End of training
Browse files
all_results.json
CHANGED
@@ -1,12 +1,12 @@
|
|
1 |
{
|
2 |
-
"epoch":
|
3 |
"eval_loss": 0.50146484375,
|
4 |
-
"eval_runtime": 168.
|
5 |
-
"eval_samples_per_second": 10.
|
6 |
-
"eval_steps_per_second": 0.
|
7 |
-
"eval_wer": 16.
|
8 |
-
"train_loss": 0.
|
9 |
-
"train_runtime":
|
10 |
-
"train_samples_per_second": 16.
|
11 |
-
"train_steps_per_second": 0.
|
12 |
}
|
|
|
1 |
{
|
2 |
+
"epoch": 12.44,
|
3 |
"eval_loss": 0.50146484375,
|
4 |
+
"eval_runtime": 168.6261,
|
5 |
+
"eval_samples_per_second": 10.058,
|
6 |
+
"eval_steps_per_second": 0.629,
|
7 |
+
"eval_wer": 16.465453194650816,
|
8 |
+
"train_loss": 0.00017441691160202026,
|
9 |
+
"train_runtime": 9920.3408,
|
10 |
+
"train_samples_per_second": 16.128,
|
11 |
+
"train_steps_per_second": 0.504
|
12 |
}
|
eval_results.json
CHANGED
@@ -1,8 +1,8 @@
|
|
1 |
{
|
2 |
-
"epoch":
|
3 |
"eval_loss": 0.50146484375,
|
4 |
-
"eval_runtime": 168.
|
5 |
-
"eval_samples_per_second": 10.
|
6 |
-
"eval_steps_per_second": 0.
|
7 |
-
"eval_wer": 16.
|
8 |
}
|
|
|
1 |
{
|
2 |
+
"epoch": 12.44,
|
3 |
"eval_loss": 0.50146484375,
|
4 |
+
"eval_runtime": 168.6261,
|
5 |
+
"eval_samples_per_second": 10.058,
|
6 |
+
"eval_steps_per_second": 0.629,
|
7 |
+
"eval_wer": 16.465453194650816
|
8 |
}
|
runs/Dec21_06-54-33_129-146-176-120/events.out.tfevents.1671616747.129-146-176-120.2641118.2
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7b04f94d8e4e12fc2210b0d84172428fcac3febc8318df4be02480122337a26c
|
3 |
+
size 358
|
train_results.json
CHANGED
@@ -1,7 +1,7 @@
|
|
1 |
{
|
2 |
-
"epoch":
|
3 |
-
"train_loss": 0.
|
4 |
-
"train_runtime":
|
5 |
-
"train_samples_per_second": 16.
|
6 |
-
"train_steps_per_second": 0.
|
7 |
}
|
|
|
1 |
{
|
2 |
+
"epoch": 12.44,
|
3 |
+
"train_loss": 0.00017441691160202026,
|
4 |
+
"train_runtime": 9920.3408,
|
5 |
+
"train_samples_per_second": 16.128,
|
6 |
+
"train_steps_per_second": 0.504
|
7 |
}
|
trainer_state.json
CHANGED
@@ -1,2515 +1,1270 @@
|
|
1 |
{
|
2 |
-
"best_metric": 16.
|
3 |
-
"best_model_checkpoint": "./checkpoint-
|
4 |
-
"epoch":
|
5 |
-
"global_step":
|
6 |
"is_hyper_param_search": false,
|
7 |
"is_local_process_zero": true,
|
8 |
"is_world_process_zero": true,
|
9 |
"log_history": [
|
10 |
{
|
11 |
"epoch": 0.06,
|
12 |
-
"learning_rate":
|
13 |
"loss": 0.0002,
|
14 |
"step": 25
|
15 |
},
|
16 |
{
|
17 |
"epoch": 0.12,
|
18 |
-
"learning_rate":
|
19 |
"loss": 0.0002,
|
20 |
"step": 50
|
21 |
},
|
22 |
{
|
23 |
"epoch": 0.19,
|
24 |
-
"learning_rate":
|
25 |
"loss": 0.0002,
|
26 |
"step": 75
|
27 |
},
|
28 |
{
|
29 |
"epoch": 0.25,
|
30 |
-
"learning_rate":
|
31 |
"loss": 0.0002,
|
32 |
"step": 100
|
33 |
},
|
34 |
{
|
35 |
"epoch": 0.31,
|
36 |
-
"learning_rate":
|
37 |
"loss": 0.0002,
|
38 |
"step": 125
|
39 |
},
|
40 |
{
|
41 |
"epoch": 0.37,
|
42 |
-
"learning_rate":
|
43 |
"loss": 0.0002,
|
44 |
"step": 150
|
45 |
},
|
46 |
{
|
47 |
"epoch": 0.44,
|
48 |
-
"learning_rate":
|
49 |
"loss": 0.0002,
|
50 |
"step": 175
|
51 |
},
|
52 |
{
|
53 |
"epoch": 0.5,
|
54 |
-
"learning_rate":
|
55 |
"loss": 0.0002,
|
56 |
"step": 200
|
57 |
},
|
58 |
{
|
59 |
"epoch": 0.56,
|
60 |
-
"learning_rate":
|
61 |
"loss": 0.0002,
|
62 |
"step": 225
|
63 |
},
|
64 |
{
|
65 |
"epoch": 0.62,
|
66 |
-
"learning_rate":
|
67 |
"loss": 0.0002,
|
68 |
"step": 250
|
69 |
},
|
70 |
{
|
71 |
"epoch": 0.68,
|
72 |
-
"learning_rate":
|
73 |
"loss": 0.0002,
|
74 |
"step": 275
|
75 |
},
|
76 |
{
|
77 |
"epoch": 0.75,
|
78 |
-
"learning_rate":
|
79 |
"loss": 0.0002,
|
80 |
"step": 300
|
81 |
},
|
82 |
{
|
83 |
"epoch": 0.81,
|
84 |
-
"learning_rate":
|
85 |
"loss": 0.0002,
|
86 |
"step": 325
|
87 |
},
|
88 |
{
|
89 |
"epoch": 0.87,
|
90 |
-
"learning_rate":
|
91 |
"loss": 0.0002,
|
92 |
"step": 350
|
93 |
},
|
94 |
{
|
95 |
"epoch": 0.93,
|
96 |
-
"learning_rate":
|
97 |
"loss": 0.0002,
|
98 |
"step": 375
|
99 |
},
|
100 |
{
|
101 |
"epoch": 1.0,
|
102 |
-
"learning_rate":
|
103 |
