|
959 |
-
{
|
960 |
-
"epoch": 0.34,
|
961 |
-
"learning_rate": 0.00019642423344668218,
|
962 |
-
"loss": 2.2259,
|
963 |
-
"step": 318
|
964 |
-
},
|
965 |
-
{
|
966 |
-
"epoch": 0.34,
|
967 |
-
"learning_rate": 0.00019636295986435003,
|
968 |
-
"loss": 2.1723,
|
969 |
-
"step": 320
|
970 |
-
},
|
971 |
-
{
|
972 |
-
"epoch": 0.34,
|
973 |
-
"learning_rate": 0.00019630117546823759,
|
974 |
-
"loss": 2.294,
|
975 |
-
"step": 322
|
976 |
-
},
|
977 |
-
{
|
978 |
-
"epoch": 0.35,
|
979 |
-
"learning_rate": 0.00019623888058585993,
|
980 |
-
"loss": 2.2241,
|
981 |
-
"step": 324
|
982 |
-
},
|
983 |
-
{
|
984 |
-
"epoch": 0.35,
|
985 |
-
"learning_rate": 0.00019617607554743818,
|
986 |
-
"loss": 2.1496,
|
987 |
-
"step": 326
|
988 |
-
},
|
989 |
-
{
|
990 |
-
"epoch": 0.35,
|
991 |
-
"learning_rate": 0.00019611276068589776,
|
992 |
-
"loss": 2.1116,
|
993 |
-
"step": 328
|
994 |
-
},
|
995 |
-
{
|
996 |
-
"epoch": 0.35,
|
997 |
-
"learning_rate": 0.00019604893633686662,
|
998 |
-
"loss": 2.1723,
|
999 |
-
"step": 330
|
1000 |
-
},
|
1001 |
-
{
|
1002 |
-
"epoch": 0.35,
|
1003 |
-
"learning_rate": 0.0001959846028386735,
|
1004 |
-
"loss": 2.0301,
|
1005 |
-
"step": 332
|
1006 |
-
},
|
1007 |
-
{
|
1008 |
-
"epoch": 0.36,
|
1009 |
-
"learning_rate": 0.00019591976053234608,
|
1010 |
-
"loss": 1.8651,
|
1011 |
-
"step": 334
|
1012 |
-
},
|
1013 |
-
{
|
1014 |
-
"epoch": 0.36,
|
1015 |
-
"learning_rate": 0.0001958544097616092,
|
1016 |
-
"loss": 1.8286,
|
1017 |
-
"step": 336
|
1018 |
-
},
|
1019 |
-
{
|
1020 |
-
"epoch": 0.36,
|
1021 |
-
"learning_rate": 0.00019578855087288302,
|
1022 |
-
"loss": 1.9081,
|
1023 |
-
"step": 338
|
1024 |
-
},
|
1025 |
-
{
|
1026 |
-
"epoch": 0.36,
|
1027 |
-
"learning_rate": 0.0001957221842152813,
|
1028 |
-
"loss": 1.8546,
|
1029 |
-
"step": 340
|
1030 |
-
},
|
1031 |
-
{
|
1032 |
-
"epoch": 0.36,
|
1033 |
-
"learning_rate": 0.0001956553101406093,
|
1034 |
-
"loss": 1.7914,
|
1035 |
-
"step": 342
|
1036 |
-
},
|
1037 |
-
{
|
1038 |
-
"epoch": 0.37,
|
1039 |
-
"learning_rate": 0.00019558792900336216,
|
1040 |
-
"loss": 1.7997,
|
1041 |
-
"step": 344
|
1042 |
-
},
|
1043 |
-
{
|
1044 |
-
"epoch": 0.37,
|
1045 |
-
"learning_rate": 0.00019552004116072294,
|
1046 |
-
"loss": 1.3596,
|
1047 |
-
"step": 346
|
1048 |
-
},
|
1049 |
-
{
|
1050 |
-
"epoch": 0.37,
|
1051 |
-
"learning_rate": 0.0001954516469725606,
|
1052 |
-
"loss": 1.6549,
|
1053 |
-
"step": 348
|
1054 |
-
},
|
1055 |
-
{
|
1056 |
-
"epoch": 0.37,
|
1057 |
-
"learning_rate": 0.00019538274680142834,
|
1058 |
-
"loss": 1.6592,
|
1059 |
-
"step": 350
|
1060 |
-
},
|
1061 |
-
{
|
1062 |
-
"epoch": 0.38,
|
1063 |
-
"learning_rate": 0.00019531334101256147,
|
1064 |
-
"loss": 2.2759,
|
1065 |
-
"step": 352
|
1066 |
-
},
|
1067 |
-
{
|
1068 |
-
"epoch": 0.38,
|
1069 |
-
"learning_rate": 0.00019524342997387557,
|
1070 |
-
"loss": 2.1805,
|
1071 |
-
"step": 354
|
1072 |
-
},
|
1073 |
-
{
|
1074 |
-
"epoch": 0.38,
|
1075 |
-
"learning_rate": 0.0001951730140559645,
|
1076 |
-
"loss": 2.2585,
|
1077 |
-
"step": 356
|
1078 |
-
},
|
1079 |
-
{
|
1080 |
-
"epoch": 0.38,
|
1081 |
-
"learning_rate": 0.00019510209363209847,
|
1082 |
-
"loss": 2.2573,
|
1083 |
-
"step": 358
|
1084 |
-
},
|
1085 |
-
{
|
1086 |
-
"epoch": 0.38,
|
1087 |
-
"learning_rate": 0.00019503066907822198,
|
1088 |
-
"loss": 2.3154,
|
1089 |
-
"step": 360
|
1090 |
-
},
|
1091 |
-
{
|
1092 |
-
"epoch": 0.39,
|
1093 |
-
"learning_rate": 0.000194958740772952,
|
1094 |
-
"loss": 2.2306,
|
1095 |
-
"step": 362
|
1096 |
-
},
|
1097 |
-
{
|
1098 |
-
"epoch": 0.39,
|
1099 |
-
"learning_rate": 0.00019488630909757579,
|
1100 |
-
"loss": 2.3178,
|
1101 |
-
"step": 364
|
1102 |
-
},
|
1103 |
-
{
|
1104 |
-
"epoch": 0.39,
|
1105 |
-
"learning_rate": 0.00019481337443604893,
|
1106 |
-
"loss": 2.2691,
|
1107 |
-
"step": 366
|
1108 |
-
},
|
1109 |
-
{
|
1110 |
-
"epoch": 0.39,
|
1111 |
-
"learning_rate": 0.0001947399371749933,
|
1112 |
-
"loss": 2.153,
|
1113 |
-
"step": 368
|
1114 |
-
},
|
1115 |
-
{
|
1116 |
-
"epoch": 0.39,
|
1117 |
-
"learning_rate": 0.00019466599770369509,
|
1118 |
-
"loss": 2.0471,
|
1119 |
-
"step": 370
|
1120 |
-
},
|
1121 |
-
{
|
1122 |
-
"epoch": 0.4,
|
1123 |
-
"learning_rate": 0.00019459155641410257,
|
1124 |
-
"loss": 2.3142,
|
1125 |
-
"step": 372
|
1126 |
-
},
|
1127 |
-
{
|
1128 |
-
"epoch": 0.4,
|
1129 |
-
"learning_rate": 0.00019451661370082426,
|
1130 |
-
"loss": 2.1241,
|
1131 |
-
"step": 374
|
1132 |
-
},
|
1133 |
-
{
|
1134 |
-
"epoch": 0.4,
|
1135 |
-
"learning_rate": 0.0001944411699611265,
|
1136 |
-
"loss": 2.1079,
|
1137 |
-
"step": 376
|
1138 |
-
},
|
1139 |
-
{
|
1140 |
-
"epoch": 0.4,
|
1141 |
-
"learning_rate": 0.0001943652255949317,
|
1142 |
-
"loss": 2.063,
|
1143 |
-
"step": 378
|
1144 |
-
},
|
1145 |
-
{
|
1146 |
-
"epoch": 0.41,
|
1147 |
-
"learning_rate": 0.00019428878100481606,
|
1148 |
-
"loss": 2.0062,
|
1149 |
-
"step": 380
|
1150 |
-
},
|
1151 |
-
{
|
1152 |
-
"epoch": 0.41,
|
1153 |
-
"learning_rate": 0.00019421183659600725,
|
1154 |
-
"loss": 1.9,
|
1155 |
-
"step": 382
|
1156 |
-
},
|
1157 |
-
{
|
1158 |
-
"epoch": 0.41,
|
1159 |
-
"learning_rate": 0.00019413439277638265,
|
1160 |
-
"loss": 1.9723,
|
1161 |
-
"step": 384
|
1162 |
-
},
|
1163 |
-
{
|
1164 |
-
"epoch": 0.41,
|
1165 |
-
"learning_rate": 0.00019405644995646696,
|
1166 |
-
"loss": 1.969,
|
1167 |
-
"step": 386
|
1168 |
-
},
|
1169 |
-
{
|
1170 |
-
"epoch": 0.41,
|
1171 |
-
"learning_rate": 0.00019397800854942986,
|
1172 |
-
"loss": 2.0202,
|
1173 |
-
"step": 388
|
1174 |
-
},
|
1175 |
-
{
|
1176 |
-
"epoch": 0.42,
|
1177 |
-
"learning_rate": 0.00019389906897108428,
|
1178 |
-
"loss": 1.9028,
|
1179 |
-
"step": 390
|
1180 |
-
},
|
1181 |
-
{
|
1182 |
-
"epoch": 0.42,
|
1183 |
-
"learning_rate": 0.0001938196316398837,
|
1184 |
-
"loss": 1.8172,
|
1185 |
-
"step": 392
|
1186 |
-
},
|
1187 |
-
{
|
1188 |
-
"epoch": 0.42,
|
1189 |
-
"learning_rate": 0.00019373969697692028,
|
1190 |
-
"loss": 1.7243,
|
1191 |
-
"step": 394
|
1192 |
-
},
|
1193 |
-
{
|
1194 |
-
"epoch": 0.42,
|
1195 |
-
"learning_rate": 0.00019365926540592247,
|
1196 |
-
"loss": 1.4621,
|
1197 |
-
"step": 396
|
1198 |
-
},
|
1199 |
-
{
|
1200 |
-
"epoch": 0.42,
|
1201 |
-
"learning_rate": 0.0001935783373532528,
|
1202 |
-
"loss": 1.7987,
|
1203 |
-
"step": 398
|
1204 |
-
},
|
1205 |
-
{
|
1206 |
-
"epoch": 0.43,
|
1207 |
-
"learning_rate": 0.00019349691324790555,
|
1208 |
-
"loss": 1.7935,
|
1209 |
-
"step": 400
|
1210 |
-
},
|
1211 |
-
{
|
1212 |
-
"epoch": 0.43,
|
1213 |
-
"learning_rate": 0.0001934149935215047,
|
1214 |
-
"loss": 2.2858,
|
1215 |
-
"step": 402
|
1216 |
-
},
|
1217 |
-
{
|
1218 |
-
"epoch": 0.43,
|
1219 |
-
"learning_rate": 0.00019333257860830135,
|
1220 |
-
"loss": 2.3798,
|
1221 |
-
"step": 404
|
1222 |
-
},
|
1223 |
-
{
|
1224 |
-
"epoch": 0.43,
|
1225 |
-
"learning_rate": 0.00019324966894517155,
|
1226 |
-
"loss": 2.1534,
|
1227 |
-
"step": 406
|
1228 |
-
},
|
1229 |
-
{
|
1230 |
-
"epoch": 0.43,
|
1231 |
-
"learning_rate": 0.00019316626497161408,
|
1232 |
-
"loss": 2.2648,
|
1233 |
-
"step": 408
|
1234 |
-
},
|
1235 |
-
{
|
1236 |
-
"epoch": 0.44,
|
1237 |
-
"learning_rate": 0.00019308236712974795,
|
1238 |
-
"loss": 2.2946,
|
1239 |
-
"step": 410
|
1240 |
-
},
|
1241 |
-
{
|
1242 |
-
"epoch": 0.44,
|
1243 |
-
"learning_rate": 0.0001929979758643102,
|
1244 |
-
"loss": 2.319,
|
1245 |
-
"step": 412
|
1246 |
-
},
|
1247 |
-
{
|
1248 |
-
"epoch": 0.44,
|
1249 |
-
"learning_rate": 0.00019291309162265338,
|
1250 |
-
"loss": 2.2271,
|
1251 |
-
"step": 414
|
1252 |
-
},
|
1253 |
-
{
|
1254 |
-
"epoch": 0.44,
|
1255 |
-
"learning_rate": 0.0001928277148547434,
|
1256 |
-
"loss": 2.0746,
|
1257 |
-
"step": 416
|
1258 |
-
},
|
1259 |
-
{
|
1260 |
-
"epoch": 0.45,
|
1261 |
-
"learning_rate": 0.00019274184601315687,
|
1262 |
-
"loss": 2.1231,
|
1263 |
-
"step": 418
|
1264 |
-
},
|
1265 |
-
{
|
1266 |
-
"epoch": 0.45,
|
1267 |
-
"learning_rate": 0.000192655485553079,
|
1268 |
-
"loss": 2.3176,
|
1269 |
-
"step": 420
|
1270 |
-
},
|
1271 |
-
{
|
1272 |
-
"epoch": 0.45,
|
1273 |
-
"learning_rate": 0.00019256863393230094,
|
1274 |
-
"loss": 2.1657,
|
1275 |
-
"step": 422
|
1276 |
-
},
|
1277 |
-
{
|
1278 |
-
"epoch": 0.45,
|
1279 |
-
"learning_rate": 0.00019248129161121748,
|
1280 |
-
"loss": 2.2696,
|
1281 |
-
"step": 424
|
1282 |
-
},
|
1283 |
-
{
|
1284 |
-
"epoch": 0.45,
|
1285 |
-
"learning_rate": 0.0001923934590528246,
|
1286 |
-
"loss": 2.2132,
|
1287 |
-
"step": 426
|
1288 |
-
},
|
1289 |
-
{
|
1290 |
-
"epoch": 0.46,
|
1291 |
-
"learning_rate": 0.00019230513672271698,
|
1292 |
-
"loss": 2.104,
|
1293 |
-
"step": 428
|
1294 |
-
},
|
1295 |
-
{
|
1296 |
-
"epoch": 0.46,
|
1297 |
-
"learning_rate": 0.0001922163250890855,
|
1298 |
-
"loss": 1.9319,
|
1299 |
-
"step": 430
|
1300 |
-
},
|