"loss": 0.0002,
|
104 |
"step": 400
|
105 |
},
|
106 |
{
|
107 |
"epoch": 1.06,
|
108 |
-
"learning_rate":
|
109 |
"loss": 0.0002,
|
110 |
"step": 425
|
111 |
},
|
112 |
{
|
113 |
"epoch": 1.12,
|
114 |
-
"learning_rate":
|
115 |
"loss": 0.0002,
|
116 |
"step": 450
|
117 |
},
|
118 |
{
|
119 |
"epoch": 1.18,
|
120 |
-
"learning_rate":
|
121 |
"loss": 0.0002,
|
122 |
"step": 475
|
123 |
},
|
124 |
{
|
125 |
"epoch": 1.24,
|
126 |
-
"learning_rate":
|
127 |
"loss": 0.0002,
|
128 |
"step": 500
|
129 |
},
|
130 |
{
|
131 |
"epoch": 1.31,
|
132 |
-
"learning_rate":
|
133 |
"loss": 0.0002,
|
134 |
"step": 525
|
135 |
},
|
136 |
{
|
137 |
"epoch": 1.37,
|
138 |
-
"learning_rate":
|
139 |
"loss": 0.0002,
|
140 |
"step": 550
|
141 |
},
|
142 |
{
|
143 |
"epoch": 1.43,
|
144 |
-
"learning_rate":
|
145 |
"loss": 0.0002,
|
146 |
"step": 575
|
147 |
},
|
148 |
{
|
149 |
"epoch": 1.49,
|
150 |
-
"learning_rate":
|
151 |
"loss": 0.0002,
|
152 |
"step": 600
|
153 |
},
|
154 |
{
|
155 |
"epoch": 1.55,
|
156 |
-
"learning_rate":
|
157 |
"loss": 0.0002,
|
158 |
"step": 625
|
159 |
},
|
160 |
{
|
161 |
"epoch": 1.62,
|
162 |
-
"learning_rate":
|
163 |
"loss": 0.0002,
|
164 |
"step": 650
|
165 |
},
|
166 |
{
|
167 |
"epoch": 1.68,
|
168 |
-
"learning_rate":
|
169 |
"loss": 0.0002,
|
170 |
"step": 675
|
171 |
},
|
172 |
{
|
173 |
"epoch": 1.74,
|
174 |
-
"learning_rate":
|
175 |
"loss": 0.0002,
|
176 |
"step": 700
|
177 |
},
|
178 |
{
|
179 |
"epoch": 1.8,
|
180 |
-
"learning_rate":
|
181 |
"loss": 0.0002,
|
182 |
"step": 725
|
183 |
},
|
184 |
{
|
185 |
"epoch": 1.87,
|
186 |
-
"learning_rate":
|
187 |
"loss": 0.0002,
|
188 |
"step": 750
|
189 |
},
|
190 |
{
|
191 |
"epoch": 1.93,
|
192 |
-
"learning_rate":
|
193 |
"loss": 0.0002,
|
194 |
"step": 775
|
195 |
},
|
196 |
{
|
197 |
"epoch": 1.99,
|
198 |
-
"learning_rate":
|
199 |
"loss": 0.0002,
|
200 |
"step": 800
|
201 |
},
|
202 |
{
|
203 |
"epoch": 2.05,
|
204 |
-
"learning_rate":
|
205 |
"loss": 0.0002,
|
206 |
"step": 825
|
207 |
},
|
208 |
{
|
209 |
"epoch": 2.11,
|
210 |
-
"learning_rate":
|
211 |
"loss": 0.0002,
|
212 |
"step": 850
|
213 |
},
|
214 |
{
|
215 |
"epoch": 2.18,
|
216 |
-
"learning_rate":
|
217 |
"loss": 0.0002,
|
218 |
"step": 875
|
219 |
},
|
220 |
{
|
221 |
"epoch": 2.24,
|
222 |
-
"learning_rate":
|
223 |
"loss": 0.0002,
|
224 |
"step": 900
|
225 |
},
|
226 |
{
|
227 |
"epoch": 2.3,
|
228 |
-
"learning_rate":
|
229 |
"loss": 0.0002,
|
230 |
"step": 925
|
231 |
},
|
232 |
{
|
233 |
"epoch": 2.36,
|
234 |
-
"learning_rate":
|
235 |
"loss": 0.0002,
|
236 |
"step": 950
|
237 |
},
|
238 |
{
|
239 |
"epoch": 2.43,
|
240 |
-
"learning_rate":
|
241 |
"loss": 0.0002,
|
242 |
"step": 975
|
243 |
},
|
244 |
{
|
245 |
"epoch": 2.49,
|
246 |
-
"learning_rate":
|
247 |
"loss": 0.0002,
|
248 |
"step": 1000
|
249 |
},
|
250 |
{
|
251 |
"epoch": 2.49,
|
252 |
"eval_loss": 0.50146484375,
|
253 |
-
"eval_runtime": 168.
|
254 |
-
"eval_samples_per_second": 10.
|
255 |
-
"eval_steps_per_second": 0.
|
256 |
-
"eval_wer": 16.
|
257 |
"step": 1000
|
258 |
},
|
259 |
{
|
260 |
"epoch": 2.55,
|
261 |
-
"learning_rate":
|
262 |
"loss": 0.0002,
|
263 |
"step": 1025
|
264 |
},
|
265 |
{
|
266 |
"epoch": 2.61,
|
267 |
-
"learning_rate":
|
268 |
"loss": 0.0002,
|
269 |
"step": 1050
|
270 |
},
|
271 |
{
|
272 |
"epoch": 2.67,
|
273 |
-
"learning_rate":
|
274 |
"loss": 0.0002,
|
275 |
"step": 1075
|
276 |
},
|
277 |
{
|
278 |
"epoch": 2.74,
|
279 |
-
"learning_rate":
|
280 |
"loss": 0.0002,
|
281 |
"step": 1100
|
282 |
},
|
283 |
{
|
284 |
"epoch": 2.8,
|
285 |
-
"learning_rate":
|
286 |
"loss": 0.0002,
|
287 |
"step": 1125
|
288 |
},
|
289 |
{
|
290 |
"epoch": 2.86,
|
291 |
-
"learning_rate":
|
292 |
"loss": 0.0002,
|
293 |
"step": 1150
|
294 |
},
|
295 |
{
|
296 |
"epoch": 2.92,
|
297 |
-
"learning_rate":
|
298 |
"loss": 0.0002,
|
299 |
"step": 1175
|
300 |
},
|
301 |
{
|
302 |
"epoch": 2.99,
|
303 |
-
"learning_rate":
|
304 |
"loss": 0.0002,
|
305 |
"step": 1200
|
306 |
},
|
307 |
{
|
308 |
"epoch": 3.05,
|
309 |
-
"learning_rate":
|
310 |
"loss": 0.0002,
|
311 |
"step": 1225
|
312 |
},
|
313 |
{
|
314 |
"epoch": 3.11,
|
315 |
-
"learning_rate":
|
316 |
"loss": 0.0002,
|
317 |
"step": 1250
|
318 |
},
|
319 |
{
|
320 |
"epoch": 3.17,
|
321 |
-
"learning_rate":
|
322 |
"loss": 0.0002,
|
323 |
"step": 1275
|
324 |
},
|
325 |
{
|
326 |
"epoch": 3.23,
|
327 |
-
"learning_rate":
|
328 |
"loss": 0.0002,
|
329 |
"step": 1300
|
330 |
},
|
331 |
{
|
332 |
"epoch": 3.3,
|
333 |
-
"learning_rate":
|
334 |
"loss": 0.0002,
|
335 |
"step": 1325
|
336 |
},
|
337 |
{
|
338 |
"epoch": 3.36,
|
339 |
-
"learning_rate":
|
340 |
"loss": 0.0002,
|
341 |
"step": 1350
|
342 |
},
|
343 |
{
|
344 |
"epoch": 3.42,
|
345 |
-
"learning_rate":
|
346 |
"loss": 0.0002,
|
347 |
"step": 1375
|
348 |
},
|
349 |
{
|
350 |
"epoch": 3.48,
|
351 |
-
"learning_rate":
|
352 |
"loss": 0.0002,
|
353 |
"step": 1400
|
354 |
},
|
355 |
{
|
356 |
"epoch": 3.54,
|
357 |
-
"learning_rate":
|
358 |
"loss": 0.0002,
|
359 |
"step": 1425
|
360 |
},
|
361 |
{
|
362 |
"epoch": 3.61,
|
363 |
-
"learning_rate":
|
364 |
"loss": 0.0002,
|
365 |
"step": 1450
|
366 |
},
|
367 |
{
|
368 |
"epoch": 3.67,
|
369 |
-
"learning_rate":
|
370 |
"loss": 0.0002,
|
371 |
"step": 1475
|
372 |
},
|
373 |
{
|
374 |
"epoch": 3.73,
|
375 |
-
"learning_rate":
|
376 |
"loss": 0.0002,
|
377 |
"step": 1500
|
378 |
},
|
379 |
{
|
380 |
"epoch": 3.79,
|
381 |
-
"learning_rate":
|
382 |
"loss": 0.0002,
|
383 |
"step": 1525
|
384 |
},
|
385 |
{
|
386 |
"epoch": 3.86,
|
387 |
-
"learning_rate":
|
388 |
"loss": 0.0002,
|
389 |
"step": 1550
|
390 |
},
|
391 |
{
|
392 |
"epoch": 3.92,
|
393 |
-
"learning_rate":
|
394 |
"loss": 0.0002,
|
395 |
"step": 1575
|
396 |
},
|
397 |
{
|
398 |
"epoch": 3.98,
|
399 |
-
"learning_rate":
|
400 |
"loss": 0.0002,
|
401 |
"step": 1600
|
402 |
},
|
403 |
{
|
404 |
"epoch": 4.04,
|
405 |
-
"learning_rate":
|
406 |
"loss": 0.0002,
|
407 |
"step": 1625
|
408 |
},
|
409 |
{
|
410 |
"epoch": 4.1,
|
411 |
-
"learning_rate":
|
412 |
"loss": 0.0002,
|
413 |
"step": 1650
|
414 |
},
|
415 |
{
|
416 |
"epoch": 4.17,
|
417 |
-
"learning_rate":
|
418 |
"loss": 0.0002,
|
419 |
"step": 1675
|
420 |
},
|
421 |
{
|
422 |
"epoch": 4.23,
|
423 |
-
"learning_rate":
|
424 |
"loss": 0.0002,
|
425 |
"step": 1700
|
426 |
},
|
427 |
{
|
428 |
"epoch": 4.29,
|
429 |
-
"learning_rate":
|
430 |
"loss": 0.0002,
|
431 |
"step": 1725
|
432 |
},
|
433 |
{
|
434 |
"epoch": 4.35,
|
435 |
-
"learning_rate":
|
436 |
"loss": 0.0002,
|
437 |
"step": 1750
|
438 |
},
|
439 |
{
|
440 |
"epoch": 4.42,
|
441 |
-
"learning_rate":
|
442 |
"loss": 0.0002,
|
443 |
"step": 1775
|
444 |
},
|
445 |
{
|
446 |
"epoch": 4.48,
|
447 |
-
"learning_rate":
|
448 |
"loss": 0.0002,
|
449 |
"step": 1800
|
450 |
},
|
451 |
{
|
452 |
"epoch": 4.54,
|
453 |
-
"learning_rate":
|
454 |
"loss": 0.0002,
|
455 |
"step": 1825
|
456 |
},
|
457 |
{
|
458 |
"epoch": 4.6,
|
459 |
-
"learning_rate":
|
460 |
"loss": 0.0002,
|
461 |
"step": 1850
|
462 |
},
|
463 |
{
|
464 |
"epoch": 4.66,
|
465 |
-
"learning_rate":
|
466 |
"loss": 0.0002,
|
467 |
"step": 1875
|
468 |
},
|
469 |
{
|
470 |
"epoch": 4.73,
|
471 |
-
"learning_rate":
|
472 |
"loss": 0.0002,
|
473 |
"step": 1900
|
474 |
},
|
475 |
{
|
476 |
"epoch": 4.79,
|
477 |
-
"learning_rate":
|
478 |
"loss": 0.0002,
|
479 |
"step": 1925
|
480 |
},
|
481 |
{
|
482 |
"epoch": 4.85,
|
483 |
-
"learning_rate":
|
484 |
"loss": 0.0002,
|
485 |
"step": 1950
|
486 |
},
|
487 |
{
|
488 |
"epoch": 4.91,
|
489 |
-
"learning_rate":
|
490 |
"loss": 0.0002,
|
491 |
"step": 1975
|
492 |
},
|
493 |
{
|
494 |
"epoch": 4.98,
|
495 |
-
"learning_rate":
|
496 |
"loss": 0.0002,
|
497 |
"step": 2000
|
498 |
},
|
499 |
{
|
500 |
"epoch": 4.98,
|
501 |
"eval_loss": 0.50146484375,
|
502 |
-
"eval_runtime":
|
503 |
-
"eval_samples_per_second": 10.
|
504 |
-
"eval_steps_per_second": 0.
|
505 |
-
"eval_wer": 16.