1301 |
-
{
|
1302 |
-
"epoch": 0.46,
|
1303 |
-
"learning_rate": 0.0001921270246227149,
|
1304 |
-
"loss": 1.9152,
|
1305 |
-
"step": 432
|
1306 |
-
},
|
1307 |
-
{
|
1308 |
-
"epoch": 0.46,
|
1309 |
-
"learning_rate": 0.00019203723579698108,
|
1310 |
-
"loss": 1.8615,
|
1311 |
-
"step": 434
|
1312 |
-
},
|
1313 |
-
{
|
1314 |
-
"epoch": 0.46,
|
1315 |
-
"learning_rate": 0.00019194695908784882,
|
1316 |
-
"loss": 2.0695,
|
1317 |
-
"step": 436
|
1318 |
-
},
|
1319 |
-
{
|
1320 |
-
"epoch": 0.47,
|
1321 |
-
"learning_rate": 0.0001918561949738691,
|
1322 |
-
"loss": 1.8626,
|
1323 |
-
"step": 438
|
1324 |
-
},
|
1325 |
-
{
|
1326 |
-
"epoch": 0.47,
|
1327 |
-
"learning_rate": 0.0001917649439361765,
|
1328 |
-
"loss": 1.7538,
|
1329 |
-
"step": 440
|
1330 |
-
},
|
1331 |
-
{
|
1332 |
-
"epoch": 0.47,
|
1333 |
-
"learning_rate": 0.00019167320645848695,
|
1334 |
-
"loss": 2.0082,
|
1335 |
-
"step": 442
|
1336 |
-
},
|
1337 |
-
{
|
1338 |
-
"epoch": 0.47,
|
1339 |
-
"learning_rate": 0.00019158098302709476,
|
1340 |
-
"loss": 1.6011,
|
1341 |
-
"step": 444
|
1342 |
-
},
|
1343 |
-
{
|
1344 |
-
"epoch": 0.48,
|
1345 |
-
"learning_rate": 0.00019148827413087034,
|
1346 |
-
"loss": 1.5286,
|
1347 |
-
"step": 446
|
1348 |
-
},
|
1349 |
-
{
|
1350 |
-
"epoch": 0.48,
|
1351 |
-
"learning_rate": 0.00019139508026125754,
|
1352 |
-
"loss": 1.8795,
|
1353 |
-
"step": 448
|
1354 |
-
},
|
1355 |
-
{
|
1356 |
-
"epoch": 0.48,
|
1357 |
-
"learning_rate": 0.000191301401912271,
|
1358 |
-
"loss": 1.7509,
|
1359 |
-
"step": 450
|
1360 |
-
},
|
1361 |
-
{
|
1362 |
-
"epoch": 0.48,
|
1363 |
-
"learning_rate": 0.00019120723958049353,
|
1364 |
-
"loss": 2.3155,
|
1365 |
-
"step": 452
|
1366 |
-
},
|
1367 |
-
{
|
1368 |
-
"epoch": 0.48,
|
1369 |
-
"learning_rate": 0.0001911125937650736,
|
1370 |
-
"loss": 2.2125,
|
1371 |
-
"step": 454
|
1372 |
-
},
|
1373 |
-
{
|
1374 |
-
"epoch": 0.49,
|
1375 |
-
"learning_rate": 0.00019101746496772242,
|
1376 |
-
"loss": 2.2262,
|
1377 |
-
"step": 456
|
1378 |
-
},
|
1379 |
-
{
|
1380 |
-
"epoch": 0.49,
|
1381 |
-
"learning_rate": 0.0001909218536927116,
|
1382 |
-
"loss": 2.3185,
|
1383 |
-
"step": 458
|
1384 |
-
},
|
1385 |
-
{
|
1386 |
-
"epoch": 0.49,
|
1387 |
-
"learning_rate": 0.0001908257604468703,
|
1388 |
-
"loss": 2.2496,
|
1389 |
-
"step": 460
|
1390 |
-
},
|
1391 |
-
{
|
1392 |
-
"epoch": 0.49,
|
1393 |
-
"learning_rate": 0.00019072918573958254,
|
1394 |
-
"loss": 2.3569,
|
1395 |
-
"step": 462
|
1396 |
-
},
|
1397 |
-
{
|
1398 |
-
"epoch": 0.49,
|
1399 |
-
"learning_rate": 0.0001906321300827846,
|
1400 |
-
"loss": 2.1697,
|
1401 |
-
"step": 464
|
1402 |
-
},
|
1403 |
-
{
|
1404 |
-
"epoch": 0.5,
|
1405 |
-
"learning_rate": 0.00019053459399096215,
|
1406 |
-
"loss": 2.197,
|
1407 |
-
"step": 466
|
1408 |
-
},
|
1409 |
-
{
|
1410 |
-
"epoch": 0.5,
|
1411 |
-
"learning_rate": 0.00019043657798114766,
|
1412 |
-
"loss": 1.9929,
|
1413 |
-
"step": 468
|
1414 |
-
},
|
1415 |
-
{
|
1416 |
-
"epoch": 0.5,
|
1417 |
-
"learning_rate": 0.00019033808257291768,
|
1418 |
-
"loss": 2.1488,
|
1419 |
-
"step": 470
|
1420 |
-
},
|
1421 |
-
{
|
1422 |
-
"epoch": 0.5,
|
1423 |
-
"learning_rate": 0.0001902391082883899,
|
1424 |
-
"loss": 2.2224,
|
1425 |
-
"step": 472
|
1426 |
-
},
|
1427 |
-
{
|
1428 |
-
"epoch": 0.51,
|
1429 |
-
"learning_rate": 0.00019013965565222062,
|
1430 |
-
"loss": 2.0495,
|
1431 |
-
"step": 474
|
1432 |
-
},
|
1433 |
-
{
|
1434 |
-
"epoch": 0.51,
|
1435 |
-
"learning_rate": 0.00019003972519160178,
|
1436 |
-
"loss": 2.0964,
|
1437 |
-
"step": 476
|
1438 |
-
},
|
1439 |
-
{
|
1440 |
-
"epoch": 0.51,
|
1441 |
-
"learning_rate": 0.0001899393174362582,
|
1442 |
-
"loss": 1.8708,
|
1443 |
-
"step": 478
|
1444 |
-
},
|
1445 |
-
{
|
1446 |
-
"epoch": 0.51,
|
1447 |
-
"learning_rate": 0.00018983843291844492,
|
1448 |
-
"loss": 1.9741,
|
1449 |
-
"step": 480
|
1450 |
-
},
|
1451 |
-
{
|
1452 |
-
"epoch": 0.51,
|
1453 |
-
"learning_rate": 0.00018973707217294415,
|
1454 |
-
"loss": 1.9908,
|
1455 |
-
"step": 482
|
1456 |
-
},
|
1457 |
-
{
|
1458 |
-
"epoch": 0.52,
|
1459 |
-
"learning_rate": 0.00018963523573706264,
|
1460 |
-
"loss": 1.7973,
|
1461 |
-
"step": 484
|
1462 |
-
},
|
1463 |
-
{
|
1464 |
-
"epoch": 0.52,
|
1465 |
-
"learning_rate": 0.0001895329241506287,
|
1466 |
-
"loss": 1.9293,
|
1467 |
-
"step": 486
|
1468 |
-
},
|
1469 |
-
{
|
1470 |
-
"epoch": 0.52,
|
1471 |
-
"learning_rate": 0.00018943013795598944,
|
1472 |
-
"loss": 1.94,
|
1473 |
-
"step": 488
|
1474 |
-
},
|
1475 |
-
{
|
1476 |
-
"epoch": 0.52,
|
1477 |
-
"learning_rate": 0.00018932687769800767,
|
1478 |
-
"loss": 1.9435,
|
1479 |
-
"step": 490
|
1480 |
-
},
|
1481 |
-
{
|
1482 |
-
"epoch": 0.52,
|
1483 |
-
"learning_rate": 0.00018922314392405944,
|
1484 |
-
"loss": 1.7907,
|
1485 |
-
"step": 492
|
1486 |
-
},
|
1487 |
-
{
|
1488 |
-
"epoch": 0.53,
|
1489 |
-
"learning_rate": 0.00018911893718403063,
|
1490 |
-
"loss": 1.6344,
|
1491 |
-
"step": 494
|
1492 |
-
},
|
1493 |
-
{
|
1494 |
-
"epoch": 0.53,
|
1495 |
-
"learning_rate": 0.00018901425803031447,
|
1496 |
-
"loss": 1.3563,
|
1497 |
-
"step": 496
|
1498 |
-
},
|
1499 |
-
{
|
1500 |
-
"epoch": 0.53,
|
1501 |
-
"learning_rate": 0.00018890910701780826,
|
1502 |
-
"loss": 1.7,
|
1503 |
-
"step": 498
|
1504 |
-
},
|
1505 |
-
{
|
1506 |
-
"epoch": 0.53,
|
1507 |
-
"learning_rate": 0.00018880348470391077,
|
1508 |
-
"loss": 1.7682,
|
1509 |
-
"step": 500
|
1510 |
-
},
|
1511 |
-
{
|
1512 |
-
"epoch": 0.54,
|
1513 |
-
"learning_rate": 0.00018869739164851889,
|
1514 |
-
"loss": 2.2041,
|
1515 |
-
"step": 502
|
1516 |
-
},
|
1517 |
-
{
|
1518 |
-
"epoch": 0.54,
|
1519 |
-
"learning_rate": 0.00018859082841402513,
|
1520 |
-
"loss": 2.237,
|
1521 |
-
"step": 504
|
1522 |
-
},
|
1523 |
-
{
|
1524 |
-
"epoch": 0.54,
|
1525 |
-
"learning_rate": 0.0001884837955653142,
|
1526 |
-
"loss": 2.1694,
|
1527 |
-
"step": 506
|
1528 |
-
},
|
1529 |
-
{
|
1530 |
-
"epoch": 0.54,
|
1531 |
-
"learning_rate": 0.00018837629366976025,
|
1532 |
-
"loss": 2.2373,
|
1533 |
-
"step": 508
|
1534 |
-
},
|
1535 |
-
{
|
1536 |
-
"epoch": 0.54,
|
1537 |
-
"learning_rate": 0.0001882683232972239,
|
1538 |
-
"loss": 2.2511,
|
1539 |
-
"step": 510
|
1540 |
-
},
|
1541 |
-
{
|
1542 |
-
"epoch": 0.55,
|
1543 |
-
"learning_rate": 0.000188159885020049,
|
1544 |
-
"loss": 2.2474,
|
1545 |
-
"step": 512
|
1546 |
-
},
|
1547 |
-
{
|
1548 |
-
"epoch": 0.55,
|
1549 |
-
"learning_rate": 0.00018805097941305984,
|
1550 |
-
"loss": 2.1938,
|
1551 |
-
"step": 514
|
1552 |
-
},
|
1553 |
-
{
|
1554 |
-
"epoch": 0.55,
|
1555 |
-
"learning_rate": 0.00018794160705355796,
|
1556 |
-
"loss": 2.0874,
|
1557 |
-
"step": 516
|
1558 |
-
},
|
1559 |
-
{
|
1560 |
-
"epoch": 0.55,
|
1561 |
-
"learning_rate": 0.00018783176852131908,
|
1562 |
-
"loss": 2.2795,
|
1563 |
-
"step": 518
|
1564 |
-
},
|
1565 |
-
{
|
1566 |
-
"epoch": 0.55,
|
1567 |
-
"learning_rate": 0.00018772146439859015,
|
1568 |
-
"loss": 2.179,
|
1569 |
-
"step": 520
|
1570 |
-
},
|
1571 |
-
{
|
1572 |
-
"epoch": 0.56,
|
1573 |
-
"learning_rate": 0.00018761069527008613,
|
1574 |
-
"loss": 2.0959,
|
1575 |
-
"step": 522
|
1576 |
-
},
|
1577 |
-
{
|
1578 |
-
"epoch": 0.56,
|
1579 |
-
"learning_rate": 0.00018749946172298698,
|
1580 |
-
"loss": 2.3345,
|
1581 |
-
"step": 524
|
1582 |
-
},
|
1583 |
-
{
|
1584 |
-
"epoch": 0.56,
|
1585 |
-
"learning_rate": 0.00018738776434693447,
|
1586 |
-
"loss": 2.0522,
|
1587 |
-
"step": 526
|
1588 |
-
},
|
1589 |
-
{
|
1590 |
-
"epoch": 0.56,
|
1591 |
-
"learning_rate": 0.00018727560373402917,
|
1592 |
-
"loss": 2.1055,
|
1593 |
-
"step": 528
|
1594 |
-
},
|
1595 |
-
{
|
1596 |
-
"epoch": 0.56,
|
1597 |
-
"learning_rate": 0.00018716298047882714,
|
1598 |
-
"loss": 2.0651,
|
1599 |
-
"step": 530
|
1600 |
-
},
|
1601 |
-
{
|
1602 |
-
"epoch": 0.57,
|
1603 |
-
"learning_rate": 0.00018704989517833695,
|
1604 |
-
"loss": 2.0961,
|
1605 |
-
"step": 532
|
1606 |
-
},
|
1607 |
-
{
|
1608 |
-
"epoch": 0.57,
|
1609 |
-
"learning_rate": 0.0001869363484320164,
|
1610 |
-
"loss": 1.9879,
|
1611 |
-
"step": 534
|
1612 |
-
},
|
1613 |
-
{
|
1614 |
-
"epoch": 0.57,
|
1615 |
-
"learning_rate": 0.00018682234084176945,
|
1616 |
-
"loss": 2.0118,
|
1617 |
-
"step": 536
|
1618 |
-
},
|
1619 |
-
{
|
1620 |
-
"epoch": 0.57,
|
1621 |
-
"learning_rate": 0.0001867078730119429,
|
1622 |
-
"loss": 1.9059,
|
1623 |
-
"step": 538
|
1624 |
-
},
|
1625 |
-
{
|
1626 |
-
"epoch": 0.58,
|
1627 |
-
"learning_rate": 0.00018659294554932324,
|
1628 |
-
"loss": 1.8227,
|
1629 |
-
"step": 540
|
1630 |
-
},
|
1631 |
-
{
|
1632 |
-
"epoch": 0.58,
|
1633 |
-
"learning_rate": 0.00018647755906313348,
|
1634 |
-
"loss": 1.8794,
|
1635 |
-
"step": 542
|
1636 |
-
},
|
1637 |
-
{
|
1638 |
-
"epoch": 0.58,
|
1639 |
-
"learning_rate": 0.0001863617141650299,
|
1640 |
-
"loss": 1.6243,