|
506 |
"step": 2000
|
507 |
},
|
508 |
{
|
509 |
"epoch": 5.04,
|
510 |
-
"learning_rate":
|
511 |
"loss": 0.0002,
|
512 |
"step": 2025
|
513 |
},
|
514 |
{
|
515 |
"epoch": 5.1,
|
516 |
-
"learning_rate":
|
517 |
"loss": 0.0002,
|
518 |
"step": 2050
|
519 |
},
|
520 |
{
|
521 |
"epoch": 5.16,
|
522 |
-
"learning_rate":
|
523 |
"loss": 0.0002,
|
524 |
"step": 2075
|
525 |
},
|
526 |
{
|
527 |
"epoch": 5.22,
|
528 |
-
"learning_rate":
|
529 |
"loss": 0.0002,
|
530 |
"step": 2100
|
531 |
},
|
532 |
{
|
533 |
"epoch": 5.29,
|
534 |
-
"learning_rate":
|
535 |
"loss": 0.0002,
|
536 |
"step": 2125
|
537 |
},
|
538 |
{
|
539 |
"epoch": 5.35,
|
540 |
-
"learning_rate":
|
541 |
"loss": 0.0002,
|
542 |
"step": 2150
|
543 |
},
|
544 |
{
|
545 |
"epoch": 5.41,
|
546 |
-
"learning_rate":
|
547 |
"loss": 0.0002,
|
548 |
"step": 2175
|
549 |
},
|
550 |
{
|
551 |
"epoch": 5.47,
|
552 |
-
"learning_rate":
|
553 |
"loss": 0.0002,
|
554 |
"step": 2200
|
555 |
},
|
556 |
{
|
557 |
"epoch": 5.53,
|
558 |
-
"learning_rate":
|
559 |
"loss": 0.0002,
|
560 |
"step": 2225
|
561 |
},
|
562 |
{
|
563 |
"epoch": 5.6,
|
564 |
-
"learning_rate":
|
565 |
"loss": 0.0002,
|
566 |
"step": 2250
|
567 |
},
|
568 |
{
|
569 |
"epoch": 5.66,
|
570 |
-
"learning_rate":
|
571 |
"loss": 0.0002,
|
572 |
"step": 2275
|
573 |
},
|
574 |
{
|
575 |
"epoch": 5.72,
|
576 |
-
"learning_rate":
|
577 |
"loss": 0.0002,
|
578 |
"step": 2300
|
579 |
},
|
580 |
{
|
581 |
"epoch": 5.78,
|
582 |
-
"learning_rate":
|
583 |
"loss": 0.0002,
|
584 |
"step": 2325
|
585 |
},
|
586 |
{
|
587 |
"epoch": 5.85,
|
588 |
-
"learning_rate":
|
589 |
"loss": 0.0002,
|
590 |
"step": 2350
|
591 |
},
|
592 |
{
|
593 |
"epoch": 5.91,
|
594 |
-
"learning_rate":
|
595 |
"loss": 0.0002,
|
596 |
"step": 2375
|
597 |
},
|
598 |
{
|
599 |
"epoch": 5.97,
|
600 |
-
"learning_rate":
|
601 |
"loss": 0.0002,
|
602 |
"step": 2400
|
603 |
},
|
604 |
{
|
605 |
"epoch": 6.03,
|
606 |
-
"learning_rate":
|
607 |
"loss": 0.0002,
|
608 |
"step": 2425
|
609 |
},
|
610 |
{
|
611 |
"epoch": 6.09,
|
612 |
-
"learning_rate":
|
613 |
"loss": 0.0002,
|
614 |
"step": 2450
|
615 |
},
|
616 |
{
|
617 |
"epoch": 6.16,
|
618 |
-
"learning_rate":
|
619 |
"loss": 0.0002,
|
620 |
"step": 2475
|
621 |
},
|
622 |
{
|
623 |
"epoch": 6.22,
|
624 |
-
"learning_rate":
|
625 |
"loss": 0.0002,
|
626 |
"step": 2500
|
627 |
},
|
628 |
{
|
629 |
"epoch": 6.28,
|
630 |
-
"learning_rate":
|
631 |
"loss": 0.0002,
|
632 |
"step": 2525
|
633 |
},
|
634 |
{
|
635 |
"epoch": 6.34,
|
636 |
-
"learning_rate":
|
637 |
"loss": 0.0002,
|
638 |
"step": 2550
|
639 |
},
|
640 |
{
|
641 |
"epoch": 6.41,
|
642 |
-
"learning_rate":
|
643 |
"loss": 0.0002,
|
644 |
"step": 2575
|
645 |
},
|
646 |
{
|
647 |
"epoch": 6.47,
|
648 |
-
"learning_rate":
|
649 |
"loss": 0.0002,
|
650 |
"step": 2600
|
651 |
},
|
652 |
{
|
653 |
"epoch": 6.53,
|
654 |
-
"learning_rate":
|
655 |
"loss": 0.0002,
|
656 |
"step": 2625
|
657 |
},
|
658 |
{
|
659 |
"epoch": 6.59,
|
660 |
-
"learning_rate":
|
661 |
"loss": 0.0002,
|
662 |
"step": 2650
|
663 |
},
|
664 |
{
|
665 |
"epoch": 6.65,
|
666 |
-
"learning_rate":
|
667 |
"loss": 0.0002,
|
668 |
"step": 2675
|
669 |
},
|
670 |
{
|
671 |
"epoch": 6.72,
|
672 |
-
"learning_rate":
|
673 |
"loss": 0.0002,
|
674 |
"step": 2700
|
675 |
},
|
676 |
{
|
677 |
"epoch": 6.78,
|
678 |
-
"learning_rate":
|
679 |
"loss": 0.0002,
|
680 |
"step": 2725
|
681 |
},
|
682 |
{
|
683 |
"epoch": 6.84,
|
684 |
-
"learning_rate":
|
685 |
"loss": 0.0002,
|
686 |
"step": 2750
|
687 |
},
|
688 |
{
|
689 |
"epoch": 6.9,
|
690 |
-
"learning_rate":
|
691 |
"loss": 0.0002,
|
692 |
"step": 2775
|
693 |
},
|
694 |
{
|
695 |
"epoch": 6.97,
|
696 |
-
"learning_rate":
|
697 |
"loss": 0.0002,
|
698 |
"step": 2800
|
699 |
},
|
700 |
{
|
701 |
"epoch": 7.03,
|
702 |
-
"learning_rate":
|
703 |
"loss": 0.0002,
|
704 |
"step": 2825
|
705 |
},
|
706 |
{
|
707 |
"epoch": 7.09,
|
708 |
-
"learning_rate":
|
709 |
"loss": 0.0002,
|
710 |
"step": 2850
|
711 |
},
|
712 |
{
|
713 |
"epoch": 7.15,
|
714 |
-
"learning_rate":
|
715 |
"loss": 0.0002,
|
716 |
"step": 2875
|
717 |
},
|
718 |
{
|
719 |
"epoch": 7.21,
|
720 |
-
"learning_rate":
|
721 |
"loss": 0.0002,
|
722 |
"step": 2900
|
723 |
},
|
724 |
{
|
725 |
"epoch": 7.28,
|
726 |
-
"learning_rate":
|
727 |
"loss": 0.0002,
|
728 |
"step": 2925
|
729 |
},
|
730 |
{
|
731 |
"epoch": 7.34,
|
732 |
-
"learning_rate":
|
733 |
"loss": 0.0002,
|
734 |
"step": 2950
|
735 |
},
|
736 |
{
|
737 |
"epoch": 7.4,
|
738 |
-
"learning_rate":
|
739 |
"loss": 0.0002,
|
740 |
"step": 2975
|
741 |
},
|
742 |
{
|
743 |
"epoch": 7.46,
|
744 |
-
"learning_rate":
|
745 |
"loss": 0.0002,
|
746 |
"step": 3000
|
747 |
},
|
748 |
{
|
749 |
"epoch": 7.46,
|
750 |
"eval_loss": 0.50146484375,
|
751 |
-
"eval_runtime": 167.
|
752 |
-
"eval_samples_per_second": 10.
|
753 |
"eval_steps_per_second": 0.633,
|
754 |
-
"eval_wer": 16.
|
755 |
"step": 3000
|
756 |
},
|
757 |
{
|
758 |
"epoch": 7.52,
|
759 |
-
"learning_rate":
|
760 |
"loss": 0.0002,
|
761 |
"step": 3025
|
762 |
},
|
763 |
{
|
764 |
"epoch": 7.59,
|
765 |
-
"learning_rate":
|
766 |
"loss": 0.0001,
|
767 |
"step": 3050
|
768 |
},
|
769 |
{
|
770 |
"epoch": 7.65,
|
771 |
-
"learning_rate":
|
772 |
"loss": 0.0002,
|
773 |
"step": 3075
|
774 |
},
|
775 |
{
|
776 |
"epoch": 7.71,
|
777 |
-
"learning_rate":
|
778 |
"loss": 0.0002,
|
779 |
"step": 3100
|
780 |
},
|
781 |
{
|
782 |
"epoch": 7.77,
|
783 |
-
"learning_rate":
|
784 |
"loss": 0.0002,
|
785 |
"step": 3125
|
786 |
},
|
787 |
{
|
788 |
"epoch": 7.84,
|
789 |
-
"learning_rate":
|
790 |
"loss": 0.0002,
|
791 |
"step": 3150
|
792 |
},
|
793 |
{
|
794 |
"epoch": 7.9,
|
795 |
-
"learning_rate":
|
796 |
"loss": 0.0002,
|
797 |
"step": 3175
|
798 |
},
|
799 |
{
|
800 |
"epoch": 7.96,
|
801 |
-
"learning_rate":
|
802 |
"loss": 0.0002,
|
803 |
"step": 3200
|
804 |
},
|
805 |
{
|
806 |
"epoch": 8.02,
|
807 |
-
"learning_rate":
|
808 |
"loss": 0.0002,
|
809 |
"step": 3225
|
810 |
},
|
811 |
{
|
812 |
"epoch": 8.08,
|
813 |
-
"learning_rate":
|
814 |
"loss": 0.0002,
|
815 |
"step": 3250
|
816 |
},
|
817 |
{
|
818 |
"epoch": 8.15,
|
819 |
-
"learning_rate":
|
820 |
"loss": 0.0002,
|
821 |
"step": 3275
|
822 |
},
|
823 |
{
|
824 |
"epoch": 8.21,
|
825 |
-
"learning_rate":
|
826 |
"loss": 0.0002,
|
827 |
"step": 3300
|
828 |
},
|
829 |
{
|
830 |
"epoch": 8.27,
|
831 |
-
"learning_rate":
|
832 |
"loss": 0.0002,
|
833 |
"step": 3325
|
834 |
},
|
835 |
{
|
836 |
"epoch": 8.33,
|
837 |
-
"learning_rate":
|
838 |
"loss": 0.0002,
|
839 |
"step": 3350
|
840 |
},
|
841 |
{
|
842 |
"epoch": 8.4,
|
843 |
-
"learning_rate":
|
844 |
"loss": 0.0002,
|
845 |
"step": 3375
|
846 |
},
|
847 |
{
|
848 |
"epoch": 8.46,
|
849 |
-
"learning_rate":
|
850 |
"loss": 0.0002,
|
851 |
"step": 3400
|
852 |
},
|
853 |
{
|
854 |
"epoch": 8.52,
|
855 |
-
"learning_rate":
|
856 |
"loss": 0.0002,
|
857 |
"step": 3425
|
858 |
},
|
859 |
{
|
860 |
"epoch": 8.58,
|
861 |
-
"learning_rate":
|
862 |
"loss": 0.0002,
|
863 |
"step": 3450
|
864 |
},
|
865 |
{
|
866 |
"epoch": 8.64,
|
867 |
-
"learning_rate":
|
868 |
"loss": 0.0002,
|
869 |
"step": 3475
|
870 |
},
|
871 |
{
|
872 |
"epoch": 8.71,
|
873 |
-
"learning_rate":
|
874 |
"loss": 0.0002,
|
875 |
"step": 3500
|
876 |
},
|
877 |
{
|
878 |
"epoch": 8.77,
|
879 |
-
"learning_rate":
|
880 |
-
"loss": 0.
|
881 |
"step": 3525
|
882 |
},
|
883 |
{
|
884 |
"epoch": 8.83,
|
885 |
-
"learning_rate":
|
886 |
"loss": 0.0002,
|
887 |
"step": 3550
|
888 |
},
|
889 |
{
|
890 |
"epoch": 8.89,
|
891 |
-
"learning_rate":
|
892 |
"loss": 0.0002,
|
893 |
"step": 3575
|
894 |
},
|
895 |
{
|
896 |
"epoch": 8.96,
|
897 |
-
"learning_rate":
|
898 |
"loss": 0.0002,
|
899 |
"step": 3600
|
900 |
},
|
901 |
{
|
902 |
"epoch": 9.02,
|
903 |
-
"learning_rate":
|
904 |
"loss": 0.0002,
|
905 |
"step": 3625
|
906 |
},
|
907 |
{
|
908 |
"epoch": 9.08,
|
909 |
-
"learning_rate":
|
910 |
"loss": 0.0002,
|
911 |
"step": 3650
|
912 |
},
|
913 |
{
|
914 |
"epoch": 9.14,
|
915 |
-
"learning_rate": 2.
|
916 |
"loss": 0.0002,
|
917 |
"step": 3675
|
918 |
},
|
919 |
{
|
920 |
"epoch": 9.2,
|
921 |
-
"learning_rate":
|
922 |
"loss": 0.0002,
|
923 |
"step": 3700
|
924 |
},
|
925 |
{
|
926 |
"epoch": 9.27,
|
927 |
-
"learning_rate":
|
928 |
"loss": 0.0002,
|
929 |
"step": 3725
|
930 |
},
|
931 |
{
|
932 |
"epoch": 9.33,
|
933 |
-
"learning_rate":
|
934 |
"loss": 0.0002,
|
935 |
"step": 3750
|
936 |
},
|
937 |
{
|
938 |
"epoch": 9.39,
|
939 |
-
"learning_rate":
|
940 |
"loss": 0.0002,
|
941 |
"step": 3775
|
942 |
},
|
943 |
{
|
944 |
"epoch": 9.45,
|
945 |
-
"learning_rate":
|
946 |
"loss": 0.0002,
|
947 |
"step": 3800
|
948 |
},
|
949 |
{
|
950 |
"epoch": 9.51,
|
951 |
-
"learning_rate":
|
952 |
"loss": 0.0002,
|
953 |
"step": 3825
|
954 |
},
|
955 |
{
|
956 |
"epoch": 9.58,
|
957 |
-
"learning_rate":
|
958 |
"loss": 0.0002,
|
959 |
"step": 3850
|
960 |
},
|
961 |
{
|
962 |
"epoch": 9.64,
|
963 |
-
"learning_rate":
|
964 |
"loss": 0.0002,
|
965 |
"step": 3875
|
966 |
},
|
967 |
{
|
968 |
"epoch": 9.7,
|
969 |
-
"learning_rate":
|
970 |
"loss": 0.0002,
|
971 |
"step": 3900
|
972 |
},
|
973 |
{
|
974 |
"epoch": 9.76,
|
975 |
-
"learning_rate":
|
976 |
"loss": 0.0002,
|
977 |
"step": 3925
|
978 |
},
|
979 |
{
|
980 |
"epoch": 9.83,
|
981 |
-
"learning_rate":
|
982 |
"loss": 0.0002,
|
983 |
"step": 3950
|
984 |
},
|
985 |
{
|
986 |
"epoch": 9.89,
|
987 |
-
"learning_rate":
|
988 |
"loss": 0.0002,
|
989 |
"step": 3975
|
990 |
},
|
991 |
{
|
992 |
"epoch": 9.95,
|
993 |
-
"learning_rate":
|
994 |
"loss": 0.0002,
|
995 |
"step": 4000
|
996 |
},
|
997 |
{
|
998 |
"epoch": 9.95,
|
999 |
"eval_loss": 0.50146484375,
|
1000 |
-
"eval_runtime": 167.