|
1641 |
-
"step": 544
|
1642 |
-
},
|
1643 |
-
{
|
1644 |
-
"epoch": 0.58,
|
1645 |
-
"learning_rate": 0.00018624541146909873,
|
1646 |
-
"loss": 1.6429,
|
1647 |
-
"step": 546
|
1648 |
-
},
|
1649 |
-
{
|
1650 |
-
"epoch": 0.58,
|
1651 |
-
"learning_rate": 0.00018612865159185304,
|
1652 |
-
"loss": 2.03,
|
1653 |
-
"step": 548
|
1654 |
-
},
|
1655 |
-
{
|
1656 |
-
"epoch": 0.59,
|
1657 |
-
"learning_rate": 0.0001860114351522293,
|
1658 |
-
"loss": 1.7897,
|
1659 |
-
"step": 550
|
1660 |
-
},
|
1661 |
-
{
|
1662 |
-
"epoch": 0.59,
|
1663 |
-
"learning_rate": 0.00018589376277158425,
|
1664 |
-
"loss": 2.2226,
|
1665 |
-
"step": 552
|
1666 |
-
},
|
1667 |
-
{
|
1668 |
-
"epoch": 0.59,
|
1669 |
-
"learning_rate": 0.00018577563507369153,
|
1670 |
-
"loss": 2.2998,
|
1671 |
-
"step": 554
|
1672 |
-
},
|
1673 |
-
{
|
1674 |
-
"epoch": 0.59,
|
1675 |
-
"learning_rate": 0.00018565705268473837,
|
1676 |
-
"loss": 2.1385,
|
1677 |
-
"step": 556
|
1678 |
-
},
|
1679 |
-
{
|
1680 |
-
"epoch": 0.59,
|
1681 |
-
"learning_rate": 0.0001855380162333223,
|
1682 |
-
"loss": 2.1793,
|
1683 |
-
"step": 558
|
1684 |
-
},
|
1685 |
-
{
|
1686 |
-
"epoch": 0.6,
|
1687 |
-
"learning_rate": 0.0001854185263504478,
|
1688 |
-
"loss": 2.2456,
|
1689 |
-
"step": 560
|
1690 |
-
},
|
1691 |
-
{
|
1692 |
-
"epoch": 0.6,
|
1693 |
-
"learning_rate": 0.00018529858366952298,
|
1694 |
-
"loss": 2.162,
|
1695 |
-
"step": 562
|
1696 |
-
},
|
1697 |
-
{
|
1698 |
-
"epoch": 0.6,
|
1699 |
-
"learning_rate": 0.00018517818882635617,
|
1700 |
-
"loss": 2.1046,
|
1701 |
-
"step": 564
|
1702 |
-
},
|
1703 |
-
{
|
1704 |
-
"epoch": 0.6,
|
1705 |
-
"learning_rate": 0.0001850573424591526,
|
1706 |
-
"loss": 2.1761,
|
1707 |
-
"step": 566
|
1708 |
-
},
|
1709 |
-
{
|
1710 |
-
"epoch": 0.61,
|
1711 |
-
"learning_rate": 0.00018493604520851097,
|
1712 |
-
"loss": 2.2593,
|
1713 |
-
"step": 568
|
1714 |
-
},
|
1715 |
-
{
|
1716 |
-
"epoch": 0.61,
|
1717 |
-
"learning_rate": 0.00018481429771742018,
|
1718 |
-
"loss": 2.2067,
|
1719 |
-
"step": 570
|
1720 |
-
},
|
1721 |
-
{
|
1722 |
-
"epoch": 0.61,
|
1723 |
-
"learning_rate": 0.00018469210063125572,
|
1724 |
-
"loss": 2.1257,
|
1725 |
-
"step": 572
|
1726 |
-
},
|
1727 |
-
{
|
1728 |
-
"epoch": 0.61,
|
1729 |
-
"learning_rate": 0.00018456945459777643,
|
1730 |
-
"loss": 2.2823,
|
1731 |
-
"step": 574
|
1732 |
-
},
|
1733 |
-
{
|
1734 |
-
"epoch": 0.61,
|
1735 |
-
"learning_rate": 0.0001844463602671209,
|
1736 |
-
"loss": 2.2942,
|
1737 |
-
"step": 576
|
1738 |
-
},
|
1739 |
-
{
|
1740 |
-
"epoch": 0.62,
|
1741 |
-
"learning_rate": 0.0001843228182918042,
|
1742 |
-
"loss": 2.1126,
|
1743 |
-
"step": 578
|
1744 |
-
},
|
1745 |
-
{
|
1746 |
-
"epoch": 0.62,
|
1747 |
-
"learning_rate": 0.0001841988293267143,
|
1748 |
-
"loss": 2.2014,
|
1749 |
-
"step": 580
|
1750 |
-
},
|
1751 |
-
{
|
1752 |
-
"epoch": 0.62,
|
1753 |
-
"learning_rate": 0.00018407439402910858,
|
1754 |
-
"loss": 2.0244,
|
1755 |
-
"step": 582
|
1756 |
-
},
|
1757 |
-
{
|
1758 |
-
"epoch": 0.62,
|
1759 |
-
"learning_rate": 0.00018394951305861055,
|
1760 |
-
"loss": 1.9842,
|
1761 |
-
"step": 584
|
1762 |
-
},
|
1763 |
-
{
|
1764 |
-
"epoch": 0.62,
|
1765 |
-
"learning_rate": 0.00018382418707720604,
|
1766 |
-
"loss": 1.8637,
|
1767 |
-
"step": 586
|
1768 |
-
},
|
1769 |
-
{
|
1770 |
-
"epoch": 0.63,
|
1771 |
-
"learning_rate": 0.00018369841674923998,
|
1772 |
-
"loss": 1.8229,
|
1773 |
-
"step": 588
|
1774 |
-
},
|
1775 |
-
{
|
1776 |
-
"epoch": 0.63,
|
1777 |
-
"learning_rate": 0.00018357220274141262,
|
1778 |
-
"loss": 1.8023,
|
1779 |
-
"step": 590
|
1780 |
-
},
|
1781 |
-
{
|
1782 |
-
"epoch": 0.63,
|
1783 |
-
"learning_rate": 0.00018344554572277628,
|
1784 |
-
"loss": 1.6493,
|
1785 |
-
"step": 592
|
1786 |
-
},
|
1787 |
-
{
|
1788 |
-
"epoch": 0.63,
|
1789 |
-
"learning_rate": 0.00018331844636473152,
|
1790 |
-
"loss": 1.4272,
|
1791 |
-
"step": 594
|
1792 |
-
},
|
1793 |
-
{
|
1794 |
-
"epoch": 0.64,
|
1795 |
-
"learning_rate": 0.00018319090534102381,
|
1796 |
-
"loss": 1.5226,
|
1797 |
-
"step": 596
|
1798 |
-
},
|
1799 |
-
{
|
1800 |
-
"epoch": 0.64,
|
1801 |
-
"learning_rate": 0.0001830629233277398,
|
1802 |
-
"loss": 1.5868,
|
1803 |
-
"step": 598
|
1804 |
-
},
|
1805 |
-
{
|
1806 |
-
"epoch": 0.64,
|
1807 |
-
"learning_rate": 0.00018293450100330375,
|
1808 |
-
"loss": 1.6859,
|
1809 |
-
"step": 600
|
1810 |
-
},
|
1811 |
-
{
|
1812 |
-
"epoch": 0.64,
|
1813 |
-
"learning_rate": 0.00018280563904847415,
|
1814 |
-
"loss": 2.2781,
|
1815 |
-
"step": 602
|
1816 |
-
},
|
1817 |
-
{
|
1818 |
-
"epoch": 0.64,
|
1819 |
-
"learning_rate": 0.0001826763381463398,
|
1820 |
-
"loss": 2.2743,
|
1821 |
-
"step": 604
|
1822 |
-
},
|
1823 |
-
{
|
1824 |
-
"epoch": 0.65,
|
1825 |
-
"learning_rate": 0.0001825465989823164,
|
1826 |
-
"loss": 2.2974,
|
1827 |
-
"step": 606
|
1828 |
-
},
|
1829 |
-
{
|
1830 |
-
"epoch": 0.65,
|
1831 |
-
"learning_rate": 0.00018241642224414272,
|
1832 |
-
"loss": 2.089,
|
1833 |
-
"step": 608
|
1834 |
-
},
|
1835 |
-
{
|
1836 |
-
"epoch": 0.65,
|
1837 |
-
"learning_rate": 0.00018228580862187727,
|
1838 |
-
"loss": 2.3559,
|
1839 |
-
"step": 610
|
1840 |
-
},
|
1841 |
-
{
|
1842 |
-
"epoch": 0.65,
|
1843 |
-
"learning_rate": 0.00018215475880789433,
|
1844 |
-
"loss": 2.2152,
|
1845 |
-
"step": 612
|
1846 |
-
},
|
1847 |
-
{
|
1848 |
-
"epoch": 0.65,
|
1849 |
-
"learning_rate": 0.00018202327349688043,
|
1850 |
-
"loss": 2.1726,
|
1851 |
-
"step": 614
|
1852 |
-
},
|
1853 |
-
{
|
1854 |
-
"epoch": 0.66,
|
1855 |
-
"learning_rate": 0.00018189135338583066,
|
1856 |
-
"loss": 2.2242,
|
1857 |
-
"step": 616
|
1858 |
-
},
|
1859 |
-
{
|
1860 |
-
"epoch": 0.66,
|
1861 |
-
"learning_rate": 0.00018175899917404492,
|
1862 |
-
"loss": 2.2506,
|
1863 |
-
"step": 618
|
1864 |
-
},
|
1865 |
-
{
|
1866 |
-
"epoch": 0.66,
|
1867 |
-
"learning_rate": 0.00018162621156312433,
|
1868 |
-
"loss": 2.1324,
|
1869 |
-
"step": 620
|
1870 |
-
},
|
1871 |
-
{
|
1872 |
-
"epoch": 0.66,
|
1873 |
-
"learning_rate": 0.00018149299125696735,
|
1874 |
-
"loss": 2.006,
|
1875 |
-
"step": 622
|
1876 |
-
},
|
1877 |
-
{
|
1878 |
-
"epoch": 0.67,
|
1879 |
-
"learning_rate": 0.00018135933896176612,
|
1880 |
-
"loss": 2.2178,
|
1881 |
-
"step": 624
|
1882 |
-
},
|
1883 |
-
{
|
1884 |
-
"epoch": 0.67,
|
1885 |
-
"learning_rate": 0.00018122525538600282,
|
1886 |
-
"loss": 2.0817,
|
1887 |
-
"step": 626
|
1888 |
-
},
|
1889 |
-
{
|
1890 |
-
"epoch": 0.67,
|
1891 |
-
"learning_rate": 0.00018109074124044572,
|
1892 |
-
"loss": 2.0006,
|
1893 |
-
"step": 628
|
1894 |
-
},
|
1895 |
-
{
|
1896 |
-
"epoch": 0.67,
|
1897 |
-
"learning_rate": 0.00018095579723814557,
|
1898 |
-
"loss": 1.9424,
|
1899 |
-
"step": 630
|
1900 |
-
},
|
1901 |
-
{
|
1902 |
-
"epoch": 0.67,
|
1903 |
-
"learning_rate": 0.00018082042409443174,
|
1904 |
-
"loss": 1.9661,
|
1905 |
-
"step": 632
|
1906 |
-
},
|
1907 |
-
{
|
1908 |
-
"epoch": 0.68,
|
1909 |
-
"learning_rate": 0.00018068462252690843,
|
1910 |
-
"loss": 1.8678,
|
1911 |
-
"step": 634
|
1912 |
-
},
|
1913 |
-
{
|
1914 |
-
"epoch": 0.68,
|
1915 |
-
"learning_rate": 0.00018054839325545096,
|
1916 |
-
"loss": 1.8343,
|
1917 |
-
"step": 636
|
1918 |
-
},
|
1919 |
-
{
|
1920 |
-
"epoch": 0.68,
|
1921 |
-
"learning_rate": 0.0001804117370022018,
|
1922 |
-
"loss": 1.8003,
|
1923 |
-
"step": 638
|
1924 |
-
},
|
1925 |
-
{
|
1926 |
-
"epoch": 0.68,
|
1927 |
-
"learning_rate": 0.0001802746544915669,
|
1928 |
-
"loss": 1.7928,
|
1929 |
-
"step": 640
|
1930 |
-
},
|
1931 |
-
{
|
1932 |
-
"epoch": 0.68,
|
1933 |
-
"learning_rate": 0.00018013714645021166,
|
1934 |
-
"loss": 1.8922,
|
1935 |
-
"step": 642
|
1936 |
-
},
|
1937 |
-
{
|
1938 |
-
"epoch": 0.69,
|
1939 |
-
"learning_rate": 0.00017999921360705733,
|
1940 |
-
"loss": 1.6961,
|
1941 |
-
"step": 644
|
1942 |
-
},
|
1943 |
-
{
|
1944 |
-
"epoch": 0.69,
|
1945 |
-
"learning_rate": 0.0001798608566932769,
|
1946 |
-
"loss": 1.3043,
|
1947 |
-
"step": 646
|
1948 |
-
},
|
1949 |
-
{
|
1950 |
-
"epoch": 0.69,
|
1951 |
-
"learning_rate": 0.00017972207644229138,
|
1952 |
-
"loss": 1.8295,
|
1953 |
-
"step": 648
|
1954 |
-
},
|
1955 |
-
{
|
1956 |
-
"epoch": 0.69,
|
1957 |
-
"learning_rate": 0.0001795828735897658,
|
1958 |
-
"loss": 1.7105,
|
1959 |
-
"step": 650
|
1960 |
-
},
|
1961 |
-
{
|
1962 |
-
"epoch": 0.69,
|
1963 |
-
"learning_rate": 0.00017944324887360553,
|
1964 |
-
"loss": 2.1881,
|
1965 |
-
"step": 652
|
1966 |
-
},
|
1967 |
-
{
|
1968 |
-
"epoch": 0.7,
|
1969 |
-
"learning_rate": 0.000179303203033952,
|
1970 |
-
"loss": 2.2434,
|
1971 |
-
"step": 654
|
1972 |
-
},
|
1973 |
-
{
|
1974 |
-
"epoch": 0.7,
|
1975 |
-
"learning_rate": 0.0001791627368131792,
|
1976 |
-
"loss": 2.3566,
|
1977 |
-
"step": 656
|
1978 |
-
},
|
1979 |
-
{
|
1980 |
-
"epoch": 0.7,
|
1981 |
-
"learning_rate": 0.00017902185095588927,
|
1982 |
-
"loss": 2.3222,
|
1983 |
-
"step": 658
|
1984 |
-
},
|
1985 |
-
{
|
1986 |
-
"epoch": 0.7,
|
1987 |
-
"learning_rate": 0.00017888054620890915,
|
1988 |
-
"loss": 2.181,
|
1989 |
-
"step": 660
|
1990 |
-
},
|
1991 |
-
{
|
1992 |
-
"epoch": 0.71,
|
1993 |
-
"learning_rate": 0.00017873882332128597,
|
1994 |
-
"loss": 2.3261,
|
1995 |
-
"step": 662
|
1996 |
-
},
|
1997 |
-
{
|
1998 |
-
"epoch": 0.71,
|
1999 |
-
"learning_rate": 0.00017859668304428365,
|
2000 |
-
"loss": 2.2798,
|
2001 |
-
"step": 664
|
2002 |
-
},
|
2003 |
-
{
|
2004 |
-
"epoch": 0.71,
|
2005 |
-
"learning_rate": 0.00017845412613137844,
|
2006 |
-
"loss": 2.0487,
|
2007 |
-
"step": 666
|
2008 |
-
},
|
2009 |
-
{
|
2010 |
-
"epoch": 0.71,
|
2011 |
-
"learning_rate": 0.00017831115333825535,
|
2012 |
-
"loss": 2.1863,
|
2013 |
-
"step": 668
|
2014 |
-
},
|
2015 |
-
{
|
2016 |
-
"epoch": 0.71,
|
2017 |
-
"learning_rate": 0.00017816776542280377,
|
2018 |
-
"loss": 2.1308,
|
2019 |
-
"step": 670
|
2020 |
-
},
|
2021 |
-
{
|
2022 |
-
"epoch": 0.72,
|
2023 |
-
"learning_rate": 0.0001780239631451138,
|
2024 |
-
"loss": 2.3122,
|
2025 |
-
"step": 672
|
2026 |
-
},
|
2027 |
-
{
|
2028 |
-
"epoch": 0.72,
|
2029 |
-
"learning_rate": 0.0001778797472674719,
|
2030 |
-
"loss": 2.2757,
|
2031 |
-
"step": 674
|
2032 |
-
},
|
2033 |
-
{
|
2034 |
-
"epoch": 0.72,
|
2035 |
-
"learning_rate": 0.00017773511855435708,
|
2036 |
-
"loss": 2.0241,
|
2037 |
-
"step": 676
|
2038 |
-
},
|
2039 |
-
{
|
2040 |
-
"epoch": 0.72,
|
2041 |
-
"learning_rate": 0.00017759007777243672,
|
2042 |
-
"loss": 1.9896,
|
2043 |
-
"step": 678
|
2044 |
-
},
|
2045 |
-
{
|
2046 |
-
"epoch": 0.72,
|
2047 |
-
"learning_rate": 0.00017744462569056256,
|
2048 |
-
"loss": 1.9607,
|
2049 |
-
"step": 680
|
2050 |
-
},
|
2051 |
-
{
|
2052 |
-
"epoch": 0.73,
|
2053 |
-
"learning_rate": 0.00017729876307976663,
|
2054 |
-
"loss": 1.9798,
|
2055 |
-
"step": 682
|
2056 |
-
},
|
2057 |
-
{
|
2058 |
-
"epoch": 0.73,
|
2059 |
-
"learning_rate": 0.00017715249071325717,
|
2060 |
-
"loss": 1.9075,
|
2061 |
-
"step": 684
|
2062 |
-
},
|
2063 |
-
{
|
2064 |
-
"epoch": 0.73,
|
2065 |
-
"learning_rate": 0.00017700580936641443,
|
2066 |
-
"loss": 2.0141,
|
2067 |
-
"step": 686
|
2068 |
-
},
|
2069 |
-
{
|
2070 |
-
"epoch": 0.73,
|
2071 |
-
"learning_rate": 0.00017685871981678672,
|
2072 |
-
"loss": 1.9238,
|
2073 |
-
"step": 688
|
2074 |
-
},
|
2075 |
-
{
|
2076 |
-
"epoch": 0.74,
|
2077 |
-
"learning_rate": 0.00017671122284408614,
|
2078 |
-
"loss": 1.9244,
|
2079 |
-
"step": 690
|
2080 |
-
},
|
2081 |
-
{
|
2082 |
-
"epoch": 0.74,
|
2083 |
-
"learning_rate": 0.00017656331923018457,
|
2084 |
-
"loss": 1.6621,
|
2085 |
-
"step": 692
|
2086 |
-
},
|
2087 |
-
{
|
2088 |
-
"epoch": 0.74,
|
2089 |
-
"learning_rate": 0.00017641500975910945,
|
2090 |
-
"loss": 1.7402,
|
2091 |
-
"step": 694
|
2092 |
-
},
|
2093 |
-
{
|
2094 |
-
"epoch": 0.74,
|
2095 |
-
"learning_rate": 0.0001762662952170396,
|
2096 |
-
"loss": 1.3913,
|
2097 |
-
"step": 696
|
2098 |
-
},
|
2099 |
-
{
|
2100 |
-
"epoch": 0.74,
|
2101 |
-
"learning_rate": 0.0001761171763923012,
|
2102 |
-
"loss": 1.5825,
|
2103 |
-
"step": 698
|
2104 |
-
},
|
2105 |
-
{
|
2106 |
-
"epoch": 0.75,
|
2107 |
-
"learning_rate": 0.0001759676540753634,
|
2108 |
-
"loss": 1.6809,
|
2109 |
-
"step": 700
|
2110 |
-
},
|
2111 |
-
{
|
2112 |
-
"epoch": 0.75,
|
2113 |
-
"learning_rate": 0.00017581772905883423,
|
2114 |
-
"loss": 2.3459,
|
2115 |
-
"step": 702
|
2116 |
-
},
|
2117 |
-
{
|
2118 |
-
"epoch": 0.75,
|
2119 |
-
"learning_rate": 0.00017566740213745648,
|
2120 |
-
"loss": 2.1963,
|
2121 |
-
"step": 704
|
2122 |
-
},
|
2123 |
-
{
|
2124 |
-
"epoch": 0.75,
|
2125 |
-
"learning_rate": 0.00017551667410810337,
|
2126 |
-
"loss": 2.1334,
|
2127 |
-
"step": 706
|
2128 |
-
},
|
2129 |
-
{
|
2130 |
-
"epoch": 0.75,
|
2131 |
-
"learning_rate": 0.00017536554576977442,
|
2132 |
-
"loss": 2.3778,
|
2133 |
-
"step": 708
|
2134 |
-
},
|
2135 |
-
{
|
2136 |
-
"epoch": 0.76,
|
2137 |
-
"learning_rate": 0.00017521401792359108,
|
2138 |
-
"loss": 2.183,
|
2139 |
-
"step": 710
|
2140 |
-
},
|
2141 |
-
{
|
2142 |
-
"epoch": 0.76,
|
2143 |
-
"learning_rate": 0.0001750620913727926,
|
2144 |
-
"loss": 2.1674,
|
2145 |
-
"step": 712
|
2146 |
-
},
|
2147 |
-
{
|
2148 |
-
"epoch": 0.76,
|
2149 |
-
"learning_rate": 0.00017490976692273176,
|
2150 |
-
"loss": 2.3534,
|
2151 |
-
"step": 714
|
2152 |
-
},
|
2153 |
-
{
|
2154 |
-
"epoch": 0.76,
|
2155 |
-
"learning_rate": 0.00017475704538087055,
|
2156 |
-
"loss": 2.1677,
|
2157 |
-
"step": 716
|
2158 |
-
},
|
2159 |
-
{
|
2160 |
-
"epoch": 0.77,
|
2161 |
-
"learning_rate": 0.00017460392755677592,
|
2162 |
-
"loss": 2.1642,
|
2163 |
-
"step": 718
|
2164 |
-
},
|
2165 |
-
{
|
2166 |
-
"epoch": 0.77,
|
2167 |
-
"learning_rate": 0.0001744504142621155,
|
2168 |
-
"loss": 2.1983,
|
2169 |
-
"step": 720
|
2170 |
-
},
|
2171 |
-
{
|
2172 |
-
"epoch": 0.77,
|
2173 |
-
"learning_rate": 0.0001742965063106533,
|
2174 |
-
"loss": 2.1661,
|
2175 |
-
"step": 722
|
2176 |
-
},
|
2177 |
-
{
|
2178 |
-
"epoch": 0.77,
|
2179 |
-
"learning_rate": 0.0001741422045182453,
|
2180 |
-
"loss": 2.2135,
|
2181 |
-
"step": 724
|
2182 |
-
},
|
2183 |
-
{
|
2184 |
-
"epoch": 0.77,
|
2185 |
-
"learning_rate": 0.00017398750970283532,
|
2186 |
-
"loss": 2.1288,
|
2187 |
-
"step": 726
|
2188 |
-
},
|
2189 |
-
{
|
2190 |
-
"epoch": 0.78,
|
2191 |
-
"learning_rate": 0.00017383242268445047,
|
2192 |
-
"loss": 1.9906,
|
2193 |
-
"step": 728
|
2194 |
-
},
|
2195 |
-
{
|
2196 |
-
"epoch": 0.78,
|
2197 |
-
"learning_rate": 0.00017367694428519696,
|
2198 |
-
"loss": 1.9031,
|
2199 |
-
"step": 730
|
2200 |
-
},
|
2201 |
-
{
|
2202 |
-
"epoch": 0.78,
|
2203 |
-
"learning_rate": 0.00017352107532925569,
|
2204 |
-
"loss": 1.831,
|
2205 |
-
"step": 732
|
2206 |
-
},
|
2207 |
-
{
|
2208 |
-
"epoch": 0.78,
|
2209 |
-
"learning_rate": 0.00017336481664287777,
|
2210 |
-
"loss": 1.9116,
|
2211 |
-
"step": 734
|
2212 |
-
},
|
2213 |
-
{
|
2214 |
-
"epoch": 0.78,
|
2215 |
-
"learning_rate": 0.00017320816905438044,
|
2216 |
-
"loss": 1.8241,
|
2217 |
-
"step": 736
|
2218 |
-
},
|
2219 |
-
{
|
2220 |
-
"epoch": 0.79,
|
2221 |
-
"learning_rate": 0.0001730511333941423,
|
2222 |
-
"loss": 1.8745,
|
2223 |
-
"step": 738
|
2224 |
-
},
|
2225 |
-
{
|
2226 |
-
"epoch": 0.79,
|
2227 |
-
"learning_rate": 0.00017289371049459922,
|
2228 |
-
"loss": 1.8306,
|
2229 |
-
"step": 740
|
2230 |
-
},
|
2231 |
-
{
|
2232 |
-
"epoch": 0.79,
|
2233 |
-
"learning_rate": 0.00017273590119023968,
|
2234 |
-
"loss": 1.6032,
|
2235 |
-
"step": 742
|
2236 |
-
},
|
2237 |
-
{
|
2238 |
-
"epoch": 0.79,
|
2239 |
-
"learning_rate": 0.00017257770631760058,
|
2240 |
-
"loss": 1.3571,
|
2241 |
-
"step": 744
|
2242 |
-
},
|
2243 |
-
{
|
2244 |
-
"epoch": 0.8,
|
2245 |
-
"learning_rate": 0.00017241912671526265,
|
2246 |
-
"loss": 1.6282,
|
2247 |
-
"step": 746
|
2248 |
-
},
|
2249 |
-
{
|
2250 |
-
"epoch": 0.8,
|
2251 |
-
"learning_rate": 0.00017226016322384604,
|
2252 |
-
"loss": 1.5942,
|
2253 |
-
"step": 748
|
2254 |
-
},
|
2255 |
-
{
|
2256 |
-
"epoch": 0.8,
|
2257 |
-
"learning_rate": 0.00017210081668600586,
|
2258 |
-
"loss": 1.6988,
|
2259 |
-
"step": 750
|
2260 |
-
},
|
2261 |
-
{
|
2262 |
-
"epoch": 0.8,
|
2263 |
-
"learning_rate": 0.00017194108794642775,
|
2264 |
-
"loss": 2.334,
|
2265 |
-
"step": 752
|
2266 |
-
},
|
2267 |
-
{
|
2268 |
-
"epoch": 0.8,
|
2269 |
-
"learning_rate": 0.00017178097785182337,
|
2270 |
-
"loss": 2.2731,
|
2271 |
-
"step": 754
|
2272 |
-
},
|
2273 |
-
{
|
2274 |
-
"epoch": 0.81,
|
2275 |
-
"learning_rate": 0.0001716204872509259,
|
2276 |
-
"loss": 2.1642,
|
2277 |
-
"step": 756
|
2278 |
-
},
|
2279 |
-
{
|
2280 |
-
"epoch": 0.81,
|
2281 |
-
"learning_rate": 0.00017145961699448559,
|
2282 |
-
"loss": 2.4107,
|
2283 |
-
"step": 758
|
2284 |
-
},
|
2285 |
-
{
|
2286 |
-
"epoch": 0.81,
|
2287 |
-
"learning_rate": 0.00017129836793526517,
|
2288 |
-
"loss": 2.2767,
|
2289 |
-
"step": 760
|
2290 |
-
},
|
2291 |
-
{
|
2292 |
-
"epoch": 0.81,
|
2293 |
-
"learning_rate": 0.00017113674092803543,
|
2294 |
-
"loss": 2.3137,
|
2295 |
-
"step": 762
|
2296 |
-
},
|
2297 |
-
{
|
2298 |
-
"epoch": 0.81,
|
2299 |
-
"learning_rate": 0.00017097473682957067,
|
2300 |
-
"loss": 2.3095,
|
2301 |
-
"step": 764
|
2302 |
-
},
|
2303 |
-
{
|
2304 |
-
"epoch": 0.82,
|
2305 |
-
"learning_rate": 0.00017081235649864395,
|
2306 |
-
"loss": 2.1327,
|
2307 |
-
"step": 766
|
2308 |
-
},
|
2309 |
-
{
|
2310 |
-
"epoch": 0.82,
|
2311 |
-
"learning_rate": 0.00017064960079602297,
|
2312 |
-
"loss": 2.2666,
|
2313 |
-
"step": 768
|
2314 |
-
},
|
2315 |
-
{
|
2316 |
-
"epoch": 0.82,
|
2317 |
-
"learning_rate": 0.00017048647058446505,
|
2318 |
-
"loss": 2.2871,
|
2319 |
-
"step": 770
|
2320 |
-
},
|
2321 |
-
{
|
2322 |
-