|
1001 |
-
"eval_samples_per_second": 10.
|
1002 |
-
"eval_steps_per_second": 0.
|
1003 |
-
"eval_wer": 16.
|
1004 |
"step": 4000
|
1005 |
},
|
1006 |
{
|
1007 |
"epoch": 10.01,
|
1008 |
-
"learning_rate":
|
1009 |
"loss": 0.0002,
|
1010 |
"step": 4025
|
1011 |
},
|
1012 |
{
|
1013 |
"epoch": 10.07,
|
1014 |
-
"learning_rate":
|
1015 |
"loss": 0.0002,
|
1016 |
"step": 4050
|
1017 |
},
|
1018 |
{
|
1019 |
"epoch": 10.14,
|
1020 |
-
"learning_rate":
|
1021 |
"loss": 0.0002,
|
1022 |
"step": 4075
|
1023 |
},
|
1024 |
{
|
1025 |
"epoch": 10.2,
|
1026 |
-
"learning_rate":
|
1027 |
"loss": 0.0002,
|
1028 |
"step": 4100
|
1029 |
},
|
1030 |
{
|
1031 |
"epoch": 10.26,
|
1032 |
-
"learning_rate": 1.
|
1033 |
"loss": 0.0002,
|
1034 |
"step": 4125
|
1035 |
},
|
1036 |
{
|
1037 |
"epoch": 10.32,
|
1038 |
-
"learning_rate": 1.
|
1039 |
"loss": 0.0002,
|
1040 |
"step": 4150
|
1041 |
},
|
1042 |
{
|
1043 |
"epoch": 10.39,
|
1044 |
-
"learning_rate": 1.
|
1045 |
"loss": 0.0002,
|
1046 |
"step": 4175
|
1047 |
},
|
1048 |
{
|
1049 |
"epoch": 10.45,
|
1050 |
-
"learning_rate": 1.
|
1051 |
"loss": 0.0002,
|
1052 |
"step": 4200
|
1053 |
},
|
1054 |
{
|
1055 |
"epoch": 10.51,
|
1056 |
-
"learning_rate": 1.
|
1057 |
"loss": 0.0002,
|
1058 |
"step": 4225
|
1059 |
},
|
1060 |
{
|
1061 |
"epoch": 10.57,
|
1062 |
-
"learning_rate": 1.
|
1063 |
"loss": 0.0002,
|
1064 |
"step": 4250
|
1065 |
},
|
1066 |
{
|
1067 |
"epoch": 10.63,
|
1068 |
-
"learning_rate": 1.
|
1069 |
"loss": 0.0002,
|
1070 |
"step": 4275
|
1071 |
},
|
1072 |
{
|
1073 |
"epoch": 10.7,
|
1074 |
-
"learning_rate": 1.
|
1075 |
"loss": 0.0002,
|
1076 |
"step": 4300
|
1077 |
},
|
1078 |
{
|
1079 |
"epoch": 10.76,
|
1080 |
-
"learning_rate": 1.
|
1081 |
"loss": 0.0002,
|
1082 |
"step": 4325
|
1083 |
},
|
1084 |
{
|
1085 |
"epoch": 10.82,
|
1086 |
-
"learning_rate": 1.
|
1087 |
"loss": 0.0002,
|
1088 |
"step": 4350
|
1089 |
},
|
1090 |
{
|
1091 |
"epoch": 10.88,
|
1092 |
-
"learning_rate": 1.
|
1093 |
"loss": 0.0002,
|
1094 |
"step": 4375
|
1095 |
},
|
1096 |
{
|
1097 |
"epoch": 10.95,
|
1098 |
-
"learning_rate": 1.
|
1099 |
"loss": 0.0002,
|
1100 |
"step": 4400
|
1101 |
},
|
1102 |
{
|
1103 |
"epoch": 11.01,
|
1104 |
-
"learning_rate": 1.
|
1105 |
"loss": 0.0002,
|
1106 |
"step": 4425
|
1107 |
},
|
1108 |
{
|
1109 |
"epoch": 11.07,
|
1110 |
-
"learning_rate": 1.
|
1111 |
"loss": 0.0002,
|
1112 |
"step": 4450
|
1113 |
},
|
1114 |
{
|
1115 |
"epoch": 11.13,
|
1116 |
-
"learning_rate": 1.
|
1117 |
"loss": 0.0002,
|
1118 |
"step": 4475
|
1119 |
},
|
1120 |
{
|
1121 |
"epoch": 11.19,
|
1122 |
-
"learning_rate": 1.
|
1123 |
"loss": 0.0002,
|
1124 |
"step": 4500
|
1125 |
},
|
1126 |
{
|
1127 |
"epoch": 11.26,
|
1128 |
-
"learning_rate": 1.
|
1129 |
"loss": 0.0002,
|
1130 |
"step": 4525
|
1131 |
},
|
1132 |
{
|
1133 |
"epoch": 11.32,
|
1134 |
-
"learning_rate": 1.
|
1135 |
"loss": 0.0002,
|
1136 |
"step": 4550
|
1137 |
},
|
1138 |
{
|
1139 |
"epoch": 11.38,
|
1140 |
-
"learning_rate":
|
1141 |
"loss": 0.0002,
|
1142 |
"step": 4575
|
1143 |
},
|
1144 |
{
|
1145 |
"epoch": 11.44,
|
1146 |
-
"learning_rate":
|
1147 |
"loss": 0.0002,
|
1148 |
"step": 4600
|
1149 |
},
|
1150 |
{
|
1151 |
"epoch": 11.5,
|
1152 |
-
"learning_rate":
|
1153 |
"loss": 0.0002,
|
1154 |
"step": 4625
|
1155 |
},
|
1156 |
{
|
1157 |
"epoch": 11.57,
|
1158 |
-
"learning_rate":
|
1159 |
"loss": 0.0002,
|
1160 |
"step": 4650
|
1161 |
},
|
1162 |
{
|
1163 |
"epoch": 11.63,
|
1164 |
-
"learning_rate":
|
1165 |
"loss": 0.0002,
|
1166 |
"step": 4675
|
1167 |
},
|
1168 |
{
|
1169 |
"epoch": 11.69,
|
1170 |
-
"learning_rate":
|
1171 |
"loss": 0.0002,
|
1172 |
"step": 4700
|
1173 |
},
|
1174 |
{
|
1175 |
"epoch": 11.75,
|
1176 |
-
"learning_rate":
|
1177 |
"loss": 0.0002,
|
1178 |
"step": 4725
|
1179 |
},
|
1180 |
{
|
1181 |
"epoch": 11.82,
|
1182 |
-
"learning_rate":
|
1183 |
"loss": 0.0002,
|
1184 |
"step": 4750
|
1185 |
},
|
1186 |
{
|
1187 |
"epoch": 11.88,
|
1188 |
-
"learning_rate":
|
1189 |
"loss": 0.0002,
|
1190 |
"step": 4775
|
1191 |
},
|
1192 |
{
|
1193 |
"epoch": 11.94,
|
1194 |
-
"learning_rate":
|
1195 |
"loss": 0.0002,
|
1196 |
"step": 4800
|
1197 |
},
|
1198 |
{
|
1199 |
"epoch": 12.0,
|
1200 |
-
"learning_rate":
|
1201 |
"loss": 0.0002,
|
1202 |
"step": 4825
|
1203 |
},
|
1204 |
{
|
1205 |
"epoch": 12.06,
|
1206 |
-
"learning_rate":
|
1207 |
"loss": 0.0002,
|
1208 |
"step": 4850
|
1209 |
},
|
1210 |
{
|
1211 |
"epoch": 12.13,
|
1212 |
-
"learning_rate":
|
1213 |
"loss": 0.0002,
|
1214 |
"step": 4875
|
1215 |
},
|
1216 |
{
|
1217 |
"epoch": 12.19,
|
1218 |
-
"learning_rate":
|
1219 |
"loss": 0.0002,
|
1220 |
"step": 4900
|
1221 |
},
|
1222 |
{
|
1223 |
"epoch": 12.25,
|
1224 |
-
"learning_rate": 1.