"epoch": 0.82,
|
2323 |
-
"learning_rate": 0.00017032296672871283,
|
2324 |
-
"loss": 2.2055,
|
2325 |
-
"step": 772
|
2326 |
-
},
|
2327 |
-
{
|
2328 |
-
"epoch": 0.82,
|
2329 |
-
"learning_rate": 0.00017015909009548966,
|
2330 |
-
"loss": 2.1066,
|
2331 |
-
"step": 774
|
2332 |
-
},
|
2333 |
-
{
|
2334 |
-
"epoch": 0.83,
|
2335 |
-
"learning_rate": 0.00016999484155349483,
|
2336 |
-
"loss": 2.1887,
|
2337 |
-
"step": 776
|
2338 |
-
},
|
2339 |
-
{
|
2340 |
-
"epoch": 0.83,
|
2341 |
-
"learning_rate": 0.00016983022197339923,
|
2342 |
-
"loss": 2.1014,
|
2343 |
-
"step": 778
|
2344 |
-
},
|
2345 |
-
{
|
2346 |
-
"epoch": 0.83,
|
2347 |
-
"learning_rate": 0.00016966523222784058,
|
2348 |
-
"loss": 2.1138,
|
2349 |
-
"step": 780
|
2350 |
-
},
|
2351 |
-
{
|
2352 |
-
"epoch": 0.83,
|
2353 |
-
"learning_rate": 0.00016949987319141868,
|
2354 |
-
"loss": 1.8663,
|
2355 |
-
"step": 782
|
2356 |
-
},
|
2357 |
-
{
|
2358 |
-
"epoch": 0.84,
|
2359 |
-
"learning_rate": 0.0001693341457406911,
|
2360 |
-
"loss": 1.7421,
|
2361 |
-
"step": 784
|
2362 |
-
},
|
2363 |
-
{
|
2364 |
-
"epoch": 0.84,
|
2365 |
-
"learning_rate": 0.00016916805075416823,
|
2366 |
-
"loss": 1.8293,
|
2367 |
-
"step": 786
|
2368 |
-
},
|
2369 |
-
{
|
2370 |
-
"epoch": 0.84,
|
2371 |
-
"learning_rate": 0.0001690015891123088,
|
2372 |
-
"loss": 1.8993,
|
2373 |
-
"step": 788
|
2374 |
-
},
|
2375 |
-
{
|
2376 |
-
"epoch": 0.84,
|
2377 |
-
"learning_rate": 0.00016883476169751518,
|
2378 |
-
"loss": 1.8486,
|
2379 |
-
"step": 790
|
2380 |
-
},
|
2381 |
-
{
|
2382 |
-
"epoch": 0.84,
|
2383 |
-
"learning_rate": 0.0001686675693941286,
|
2384 |
-
"loss": 1.7033,
|
2385 |
-
"step": 792
|
2386 |
-
},
|
2387 |
-
{
|
2388 |
-
"epoch": 0.85,
|
2389 |
-
"learning_rate": 0.00016850001308842458,
|
2390 |
-
"loss": 1.5906,
|
2391 |
-
"step": 794
|
2392 |
-
},
|
2393 |
-
{
|
2394 |
-
"epoch": 0.85,
|
2395 |
-
"learning_rate": 0.00016833209366860826,
|
2396 |
-
"loss": 1.4477,
|
2397 |
-
"step": 796
|
2398 |
-
},
|
2399 |
-
{
|
2400 |
-
"epoch": 0.85,
|
2401 |
-
"learning_rate": 0.00016816381202480946,
|
2402 |
-
"loss": 1.8339,
|
2403 |
-
"step": 798
|
2404 |
-
},
|
2405 |
-
{
|
2406 |
-
"epoch": 0.85,
|
2407 |
-
"learning_rate": 0.00016799516904907828,
|
2408 |
-
"loss": 1.6691,
|
2409 |
-
"step": 800
|
2410 |
-
},
|
2411 |
-
{
|
2412 |
-
"epoch": 0.85,
|
2413 |
-
"learning_rate": 0.00016782616563538016,
|
2414 |
-
"loss": 2.4258,
|
2415 |
-
"step": 802
|
2416 |
-
},
|
2417 |
-
{
|
2418 |
-
"epoch": 0.86,
|
2419 |
-
"learning_rate": 0.0001676568026795912,
|
2420 |
-
"loss": 2.3445,
|
2421 |
-
"step": 804
|
2422 |
-
},
|
2423 |
-
{
|
2424 |
-
"epoch": 0.86,
|
2425 |
-
"learning_rate": 0.00016748708107949337,
|
2426 |
-
"loss": 2.137,
|
2427 |
-
"step": 806
|
2428 |
-
},
|
2429 |
-
{
|
2430 |
-
"epoch": 0.86,
|
2431 |
-
"learning_rate": 0.00016731700173476988,
|
2432 |
-
"loss": 2.2656,
|
2433 |
-
"step": 808
|
2434 |
-
},
|
2435 |
-
{
|
2436 |
-
"epoch": 0.86,
|
2437 |
-
"learning_rate": 0.00016714656554700022,
|
2438 |
-
"loss": 2.2945,
|
2439 |
-
"step": 810
|
2440 |
-
},
|
2441 |
-
{
|
2442 |
-
"epoch": 0.87,
|
2443 |
-
"learning_rate": 0.0001669757734196556,
|
2444 |
-
"loss": 2.0641,
|
2445 |
-
"step": 812
|
2446 |
-
},
|
2447 |
-
{
|
2448 |
-
"epoch": 0.87,
|
2449 |
-
"learning_rate": 0.0001668046262580939,
|
2450 |
-
"loss": 2.0698,
|
2451 |
-
"step": 814
|
2452 |
-
},
|
2453 |
-
{
|
2454 |
-
"epoch": 0.87,
|
2455 |
-
"learning_rate": 0.00016663312496955517,
|
2456 |
-
"loss": 2.1787,
|
2457 |
-
"step": 816
|
2458 |
-
},
|
2459 |
-
{
|
2460 |
-
"epoch": 0.87,
|
2461 |
-
"learning_rate": 0.00016646127046315653,
|
2462 |
-
"loss": 2.0295,
|
2463 |
-
"step": 818
|
2464 |
-
},
|
2465 |
-
{
|
2466 |
-
"epoch": 0.87,
|
2467 |
-
"learning_rate": 0.0001662890636498875,
|
2468 |
-
"loss": 2.2736,
|
2469 |
-
"step": 820
|
2470 |
-
},
|
2471 |
-
{
|
2472 |
-
"epoch": 0.88,
|
2473 |
-
"learning_rate": 0.00016611650544260526,
|
2474 |
-
"loss": 2.1205,
|
2475 |
-
"step": 822
|
2476 |
-
},
|
2477 |
-
{
|
2478 |
-
"epoch": 0.88,
|
2479 |
-
"learning_rate": 0.00016594359675602962,
|
2480 |
-
"loss": 2.1249,
|
2481 |
-
"step": 824
|
2482 |
-
},
|
2483 |
-
{
|
2484 |
-
"epoch": 0.88,
|
2485 |
-
"learning_rate": 0.00016577033850673824,
|
2486 |
-
"loss": 1.9355,
|
2487 |
-
"step": 826
|
2488 |
-
},
|
2489 |
-
{
|
2490 |
-
"epoch": 0.88,
|
2491 |
-
"learning_rate": 0.00016559673161316188,
|
2492 |
-
"loss": 1.8423,
|
2493 |
-
"step": 828
|
2494 |
-
},
|
2495 |
-
{
|
2496 |
-
"epoch": 0.88,
|
2497 |
-
"learning_rate": 0.00016542277699557934,
|
2498 |
-
"loss": 2.0979,
|
2499 |
-
"step": 830
|
2500 |
-
},
|
2501 |
-
{
|
2502 |
-
"epoch": 0.89,
|
2503 |
-
"learning_rate": 0.00016524847557611278,
|
2504 |
-
"loss": 1.7747,
|
2505 |
-
"step": 832
|
2506 |
-
},
|
2507 |
-
{
|
2508 |
-
"epoch": 0.89,
|
2509 |
-
"learning_rate": 0.00016507382827872264,
|
2510 |
-
"loss": 1.7137,
|
2511 |
-
"step": 834
|
2512 |
-
},
|
2513 |
-
{
|
2514 |
-
"epoch": 0.89,
|
2515 |
-
"learning_rate": 0.0001648988360292029,
|
2516 |
-
"loss": 1.9926,
|
2517 |
-
"step": 836
|
2518 |
-
},
|
2519 |
-
{
|
2520 |
-
"epoch": 0.89,
|
2521 |
-
"learning_rate": 0.0001647234997551761,
|
2522 |
-
"loss": 2.033,
|
2523 |
-
"step": 838
|
2524 |
-
},
|
2525 |
-
{
|
2526 |
-
"epoch": 0.9,
|
2527 |
-
"learning_rate": 0.00016454782038608835,
|
2528 |
-
"loss": 1.7298,
|
2529 |
-
"step": 840
|
2530 |
-
},
|
2531 |
-
{
|
2532 |
-
"epoch": 0.9,
|
2533 |
-
"learning_rate": 0.00016437179885320466,
|
2534 |
-
"loss": 1.8326,
|
2535 |
-
"step": 842
|
2536 |
-
},
|
2537 |
-
{
|
2538 |
-
"epoch": 0.9,
|
2539 |
-
"learning_rate": 0.00016419543608960367,
|
2540 |
-
"loss": 1.5744,
|
2541 |
-
"step": 844
|
2542 |
-
},
|
2543 |
-
{
|
2544 |
-
"epoch": 0.9,
|
2545 |
-
"learning_rate": 0.00016401873303017287,
|
2546 |
-
"loss": 1.471,
|
2547 |
-
"step": 846
|
2548 |
-
},
|
2549 |
-
{
|
2550 |
-
"epoch": 0.9,
|
2551 |
-
"learning_rate": 0.00016384169061160376,
|
2552 |
-
"loss": 1.7169,
|
2553 |
-
"step": 848
|
2554 |
-
},
|
2555 |
-
{
|
2556 |
-
"epoch": 0.91,
|
2557 |
-
"learning_rate": 0.00016366430977238667,
|
2558 |
-
"loss": 1.5636,
|
2559 |
-
"step": 850
|
2560 |
-
},
|
2561 |
-
{
|
2562 |
-
"epoch": 0.91,
|
2563 |
-
"learning_rate": 0.00016348659145280585,
|
2564 |
-
"loss": 2.3071,
|
2565 |
-
"step": 852
|
2566 |
-
},
|
2567 |
-
{
|
2568 |
-
"epoch": 0.91,
|
2569 |
-
"learning_rate": 0.00016330853659493456,
|
2570 |
-
"loss": 2.1993,
|
2571 |
-
"step": 854
|
2572 |
-
},
|
2573 |
-
{
|
2574 |
-
"epoch": 0.91,
|
2575 |
-
"learning_rate": 0.00016313014614263003,
|
2576 |
-
"loss": 2.1583,
|
2577 |
-
"step": 856
|
2578 |
-
},
|
2579 |
-
{
|
2580 |
-
"epoch": 0.91,
|
2581 |
-
"learning_rate": 0.0001629514210415284,
|
2582 |
-
"loss": 2.1996,
|
2583 |
-
"step": 858
|
2584 |
-
},
|
2585 |
-
{
|
2586 |
-
"epoch": 0.92,
|
2587 |
-
"learning_rate": 0.00016277236223903986,
|
2588 |
-
"loss": 2.309,
|
2589 |
-
"step": 860
|
2590 |
-
},
|
2591 |
-
{
|
2592 |
-
"epoch": 0.92,
|
2593 |
-
"learning_rate": 0.00016259297068434343,
|
2594 |
-
"loss": 2.2155,
|
2595 |
-
"step": 862
|
2596 |
-
},
|
2597 |
-
{
|
2598 |
-
"epoch": 0.92,
|
2599 |
-
"learning_rate": 0.0001624132473283821,
|
2600 |
-
"loss": 2.2474,
|
2601 |
-
"step": 864
|
2602 |
-
},
|
2603 |
-
{
|
2604 |
-
"epoch": 0.92,
|
2605 |
-
"learning_rate": 0.00016223319312385766,
|
2606 |
-
"loss": 2.2034,
|
2607 |
-
"step": 866
|
2608 |
-
},
|
2609 |
-
{
|
2610 |
-
"epoch": 0.93,
|
2611 |
-
"learning_rate": 0.00016205280902522576,
|
2612 |
-
"loss": 2.2016,
|
2613 |
-
"step": 868
|
2614 |
-
},
|
2615 |
-
{
|
2616 |
-
"epoch": 0.93,
|
2617 |
-
"learning_rate": 0.00016187209598869074,
|
2618 |
-
"loss": 2.133,
|
2619 |
-
"step": 870
|
2620 |
-
},
|
2621 |
-
{
|
2622 |
-
"epoch": 0.93,
|
2623 |
-
"learning_rate": 0.00016169105497220064,
|
2624 |
-
"loss": 2.1746,
|
2625 |
-
"step": 872
|
2626 |
-
},
|
2627 |
-
{
|
2628 |
-
"epoch": 0.93,
|
2629 |
-
"learning_rate": 0.00016150968693544215,
|
2630 |
-
"loss": 2.0806,
|
2631 |
-
"step": 874
|
2632 |
-
},
|
2633 |
-
{
|
2634 |
-
"epoch": 0.93,
|
2635 |
-
"learning_rate": 0.00016132799283983542,
|
2636 |
-
"loss": 2.2318,
|
2637 |
-
"step": 876
|
2638 |
-
},
|
2639 |
-
{
|
2640 |
-
"epoch": 0.94,
|
2641 |
-
"learning_rate": 0.000161145973648529,
|
2642 |
-
"loss": 2.0729,
|
2643 |
-
"step": 878
|
2644 |
-
},
|
2645 |
-
{
|
2646 |
-
"epoch": 0.94,
|
2647 |
-
"learning_rate": 0.0001609636303263948,
|
2648 |
-
"loss": 1.8199,
|
2649 |
-
"step": 880
|
2650 |
-
},
|
2651 |
-
{
|
2652 |
-
"epoch": 0.94,
|
2653 |
-
"learning_rate": 0.00016078096384002292,
|
2654 |
-
"loss": 2.0973,
|
2655 |
-
"step": 882
|
2656 |
-
},
|
2657 |
-
{
|
2658 |
-
"epoch": 0.94,
|
2659 |
-
"learning_rate": 0.00016059797515771652,
|
2660 |
-
"loss": 1.7302,
|
2661 |
-
"step": 884
|
2662 |
-
},
|
2663 |
-
{
|
2664 |
-
"epoch": 0.94,
|
2665 |
-
"learning_rate": 0.00016041466524948663,
|
2666 |
-
"loss": 1.9608,
|
2667 |
-
"step": 886
|
2668 |
-
},
|
2669 |
-
{
|
2670 |
-
"epoch": 0.95,
|
2671 |
-
"learning_rate": 0.00016023103508704725,
|
2672 |
-
"loss": 1.8483,
|
2673 |
-
"step": 888
|
2674 |
-
},
|
2675 |
-
{
|
2676 |
-
"epoch": 0.95,
|
2677 |
-
"learning_rate": 0.00016004708564380985,
|
2678 |
-
"loss": 1.9501,
|
2679 |
-
"step": 890
|
2680 |
-
},
|
2681 |
-
{
|
2682 |
-
"epoch": 0.95,
|
2683 |
-
"learning_rate": 0.0001598628178948785,
|
2684 |
-
"loss": 1.6526,
|
2685 |
-
"step": 892
|
2686 |
-
},
|
2687 |
-
{
|
2688 |
-
"epoch": 0.95,
|
2689 |
-
"learning_rate": 0.0001596782328170445,
|
2690 |
-
"loss": 1.4811,
|
2691 |
-
"step": 894
|
2692 |
-
},
|
2693 |
-
{
|
2694 |
-
"epoch": 0.96,
|
2695 |
-
"learning_rate": 0.00015949333138878138,
|
2696 |
-
"loss": 1.4048,
|
2697 |
-
"step": 896
|
2698 |
-
},
|
2699 |
-
{
|
2700 |
-
"epoch": 0.96,
|
2701 |
-
"learning_rate": 0.00015930811459023957,
|
2702 |
-
"loss": 1.5489,
|
2703 |
-
"step": 898
|
2704 |
-
},
|
2705 |
-
{
|
2706 |
-
"epoch": 0.96,
|
2707 |
-
"learning_rate": 0.00015912258340324126,
|
2708 |
-
"loss": 1.7197,
|
2709 |
-
"step": 900
|
2710 |
-
},
|
2711 |
-
{
|
2712 |
-
"epoch": 0.96,
|
2713 |
-
"learning_rate": 0.00015893673881127524,
|
2714 |
-
"loss": 2.1217,
|
2715 |
-
"step": 902
|
2716 |
-
},
|
2717 |
-
{
|
2718 |
-
"epoch": 0.96,
|
2719 |
-
"learning_rate": 0.00015875058179949151,
|
2720 |
-
"loss": 2.2086,
|
2721 |
-
"step": 904
|
2722 |
-
},
|
2723 |
-
{
|
2724 |
-
"epoch": 0.97,
|
2725 |
-
"learning_rate": 0.00015856411335469638,
|
2726 |
-
"loss": 2.094,
|
2727 |
-
"step": 906
|
2728 |
-
},
|
2729 |
-
{
|
2730 |
-
"epoch": 0.97,
|
2731 |
-
"learning_rate": 0.00015837733446534688,
|
2732 |
-
"loss": 2.3415,
|
2733 |
-
"step": 908
|
2734 |
-
},
|
2735 |
-
{
|
2736 |
-
"epoch": 0.97,
|
2737 |
-
"learning_rate": 0.00015819024612154575,
|
2738 |
-
"loss": 2.2378,
|
2739 |
-
"step": 910
|
2740 |
-
},
|
2741 |
-
{
|
2742 |
-
"epoch": 0.97,
|
2743 |
-
"learning_rate": 0.00015800284931503618,
|
2744 |
-
"loss": 2.1351,
|
2745 |
-
"step": 912
|
2746 |
-
},
|
2747 |
-
{
|
2748 |
-
"epoch": 0.97,
|
2749 |
-
"learning_rate": 0.0001578151450391964,
|
2750 |
-
"loss": 2.1795,
|
2751 |
-
"step": 914
|
2752 |
-
},
|
2753 |
-
{
|
2754 |
-
"epoch": 0.98,
|
2755 |
-
"learning_rate": 0.00015762713428903454,
|
2756 |
-
"loss": 2.0282,
|
2757 |
-
"step": 916
|
2758 |
-
},
|
2759 |
-
{
|
2760 |
-
"epoch": 0.98,
|
2761 |
-
"learning_rate": 0.00015743881806118342,
|
2762 |
-
"loss": 2.2858,
|
2763 |
-
"step": 918
|
2764 |
-
},
|
2765 |
-
{
|
2766 |
-
"epoch": 0.98,
|
2767 |
-
"learning_rate": 0.00015725019735389503,
|
2768 |
-
"loss": 2.0929,
|
2769 |
-
"step": 920
|
2770 |
-
},
|
2771 |
-
{
|
2772 |
-
"epoch": 0.98,
|
2773 |
-
"learning_rate": 0.00015706127316703557,
|
2774 |
-
"loss": 1.9913,
|
2775 |
-
"step": 922
|
2776 |
-
},
|
2777 |
-
{
|
2778 |
-
"epoch": 0.98,
|
2779 |
-
"learning_rate": 0.0001568720465020798,
|
2780 |
-
"loss": 2.0128,
|
2781 |
-
"step": 924
|
2782 |
-
},
|
2783 |
-
{
|
2784 |
-
"epoch": 0.99,
|
2785 |
-
"learning_rate": 0.00015668251836210595,
|
2786 |
-
"loss": 1.8792,
|
2787 |
-
"step": 926
|
2788 |
-
},
|
2789 |
-
{
|
2790 |
-
"epoch": 0.99,
|
2791 |
-
"learning_rate": 0.0001564926897517904,
|
2792 |
-
"loss": 1.6978,
|
2793 |
-
"step": 928
|
2794 |
-
},
|
2795 |
-
{
|
2796 |
-
"epoch": 0.99,
|
2797 |
-
"learning_rate": 0.0001563025616774022,
|
2798 |
-
"loss": 2.0335,
|
2799 |
-
"step": 930
|
2800 |
-
},
|
2801 |
-
{
|
2802 |
-
"epoch": 0.99,
|
2803 |
-
"learning_rate": 0.0001561121351467979,
|
2804 |
-
"loss": 1.845,
|
2805 |
-
"step": 932
|
2806 |
-
},
|
2807 |
-
{
|
2808 |
-
"epoch": 1.0,
|
2809 |
-
"learning_rate": 0.00015592141116941628,
|
2810 |
-
"loss": 1.3927,
|
2811 |
-
"step": 934
|
2812 |
-
},
|
2813 |
-
{
|
2814 |
-
"epoch": 1.0,
|
2815 |
-
"learning_rate": 0.00015573039075627256,
|
2816 |
-
"loss": 1.6165,
|
2817 |
-
"step": 936
|
2818 |
-
},
|
2819 |
-
{
|
2820 |
-
"epoch": 1.0,
|
2821 |
-
"learning_rate": 0.00015553907491995365,
|
2822 |
-
"loss": 1.6737,
|
2823 |
-
"step": 938
|
2824 |
}
|
2825 |
],
|
2826 |
"logging_steps": 2,
|
2827 |
-
"max_steps":
|
2828 |
"num_train_epochs": 3,
|
2829 |
"save_steps": 500,
|
2830 |
-
"total_flos":
|
2831 |
"trial_name": null,
|
2832 |
"trial_params": null
|
2833 |
}
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 2.983219390926041,
|
5 |
"eval_steps": 500,
|
6 |
+
"global_step": 300,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
10 |
"log_history": [
|
11 |
{
|
12 |
+
"epoch": 0.02,
|
13 |
+
"learning_rate": 4.4444444444444447e-05,
|
14 |
+
"loss": 2.7675,
|
15 |
"step": 2
|
16 |
},
|
17 |
{
|
18 |
+
"epoch": 0.04,
|
19 |
+
"learning_rate": 8.888888888888889e-05,
|
20 |
+
"loss": 2.6931,
|
21 |
"step": 4
|
22 |
},
|
23 |
{
|
24 |
+
"epoch": 0.06,
|
25 |
+
"learning_rate": 0.00013333333333333334,
|
26 |
+
"loss": 2.6733,
|
27 |
"step": 6
|
28 |
},
|
29 |
{
|
30 |
+
"epoch": 0.08,
|
31 |
+
"learning_rate": 0.00017777777777777779,
|
32 |
+
"loss": 2.5515,
|
33 |
"step": 8
|
34 |
},
|
35 |
{
|
36 |
+
"epoch": 0.1,
|
37 |
+
"learning_rate": 0.00019999417253661235,
|
38 |
+
"loss": 2.4826,
|
39 |
"step": 10
|
40 |
},
|
41 |
{
|
42 |
+
"epoch": 0.12,
|
43 |
+
"learning_rate": 0.00019994755690455152,
|
44 |
+
"loss": 2.5387,
|
45 |
"step": 12
|
46 |
},
|
47 |
{
|
48 |
+
"epoch": 0.14,
|
49 |
+
"learning_rate": 0.0001998543473718677,
|
50 |
+
"loss": 2.4596,
|
51 |
"step": 14
|
52 |
},
|
53 |
{
|
54 |
+
"epoch": 0.16,
|
55 |
+
"learning_rate": 0.00019971458739130598,
|
56 |
+
"loss": 2.4936,
|
57 |
"step": 16
|
58 |
},
|
59 |
{
|
60 |
+
"epoch": 0.18,
|
61 |
+
"learning_rate": 0.0001995283421166614,
|
62 |
+
"loss": 2.1189,
|
63 |
"step": 18
|
64 |
},
|
65 |
{
|
66 |
+
"epoch": 0.2,
|
67 |
+
"learning_rate": 0.00019929569837240564,
|
68 |
+
"loss": 1.9708,
|
69 |
"step": 20
|
70 |
},
|
71 |
{
|
72 |
+
"epoch": 0.22,
|
73 |
+
"learning_rate": 0.00019901676461321068,
|
74 |
+
"loss": 2.0963,
|
75 |
"step": 22
|
76 |
},
|
77 |
{
|
78 |
+
"epoch": 0.24,
|
79 |
+
"learning_rate": 0.00019869167087338907,
|
80 |
+
"loss": 2.0887,
|
81 |
"step": 24
|
82 |
},
|
83 |
{
|
84 |
+
"epoch": 0.26,
|
85 |
+
"learning_rate": 0.00019832056870627417,
|
86 |
+
"loss": 2.218,
|
87 |
"step": 26
|
88 |
},
|
89 |
{
|
90 |
+
"epoch": 0.28,
|
91 |
+
"learning_rate": 0.00019790363111356837,
|
92 |
+
"loss": 2.358,
|
93 |
"step": 28
|
94 |
},
|
95 |
{
|
96 |
+
"epoch": 0.3,
|
97 |
+
"learning_rate": 0.00019744105246469263,
|
98 |
+
"loss": 2.334,
|
99 |
"step": 30
|
100 |
},
|
101 |
{
|
102 |
+
"epoch": 0.32,
|
103 |
+
"learning_rate": 0.00019693304840617457,
|
104 |
+
"loss": 2.1776,
|
105 |
"step": 32
|
106 |
},
|
107 |
{
|
108 |
+
"epoch": 0.34,
|
109 |
+
"learning_rate": 0.00019637985576111778,
|
110 |
+
"loss": 2.3149,
|
111 |
"step": 34
|
112 |
},
|
113 |
{
|
114 |
+
"epoch": 0.36,
|
115 |
+
"learning_rate": 0.00019578173241879872,
|
116 |
+
"loss": 2.1172,
|
117 |
"step": 36
|
118 |
},
|
119 |
{
|
120 |
+
"epoch": 0.38,
|
121 |
+
"learning_rate": 0.00019513895721444286,
|
122 |
+
"loss": 2.183,
|
123 |
"step": 38
|
124 |
},
|
125 |
{
|
126 |
+
"epoch": 0.4,
|
127 |
+
"learning_rate": 0.00019445182979923654,
|
128 |
+
"loss": 2.2688,
|
129 |
"step": 40
|
130 |
},
|
131 |
{
|
132 |
+
"epoch": 0.42,
|
133 |
+
"learning_rate": 0.00019372067050063438,
|
134 |
+
"loss": 2.2487,
|
135 |
"step": 42
|
136 |
},
|
137 |
{
|
138 |
+
"epoch": 0.44,
|
139 |
+
"learning_rate": 0.00019294582017302797,
|
140 |
+
"loss": 2.1218,
|
141 |
"step": 44
|
142 |
},
|
143 |
{
|
144 |
+
"epoch": 0.46,
|
145 |
+
"learning_rate": 0.0001921276400388451,
|
146 |
+
"loss": 2.1302,
|
147 |
"step": 46
|
148 |
},
|
149 |
{
|
150 |
+
"epoch": 0.48,
|
151 |
+
"learning_rate": 0.00019126651152015403,
|
152 |
+
"loss": 2.1458,
|
153 |
"step": 48
|
154 |
},
|
155 |
{
|
156 |
+
"epoch": 0.5,
|
157 |
+
"learning_rate": 0.00019036283606085053,
|
158 |
+
"loss": 2.1425,
|
159 |
"step": 50
|
160 |
},
|
161 |
{
|
162 |
+
"epoch": 0.52,
|
163 |
+
"learning_rate": 0.00018941703493951164,
|
164 |
+
"loss": 2.2523,
|
165 |
"step": 52
|
166 |
},
|
167 |
{
|
168 |
+
"epoch": 0.54,
|
169 |
+
"learning_rate": 0.00018842954907300236,
|
170 |
+
"loss": 2.1411,
|
171 |
"step": 54
|
172 |
},
|
173 |
{
|
174 |
+
"epoch": 0.56,
|
175 |
+
"learning_rate": 0.0001874008388109276,
|
176 |
+
"loss": 2.2491,
|
177 |
"step": 56
|
178 |
},
|
179 |
{
|
180 |
+
"epoch": 0.58,
|
181 |
+
"learning_rate": 0.00018633138372102468,
|
182 |
+
"loss": 2.1121,
|
183 |
"step": 58
|
184 |
},
|
185 |
{
|
186 |
+
"epoch": 0.6,
|
187 |
+
"learning_rate": 0.00018522168236559695,
|
188 |
+
"loss": 2.2793,
|
189 |
"step": 60
|
190 |