|
1225 |
"loss": 0.0002,
|
1226 |
"step": 4925
|
1227 |
},
|
1228 |
{
|
1229 |
"epoch": 12.31,
|
1230 |
-
"learning_rate": 1.
|
1231 |
"loss": 0.0002,
|
1232 |
"step": 4950
|
1233 |
},
|
1234 |
{
|
1235 |
"epoch": 12.38,
|
1236 |
-
"learning_rate":
|
1237 |
"loss": 0.0002,
|
1238 |
"step": 4975
|
1239 |
},
|
1240 |
{
|
1241 |
"epoch": 12.44,
|
1242 |
-
"learning_rate":
|
1243 |
"loss": 0.0002,
|
1244 |
"step": 5000
|
1245 |
},
|
1246 |
{
|
1247 |
"epoch": 12.44,
|
1248 |
"eval_loss": 0.50146484375,
|
1249 |
-
"eval_runtime": 168.
|
1250 |
-
"eval_samples_per_second": 10.
|
1251 |
-
"eval_steps_per_second": 0.
|
1252 |
"eval_wer": 16.558320950965825,
|
1253 |
"step": 5000
|
1254 |
},
|
1255 |
{
|
1256 |
-
"epoch": 12.
|
1257 |
-
"
|
1258 |
-
"
|
1259 |
-
"
|
1260 |
-
|
1261 |
-
|
1262 |
-
"
|
1263 |
-
"learning_rate": 1.5672631578947367e-09,
|
1264 |
-
"loss": 0.0002,
|
1265 |
-
"step": 5050
|
1266 |
-
},
|
1267 |
-
{
|
1268 |
-
"epoch": 12.62,
|
1269 |
-
"learning_rate": 1.5593684210526317e-09,
|
1270 |
-
"loss": 0.0002,
|
1271 |
-
"step": 5075
|
1272 |
-
},
|
1273 |
-
{
|
1274 |
-
"epoch": 12.69,
|
1275 |
-
"learning_rate": 1.5514736842105264e-09,
|
1276 |
-
"loss": 0.0002,
|
1277 |
-
"step": 5100
|
1278 |
-
},
|
1279 |
-
{
|
1280 |
-
"epoch": 12.75,
|
1281 |
-
"learning_rate": 1.543578947368421e-09,
|
1282 |
-
"loss": 0.0002,
|
1283 |
-
"step": 5125
|
1284 |
-
},
|
1285 |
-
{
|
1286 |
-
"epoch": 12.81,
|
1287 |
-
"learning_rate": 1.535684210526316e-09,
|
1288 |
-
"loss": 0.0002,
|
1289 |
-
"step": 5150
|
1290 |
-
},
|
1291 |
-
{
|
1292 |
-
"epoch": 12.87,
|
1293 |
-
"learning_rate": 1.5277894736842106e-09,
|
1294 |
-
"loss": 0.0002,
|
1295 |
-
"step": 5175
|
1296 |
-
},
|
1297 |
-
{
|
1298 |
-
"epoch": 12.94,
|
1299 |
-
"learning_rate": 1.5198947368421053e-09,
|
1300 |
-
"loss": 0.0002,
|
1301 |
-
"step": 5200
|
1302 |
-
},
|
1303 |
-
{
|
1304 |
-
"epoch": 13.0,
|
1305 |
-
"learning_rate": 1.512e-09,
|
1306 |
-
"loss": 0.0002,
|
1307 |
-
"step": 5225
|
1308 |
-
},
|
1309 |
-
{
|
1310 |
-
"epoch": 13.06,
|
1311 |
-
"learning_rate": 1.5041052631578946e-09,
|
1312 |
-
"loss": 0.0002,
|
1313 |
-
"step": 5250
|
1314 |
-
},
|
1315 |
-
{
|
1316 |
-
"epoch": 13.12,
|
1317 |
-
"learning_rate": 1.4962105263157895e-09,
|
1318 |
-
"loss": 0.0002,
|
1319 |
-
"step": 5275
|
1320 |
-
},
|
1321 |
-
{
|
1322 |
-
"epoch": 13.18,
|
1323 |
-
"learning_rate": 1.4883157894736842e-09,
|
1324 |
-
"loss": 0.0002,
|
1325 |
-
"step": 5300
|
1326 |
-
},
|
1327 |
-
{
|
1328 |
-
"epoch": 13.25,
|
1329 |
-
"learning_rate": 1.480421052631579e-09,
|
1330 |
-
"loss": 0.0002,
|
1331 |
-
"step": 5325
|
1332 |
-
},
|
1333 |
-
{
|
1334 |
-
"epoch": 13.31,
|
1335 |
-
"learning_rate": 1.4725263157894737e-09,
|
1336 |
-
"loss": 0.0002,
|
1337 |
-
"step": 5350
|
1338 |
-
},
|
1339 |
-
{
|
1340 |
-
"epoch": 13.37,
|
1341 |
-
"learning_rate": 1.4646315789473684e-09,
|
1342 |
-
"loss": 0.0002,
|
1343 |
-
"step": 5375
|
1344 |
-
},
|
1345 |
-
{
|
1346 |
-
"epoch": 13.43,
|
1347 |
-
"learning_rate": 1.4567368421052632e-09,
|
1348 |
-
"loss": 0.0002,
|
1349 |
-
"step": 5400
|
1350 |
-
},
|
1351 |
-
{
|
1352 |
-
"epoch": 13.5,
|
1353 |
-
"learning_rate": 1.448842105263158e-09,
|
1354 |
-
"loss": 0.0001,
|
1355 |
-
"step": 5425
|
1356 |
-
},
|
1357 |
-
{
|
1358 |
-
"epoch": 13.56,
|
1359 |
-
"learning_rate": 1.4409473684210526e-09,
|
1360 |
-
"loss": 0.0002,
|
1361 |
-
"step": 5450
|
1362 |
-
},
|
1363 |
-
{
|
1364 |
-
"epoch": 13.62,
|
1365 |
-
"learning_rate": 1.4330526315789472e-09,
|
1366 |
-
"loss": 0.0002,
|
1367 |
-
"step": 5475
|
1368 |
-
},
|
1369 |
-
{
|
1370 |
-
"epoch": 13.68,
|
1371 |
-
"learning_rate": 1.4251578947368421e-09,
|
1372 |
-
"loss": 0.0002,
|
1373 |
-
"step": 5500
|
1374 |
-
},
|
1375 |
-
{
|
1376 |
-
"epoch": 13.74,
|
1377 |
-
"learning_rate": 1.417263157894737e-09,
|
1378 |
-
"loss": 0.0002,
|
1379 |
-
"step": 5525