},
|
191 |
{
|
192 |
+
"epoch": 0.62,
|
193 |
+
"learning_rate": 0.00018407225206909208,
|
194 |
+
"loss": 2.1577,
|
195 |
"step": 62
|
196 |
},
|
197 |
{
|
198 |
+
"epoch": 0.64,
|
199 |
+
"learning_rate": 0.00018288362867693414,
|
200 |
+
"loss": 2.1961,
|
201 |
"step": 64
|
202 |
},
|
203 |
{
|
204 |
+
"epoch": 0.66,
|
205 |
+
"learning_rate": 0.0001816563663057211,
|
206 |
+
"loss": 2.1825,
|
207 |
"step": 66
|
208 |
},
|
209 |
{
|
210 |
+
"epoch": 0.68,
|
211 |
+
"learning_rate": 0.000180391037084905,
|
212 |
+
"loss": 2.0917,
|
213 |
"step": 68
|
214 |
},
|
215 |
{
|
216 |
+
"epoch": 0.7,
|
217 |
+
"learning_rate": 0.00017908823089007457,
|
218 |
+
"loss": 2.0896,
|
219 |
"step": 70
|
220 |
},
|
221 |
{
|
222 |
+
"epoch": 0.72,
|
223 |
+
"learning_rate": 0.00017774855506796496,
|
224 |
+
"loss": 2.0877,
|
225 |
"step": 72
|
226 |
},
|
227 |
{
|
228 |
+
"epoch": 0.74,
|
229 |
+
"learning_rate": 0.0001763726341533227,
|
230 |
+
"loss": 2.1369,
|
231 |
"step": 74
|
232 |
},
|
233 |
{
|
234 |
+
"epoch": 0.76,
|
235 |
+
"learning_rate": 0.0001749611095777581,
|
236 |
+
"loss": 2.0917,
|
237 |
"step": 76
|
238 |
},
|
239 |
{
|
240 |
+
"epoch": 0.78,
|
241 |
+
"learning_rate": 0.00017351463937072004,
|
242 |
+
"loss": 2.3862,
|
243 |
"step": 78
|
244 |
},
|
245 |
{
|
246 |
+
"epoch": 0.8,
|
247 |
+
"learning_rate": 0.000172033897852734,
|
248 |
+
"loss": 2.1755,
|
249 |
"step": 80
|
250 |
},
|
251 |
{
|
252 |
+
"epoch": 0.82,
|
253 |
+
"learning_rate": 0.0001705195753210446,
|
254 |
+
"loss": 2.1165,
|
255 |
"step": 82
|
256 |
},
|
257 |
{
|
258 |
+
"epoch": 0.84,
|
259 |
+
"learning_rate": 0.00016897237772781044,
|
260 |
+
"loss": 2.1959,
|
261 |
"step": 84
|
262 |
},
|
263 |
{
|
264 |
+
"epoch": 0.86,
|
265 |
+
"learning_rate": 0.00016739302635100108,
|
266 |
+
"loss": 2.1884,
|
267 |
"step": 86
|
268 |
},
|
269 |
{
|
270 |
+
"epoch": 0.88,
|
271 |
+
"learning_rate": 0.00016578225745814907,
|
272 |
+
"loss": 2.2531,
|
273 |
"step": 88
|
274 |
},
|
275 |
{
|
276 |
+
"epoch": 0.89,
|
277 |
+
"learning_rate": 0.000164140821963114,
|
278 |
+
"loss": 2.2549,
|
279 |
"step": 90
|
280 |
},
|
281 |
{
|
282 |
+
"epoch": 0.91,
|
283 |
+
"learning_rate": 0.00016246948507601914,
|
284 |
+
"loss": 2.0919,
|
285 |
"step": 92
|
286 |
},
|
287 |
{
|
288 |
+
"epoch": 0.93,
|
289 |
+
"learning_rate": 0.0001607690259465229,
|
290 |
+
"loss": 2.0073,
|
291 |
"step": 94
|
292 |
},
|
293 |
{
|
294 |
+
"epoch": 0.95,
|
295 |
+
"learning_rate": 0.00015904023730059228,
|
296 |
+
"loss": 2.1638,
|
297 |
"step": 96
|
298 |
},
|
299 |
{
|
300 |
+
"epoch": 0.97,
|
301 |
+
"learning_rate": 0.000157283925070947,
|
302 |
+
"loss": 2.0956,
|
303 |
"step": 98
|
304 |
},
|
305 |
{
|
306 |
+
"epoch": 0.99,
|
307 |
+
"learning_rate": 0.000155500908021347,
|
308 |
+
"loss": 1.9043,
|
309 |
"step": 100
|
310 |
},
|
311 |
{
|
312 |
+
"epoch": 1.01,
|
313 |
+
"learning_rate": 0.0001536920173648984,
|
314 |
+
"loss": 2.0836,
|
315 |
"step": 102
|
316 |
},
|
317 |
{
|
318 |
+
"epoch": 1.03,
|
319 |
+
"learning_rate": 0.0001518580963765555,
|
320 |
+
"loss": 2.1632,
|
321 |
"step": 104
|
322 |
},
|
323 |
{
|
324 |
+
"epoch": 1.05,
|
325 |
+
"learning_rate": 0.00015000000000000001,
|
326 |
+
"loss": 2.0727,
|
327 |
"step": 106
|
328 |
},
|
329 |
{
|
330 |
+
"epoch": 1.07,
|
331 |
+
"learning_rate": 0.00014811859444908052,
|
332 |
+
"loss": 2.2505,
|
333 |
"step": 108
|
334 |
},
|
335 |
{
|
336 |
+
"epoch": 1.09,
|
337 |
+
"learning_rate": 0.0001462147568039977,
|
338 |
+
"loss": 2.1901,
|
339 |
"step": 110
|
340 |
},
|
341 |
{
|
342 |
+
"epoch": 1.11,
|
343 |
+
"learning_rate": 0.00014428937460242417,
|
344 |
+
"loss": 2.1774,
|
345 |
"step": 112
|
346 |
},
|
347 |
{
|
348 |
+
"epoch": 1.13,
|
349 |
+
"learning_rate": 0.00014234334542574906,
|
350 |
+
"loss": 2.1062,
|
351 |
"step": 114
|
352 |
},
|
353 |
{
|
354 |
+
"epoch": 1.15,
|
355 |
+
"learning_rate": 0.00014037757648064018,
|
356 |
+
"loss": 2.0187,
|
357 |
"step": 116
|
358 |
},
|
359 |
{
|
360 |
+
"epoch": 1.17,
|
361 |
+
"learning_rate": 0.00013839298417611963,
|
362 |
+
"loss": 2.0678,
|
363 |
"step": 118
|
364 |
},
|
365 |
{
|
366 |
+
"epoch": 1.19,
|
367 |
+
"learning_rate": 0.00013639049369634876,
|
368 |
+
"loss": 2.0362,
|
369 |
"step": 120
|
370 |
},
|
371 |
{
|
372 |
+
"epoch": 1.21,
|
373 |
+
"learning_rate": 0.00013437103856932264,
|
374 |
+
"loss": 2.0308,
|
375 |
"step": 122
|
376 |
},
|
377 |
{
|
378 |
+
"epoch": 1.23,
|
379 |
+
"learning_rate": 0.00013233556023167485,
|
380 |
+
"loss": 1.9568,
|
381 |
"step": 124
|
382 |
},
|
383 |
{
|
384 |
+
"epoch": 1.25,
|
385 |
+
"learning_rate": 0.00013028500758979506,
|
386 |
+
"loss": 2.1364,
|
387 |
"step": 126
|
388 |
},
|
389 |
{
|
390 |
+
"epoch": 1.27,
|
391 |
+
"learning_rate": 0.00012822033657746478,
|
392 |
+
"loss": 2.2107,
|
393 |
"step": 128
|
394 |
},
|
395 |
{
|
396 |
+
"epoch": 1.29,
|
397 |
+
"learning_rate": 0.00012614250971021657,
|
398 |
+
"loss": 2.0798,
|
399 |
"step": 130
|
400 |
},
|
401 |
{
|
402 |
+
"epoch": 1.31,
|
403 |
+
"learning_rate": 0.00012405249563662537,
|
404 |
+
"loss": 2.2118,
|
405 |
"step": 132
|
406 |
},
|
407 |
{
|
408 |
+
"epoch": 1.33,
|
409 |
+
"learning_rate": 0.00012195126868674051,
|
410 |
+
"loss": 2.1874,
|
411 |
"step": 134
|
412 |
},
|
413 |
{
|
414 |
+
"epoch": 1.35,
|
415 |
+
"learning_rate": 0.000119839808417869,
|
416 |
+
"loss": 2.1919,
|
417 |
"step": 136
|
418 |
},
|
419 |
{
|
420 |
+
"epoch": 1.37,
|
421 |
+
"learning_rate": 0.0001177190991579223,
|
422 |
+
"loss": 2.0894,
|
423 |
"step": 138
|
424 |
},
|
425 |
{
|
426 |
+
"epoch": 1.39,
|
427 |
+
"learning_rate": 0.00011559012954653865,
|
428 |
+
"loss": 2.1773,
|
429 |
"step": 140
|
430 |
},
|
431 |
{
|
432 |
+
"epoch": 1.41,
|
433 |
+
"learning_rate": 0.00011345389207419588,
|
434 |
+
"loss": 2.2115,
|
435 |
"step": 142
|
436 |
},
|
437 |
{
|
438 |
+
"epoch": 1.43,
|
439 |
+
"learning_rate": 0.00011131138261952845,
|
440 |
+
"loss": 1.9207,
|
441 |
"step": 144
|
442 |
},
|
443 |
{
|
444 |
+
"epoch": 1.45,
|
445 |
+
"learning_rate": 0.0001091635999850655,
|
446 |
+
"loss": 1.9258,
|
447 |
"step": 146
|
448 |
},
|
449 |
{
|
450 |
+
"epoch": 1.47,
|
451 |
+
"learning_rate": 0.00010701154543160541,
|
452 |
+
"loss": 1.9662,
|
453 |
"step": 148
|
454 |
},
|
455 |
{
|
456 |
+
"epoch": 1.49,
|
457 |
+
"learning_rate": 0.00010485622221144484,
|
458 |
+
"loss": 2.0598,
|
459 |
"step": 150
|
460 |
},
|
461 |
{
|
462 |
+
"epoch": 1.51,
|
463 |
+
"learning_rate": 0.00010269863510067872,
|
464 |
+
"loss": 1.933,
|
465 |
"step": 152
|
466 |
},
|
467 |
{
|
468 |
+
"epoch": 1.53,
|
469 |
+
"learning_rate": 0.00010053978993079045,
|
470 |
+
"loss": 2.2902,
|
471 |
"step": 154
|
472 |
},
|
473 |
{
|
474 |
+
"epoch": 1.55,
|
475 |
+
"learning_rate": 9.838069311974986e-05,
|
476 |
+
"loss": 2.1651,
|
477 |
"step": 156
|
478 |
},
|
479 |
{
|
480 |
+
"epoch": 1.57,
|
481 |
+
"learning_rate": 9.622235120283769e-05,
|
482 |
+
"loss": 2.0562,
|
483 |
"step": 158
|
484 |
},
|
485 |
{
|
486 |
+
"epoch": 1.59,
|
487 |
+
"learning_rate": 9.406577036341548e-05,
|
488 |
+
"loss": 2.1876,
|
489 |
"step": 160
|
490 |
},
|
491 |
{
|
492 |
+
"epoch": 1.61,
|
493 |
+
"learning_rate": 9.19119559638596e-05,
|
494 |
+
"loss": 2.1455,
|
495 |
"step": 162
|
496 |
},
|
497 |
{
|
498 |
+
"epoch": 1.63,
|
499 |
+
"learning_rate": 8.976191207687775e-05,
|
500 |
+
"loss": 2.0586,
|
501 |
"step": 164
|
502 |
},
|
503 |
{
|
504 |
+
"epoch": 1.65,
|
505 |
+
"learning_rate": 8.7616641017427e-05,
|
506 |
+
"loss": 2.0516,
|
507 |
"step": 166
|
508 |
},
|
509 |
{
|
510 |
+
"epoch": 1.67,
|
511 |
+
"learning_rate": 8.5477142875451e-05,
|
512 |
+
"loss": 2.1687,
|
513 |
"step": 168
|
514 |
},
|
515 |
{
|
516 |
+
"epoch": 1.69,
|
517 |
+
"learning_rate": 8.334441504965455e-05,
|
518 |
+
"loss": 1.8243,
|
519 |
"step": 170
|
520 |
},
|
521 |
{
|
522 |
+
"epoch": 1.71,
|
523 |
+
"learning_rate": 8.1219451782533e-05,
|
524 |
+
"loss": 2.011,
|
525 |
"step": 172
|
526 |
},
|
527 |
{
|
528 |
+
"epoch": 1.73,
|
529 |
+
"learning_rate": 7.91032436968725e-05,
|
530 |
+
"loss": 2.0671,
|
531 |
"step": 174
|
532 |
},
|
533 |
{
|
534 |
+
"epoch": 1.75,
|
535 |
+
"learning_rate": 7.699677733393826e-05,
|
536 |
+
"loss": 2.1099,
|
537 |
"step": 176
|
538 |
},
|
539 |
{
|
540 |
+
"epoch": 1.77,
|
541 |
+
"learning_rate": 7.490103469356513e-05,
|
542 |
+
"loss": 2.2192,
|
543 |
"step": 178
|
544 |
},
|
545 |
{
|
546 |
+
"epoch": 1.79,
|
547 |
+
"learning_rate": 7.281699277636572e-05,
|
548 |
+
"loss": 2.1303,
|
549 |
"step": 180
|
550 |
},
|
551 |
{
|
552 |
+
"epoch": 1.81,
|
553 |
+
"learning_rate": 7.07456231282686e-05,
|
554 |
+
"loss": 2.2478,
|
555 |
"step": 182
|
556 |
},
|
557 |
{
|
558 |
+
"epoch": 1.83,
|
559 |
+
"learning_rate": 6.868789138759976e-05,
|
560 |
+
"loss": 2.2372,
|
561 |
"step": 184
|
562 |
},
|
563 |
{
|
564 |
+
"epoch": 1.85,
|
565 |
+
"learning_rate": 6.664475683491796e-05,
|
566 |
+
"loss": 2.0905,
|
567 |
"step": 186
|
568 |
},
|
569 |
{
|
570 |
+
"epoch": 1.87,
|
571 |
+
"learning_rate": 6.461717194581393e-05,
|
572 |
+
"loss": 2.1761,
|
573 |
"step": 188
|
574 |
},
|
575 |
{
|
576 |
+
"epoch": 1.89,
|
577 |
+
"learning_rate": 6.260608194688206e-05,
|
578 |
+
"loss": 2.2156,
|
579 |
"step": 190
|
580 |
},
|
581 |
{
|
582 |
+
"epoch": 1.91,
|
583 |
+
"learning_rate": 6.061242437507131e-05,
|
584 |
+
"loss": 2.0831,
|
585 |
"step": 192
|
586 |
},
|
587 |
{
|
588 |
+
"epoch": 1.93,
|
589 |
+
"learning_rate": 5.863712864062089e-05,
|
590 |
+
"loss": 1.9666,
|
591 |
"step": 194
|
592 |
},
|
593 |
{
|
594 |
+
"epoch": 1.95,
|
595 |
+
"learning_rate": 5.668111559378471e-05,
|
596 |
+
"loss": 2.0193,
|
597 |
"step": 196
|
598 |
},
|
599 |
{
|
600 |
+
"epoch": 1.97,
|
601 |
+
"learning_rate": 5.474529709554612e-05,
|
602 |
+
"loss": 2.0009,
|
603 |
"step": 198
|
604 |
},
|
605 |
{
|
606 |
+
"epoch": 1.99,
|
607 |
+
"learning_rate": 5.283057559252341e-05,
|
608 |
+
"loss": 1.8429,
|
609 |
"step": 200
|
610 |
},
|
611 |
{
|
612 |
+
"epoch": 2.01,
|
613 |
+
"learning_rate": 5.0937843696263966e-05,
|
614 |
+
"loss": 1.825,
|
615 |
"step": 202
|
616 |
},
|
617 |
{
|
618 |
+
"epoch": 2.03,
|
619 |
+
"learning_rate": 4.9067983767123736e-05,
|
620 |
+
"loss": 2.1982,
|
621 |
"step": 204
|
622 |
},
|
623 |
{
|
624 |
+
"epoch": 2.05,
|
625 |
+
"learning_rate": 4.722186750292511e-05,
|
626 |
+
"loss": 2.2024,
|
627 |
"step": 206
|
628 |
},
|
629 |
{
|
630 |
+
"epoch": 2.07,
|
631 |
+
"learning_rate": 4.540035553258619e-05,
|
632 |
+
"loss": 2.0208,
|
633 |
"step": 208
|
634 |
},
|
635 |
{
|
636 |
+
"epoch": 2.09,
|
637 |
+
"learning_rate": 4.360429701490934e-05,
|
638 |
+
"loss": 2.0707,
|
639 |
"step": 210
|
640 |
},
|
641 |
{
|
642 |
+
"epoch": 2.11,
|
643 |
+
"learning_rate": 4.183452924271776e-05,
|
644 |
+
"loss": 2.0485,
|
645 |
"step": 212
|
646 |
},
|
647 |
{
|
648 |
+
"epoch": 2.13,
|
649 |
+
"learning_rate": 4.009187725252309e-05,
|
650 |
+
"loss": 2.1584,
|
651 |
"step": 214
|
652 |
},
|
653 |
{
|
654 |
+
"epoch": 2.15,
|
655 |
+
"learning_rate": 3.8377153439907266e-05,
|
656 |
+
"loss": 2.1151,
|
657 |
"step": 216
|
658 |
},
|
659 |
{
|
660 |
+
"epoch": 2.17,
|
661 |
+
"learning_rate": 3.669115718079702e-05,
|
662 |
+
"loss": 1.9831,
|
663 |
"step": 218
|
664 |
},
|
665 |
{
|
666 |
+
"epoch": 2.19,
|
667 |
+
"learning_rate": 3.503467445880789e-05,
|
668 |
+
"loss": 1.9593,
|
669 |
"step": 220
|
670 |
},
|
671 |
{
|
672 |
+
"epoch": 2.21,
|
673 |
+
"learning_rate": 3.340847749883191e-05,
|
674 |
+
"loss": 1.9555,
|
675 |
"step": 222
|
676 |
},
|
677 |
{
|
678 |
+
"epoch": 2.23,
|
679 |
+
"learning_rate": 3.1813324407038825e-05,
|
680 |
+
"loss": 1.9759,
|
681 |
"step": 224
|
682 |
},
|
683 |
{
|
684 |
+
"epoch": 2.25,
|
685 |
+
"learning_rate": 3.0249958817459722e-05,
|
686 |
+
"loss": 1.9367,
|
687 |
"step": 226
|
688 |
},
|
689 |
{
|
690 |
+
"epoch": 2.27,
|
691 |
+
"learning_rate": 2.8719109545317103e-05,
|
692 |
+
"loss": 2.0455,
|
693 |
"step": 228
|
694 |
},
|
695 |
{
|
696 |
+
"epoch": 2.29,
|
697 |
+
"learning_rate": 2.722149024726307e-05,
|
698 |
+
"loss": 2.0912,
|
699 |
"step": 230
|
700 |
},
|
701 |
{
|
702 |
+
"epoch": 2.31,
|
703 |
+
"learning_rate": 2.5757799088684654e-05,
|
704 |
+
"loss": 2.0974,
|
705 |
"step": 232
|
706 |
},
|
707 |
{
|
708 |
+
"epoch": 2.33,
|
709 |
+
"learning_rate": 2.432871841823047e-05,
|
710 |
+
"loss": 2.1669,
|
711 |
"step": 234
|
712 |
},
|
713 |
{
|
714 |
+
"epoch": 2.35,
|
715 |
+
"learning_rate": 2.2934914449711087e-05,
|
716 |
+
"loss": 2.0861,
|
717 |
"step": 236
|
718 |
},
|
719 |
{
|
720 |
+
"epoch": 2.37,
|
721 |
+
"learning_rate": 2.157703695152109e-05,
|
722 |
+
"loss": 2.2297,
|
723 |
"step": 238
|
724 |
},
|
725 |
{
|
726 |
+
"epoch": 2.39,
|
727 |
+
"learning_rate": 2.025571894372794e-05,
|
728 |
+
"loss": 2.078,
|
729 |
"step": 240
|
730 |
},
|
731 |
{
|
732 |
+
"epoch": 2.41,
|
733 |
+
"learning_rate": 1.897157640296825e-05,
|
734 |
+
"loss": 2.0759,
|
735 |
"step": 242
|
736 |
},
|
737 |
{
|
738 |
+
"epoch": 2.43,
|
739 |
+
"learning_rate": 1.772520797528988e-05,
|
740 |
+
"loss": 1.957,
|
741 |
"step": 244
|
742 |
},
|
743 |
{
|
744 |
+
"epoch": 2.45,
|
745 |
+
"learning_rate": 1.65171946970729e-05,
|
746 |
+
"loss": 2.0201,
|
747 |
"step": 246
|
748 |
},
|
749 |
{
|
750 |
+
"epoch": 2.47,
|
751 |
+
"learning_rate": 1.534809972415998e-05,
|
752 |
+
"loss": 1.972,
|
753 |
"step": 248
|
754 |
},
|
755 |
{
|
756 |
+
"epoch": 2.49,
|
757 |
+
"learning_rate": 1.4218468069322578e-05,
|
758 |
+
"loss": 1.9812,
|
759 |
"step": 250
|
760 |
},
|
761 |
{
|
762 |
+
"epoch": 2.51,
|
763 |
+
"learning_rate": 1.3128826348184887e-05,
|
764 |
+
"loss": 1.7967,
|
765 |
"step": 252
|
766 |
},
|
767 |
{
|
768 |
+
"epoch": 2.53,
|
769 |
+
"learning_rate": 1.2079682533724379e-05,
|
770 |
+
"loss": 2.1202,
|
771 |
"step": 254
|
772 |
},
|
773 |
{
|
774 |
+
"epoch": 2.55,
|
775 |
+
"learning_rate": 1.1071525719463095e-05,
|
776 |
+
"loss": 2.2584,
|
777 |
"step": 256
|
778 |
},
|
779 |
{
|
780 |
+
"epoch": 2.57,
|
781 |
+
"learning_rate": 1.010482589146048e-05,
|
782 |
+
"loss": 2.1405,
|
783 |
"step": 258
|
784 |
},
|
785 |
{
|
786 |
+
"epoch": 2.59,
|
787 |
+
"learning_rate": 9.180033709213454e-06,
|
788 |
+
"loss": 2.1317,
|
789 |
"step": 260
|
790 |
},
|
791 |
{
|
792 |
+
"epoch": 2.61,
|
793 |
+
"learning_rate": 8.297580295566575e-06,
|
794 |
+
"loss": 2.0388,
|
795 |
"step": 262
|
796 |
},
|
797 |
{
|
798 |
+
"epoch": 2.63,
|
799 |
+
"learning_rate": 7.457877035729588e-06,
|
800 |
+
"loss": 2.0719,
|
801 |
"step": 264
|
802 |
},
|
803 |
{
|
804 |
+
"epoch": 2.65,
|
805 |
+
"learning_rate": 6.661315385496425e-06,
|
806 |
+
"loss": 2.1417,
|
807 |
"step": 266
|
808 |
},
|
809 |
{
|
810 |
+
"epoch": 2.67,
|
811 |
+
"learning_rate": 5.908266688755049e-06,
|
812 |
+
"loss": 2.0137,
|
813 |
"step": 268
|
814 |
},
|
815 |
{
|
816 |
+
"epoch": 2.68,
|
817 |
+
"learning_rate": 5.199082004372957e-06,
|
818 |
+
"loss": 2.0492,
|
819 |
"step": 270
|
820 |
},
|
821 |
{
|
822 |
+
"epoch": 2.7,
|
823 |
+
"learning_rate": 4.534091942539475e-06,
|
824 |
+
"loss": 1.9081,
|
825 |
"step": 272
|
826 |
},
|
827 |
{
|
828 |
+
"epoch": 2.72,
|
829 |
+
"learning_rate": 3.913606510640644e-06,
|
830 |
+
"loss": 2.0606,
|
831 |
"step": 274
|
832 |
},
|
833 |
{
|
834 |
+
"epoch": 2.74,
|
835 |
+
"learning_rate": 3.3379149687388867e-06,
|
836 |
+
"loss": 1.8957,
|
837 |
"step": 276
|
838 |
},
|
839 |
{
|
840 |
+
"epoch": 2.76,
|
841 |
+
"learning_rate": 2.8072856947248037e-06,
|
842 |
+
"loss": 2.1141,
|
843 |
"step": 278
|
844 |
},
|
845 |
{
|
846 |
+
"epoch": 2.78,
|
847 |
+
"learning_rate": 2.3219660592038285e-06,
|
848 |
+
"loss": 2.002,
|
849 |
"step": 280
|
850 |
},
|
851 |
{
|
852 |
+
"epoch": 2.8,
|
853 |
+
"learning_rate": 1.882182310176095e-06,
|
854 |
+
"loss": 2.0662,
|
855 |
"step": 282
|
856 |
},
|
857 |
{
|
858 |
+
"epoch": 2.82,
|
859 |
+
"learning_rate": 1.488139467563354e-06,
|
860 |
+
"loss": 1.9553,
|
861 |
"step": 284
|
862 |
},
|
863 |
{
|
864 |
+
"epoch": 2.84,
|
865 |
+
"learning_rate": 1.1400212276321376e-06,
|
866 |
+
"loss": 2.2417,
|
867 |
"step": 286
|
868 |
},
|
869 |
{
|
870 |
+
"epoch": 2.86,
|
871 |
+
"learning_rate": 8.379898773574924e-07,
|
872 |
+
"loss": 2.2588,
|
873 |
"step": 288
|
874 |
},
|
875 |
{
|
876 |
+
"epoch": 2.88,
|
877 |
+
"learning_rate": 5.821862187675775e-07,
|
878 |
+
"loss": 2.0636,
|
879 |
"step": 290
|
880 |
},
|
881 |
{
|
882 |
+
"epoch": 2.9,
|
883 |
+
"learning_rate": 3.727295033040035e-07,
|
884 |
+
"loss": 2.1176,
|
885 |
"step": 292
|
886 |
},
|
887 |
{
|
888 |
+
"epoch": 2.92,
|
889 |
+
"learning_rate": 2.0971737622883515e-07,
|
890 |
+
"loss": 1.9134,
|
891 |
"step": 294
|
892 |
},
|
893 |
{
|
894 |
+
"epoch": 2.94,
|
895 |
+
"learning_rate": 9.32258311039269e-08,
|
896 |
+
"loss": 1.8922,
|
897 |
"step": 296
|
898 |
},
|
899 |
{
|
900 |
+
"epoch": 2.96,
|
901 |
+
"learning_rate": 2.3309174364027907e-08,
|
902 |
+
"loss": 1.9239,
|
903 |
"step": 298
|
904 |
},
|
905 |
{
|
906 |
+
"epoch": 2.98,
|
907 |
+
"learning_rate": 0.0,
|
908 |
+
"loss": 1.9386,
|
909 |
"step": 300
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
910 |
}
|
911 |
],
|
912 |
"logging_steps": 2,
|
913 |
+
"max_steps": 300,
|
914 |
"num_train_epochs": 3,
|
915 |
"save_steps": 500,
|
916 |
+
"total_flos": 2.587049079238656e+16,
|
917 |
"trial_name": null,
|
918 |
"trial_params": null
|
919 |
}
|
training_args.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 4600
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:10710bc8fc564216df8d9aa27d769ef2fbb655a2fa656346c4949c447c166619
|
3 |
size 4600
|