|
1380 |
-
},
|
1381 |
-
{
|
1382 |
-
"epoch": 13.81,
|
1383 |
-
"learning_rate": 1.4093684210526316e-09,
|
1384 |
-
"loss": 0.0002,
|
1385 |
-
"step": 5550
|
1386 |
-
},
|
1387 |
-
{
|
1388 |
-
"epoch": 13.87,
|
1389 |
-
"learning_rate": 1.4014736842105263e-09,
|
1390 |
-
"loss": 0.0002,
|
1391 |
-
"step": 5575
|
1392 |
-
},
|
1393 |
-
{
|
1394 |
-
"epoch": 13.93,
|
1395 |
-
"learning_rate": 1.393578947368421e-09,
|
1396 |
-
"loss": 0.0002,
|
1397 |
-
"step": 5600
|
1398 |
-
},
|
1399 |
-
{
|
1400 |
-
"epoch": 13.99,
|
1401 |
-
"learning_rate": 1.3856842105263156e-09,
|
1402 |
-
"loss": 0.0002,
|
1403 |
-
"step": 5625
|
1404 |
-
},
|
1405 |
-
{
|
1406 |
-
"epoch": 14.05,
|
1407 |
-
"learning_rate": 1.3777894736842105e-09,
|
1408 |
-
"loss": 0.0002,
|
1409 |
-
"step": 5650
|
1410 |
-
},
|
1411 |
-
{
|
1412 |
-
"epoch": 14.12,
|
1413 |
-
"learning_rate": 1.3698947368421054e-09,
|
1414 |
-
"loss": 0.0002,
|
1415 |
-
"step": 5675
|
1416 |
-
},
|
1417 |
-
{
|
1418 |
-
"epoch": 14.18,
|
1419 |
-
"learning_rate": 1.362e-09,
|
1420 |
-
"loss": 0.0002,
|
1421 |
-
"step": 5700
|
1422 |
-
},
|
1423 |
-
{
|
1424 |
-
"epoch": 14.24,
|
1425 |
-
"learning_rate": 1.3541052631578947e-09,
|
1426 |
-
"loss": 0.0002,
|
1427 |
-
"step": 5725
|
1428 |
-
},
|
1429 |
-
{
|
1430 |
-
"epoch": 14.3,
|
1431 |
-
"learning_rate": 1.3462105263157894e-09,
|
1432 |
-
"loss": 0.0002,
|
1433 |
-
"step": 5750
|
1434 |
-
},
|
1435 |
-
{
|
1436 |
-
"epoch": 14.37,
|
1437 |
-
"learning_rate": 1.3383157894736843e-09,
|
1438 |
-
"loss": 0.0002,
|
1439 |
-
"step": 5775
|
1440 |
-
},
|
1441 |
-
{
|
1442 |
-
"epoch": 14.43,
|
1443 |
-
"learning_rate": 1.330421052631579e-09,
|
1444 |
-
"loss": 0.0002,
|
1445 |
-
"step": 5800
|
1446 |
-
},
|
1447 |
-
{
|
1448 |
-
"epoch": 14.49,
|
1449 |
-
"learning_rate": 1.3225263157894736e-09,
|
1450 |
-
"loss": 0.0002,
|
1451 |
-
"step": 5825
|
1452 |
-
},
|
1453 |
-
{
|
1454 |
-
"epoch": 14.55,
|
1455 |
-
"learning_rate": 1.3146315789473685e-09,
|
1456 |
-
"loss": 0.0002,
|
1457 |
-
"step": 5850
|
1458 |
-
},
|
1459 |
-
{
|
1460 |
-
"epoch": 14.61,
|
1461 |
-
"learning_rate": 1.3067368421052631e-09,
|
1462 |
-
"loss": 0.0002,
|
1463 |
-
"step": 5875
|
1464 |
-
},
|
1465 |
-
{
|
1466 |
-
"epoch": 14.68,
|
1467 |
-
"learning_rate": 1.298842105263158e-09,
|
1468 |
-
"loss": 0.0002,
|
1469 |
-
"step": 5900
|
1470 |
-
},
|
1471 |
-
{
|
1472 |
-
"epoch": 14.74,
|
1473 |
-
"learning_rate": 1.2909473684210527e-09,
|
1474 |
-
"loss": 0.0002,
|
1475 |
-
"step": 5925
|
1476 |
-
},
|
1477 |
-
{
|
1478 |
-
"epoch": 14.8,
|
1479 |
-
"learning_rate": 1.2830526315789473e-09,
|
1480 |
-
"loss": 0.0002,
|
1481 |
-
"step": 5950
|
1482 |
-
},
|
1483 |
-
{
|
1484 |
-
"epoch": 14.86,
|
1485 |
-
"learning_rate": 1.275157894736842e-09,
|
1486 |
-
"loss": 0.0002,
|
1487 |
-
"step": 5975
|
1488 |
-
},
|
1489 |
-
{
|
1490 |
-
"epoch": 14.93,
|
1491 |
-
"learning_rate": 1.2672631578947369e-09,
|
1492 |
-
"loss": 0.0002,
|
1493 |
-
"step": 6000
|
1494 |
-
},
|
1495 |
-
{
|
1496 |
-
"epoch": 14.93,
|
1497 |
-
"eval_loss": 0.50146484375,
|
1498 |
-
"eval_runtime": 167.7376,
|
1499 |
-
"eval_samples_per_second": 10.111,
|
1500 |
-
"eval_steps_per_second": 0.632,
|
1501 |
-
"eval_wer": 16.576894502228825,
|
1502 |
-
"step": 6000
|
1503 |
-
},
|
1504 |
-
{
|
1505 |
-
"epoch": 14.99,
|
1506 |
-
"learning_rate": 1.26e-09,
|
1507 |
-
"loss": 0.0002,
|
1508 |
-
"step": 6025
|
1509 |
-
},
|
1510 |
-
{
|
1511 |
-
"epoch": 15.05,
|
1512 |
-
"learning_rate": 1.2521052631578947e-09,
|
1513 |
-
"loss": 0.0002,
|
1514 |
-
"step": 6050
|
1515 |
-
},
|
1516 |
-
{
|
1517 |
-
"epoch": 15.11,
|
1518 |
-
"learning_rate": 1.2442105263157895e-09,
|
1519 |
-
"loss": 0.0002,
|
1520 |
-
"step": 6075
|
1521 |
-
},
|
1522 |
-
{
|
1523 |
-
"epoch": 15.17,
|
1524 |
-
"learning_rate": 1.2363157894736842e-09,
|
1525 |
-
"loss": 0.0002,
|
1526 |
-
"step": 6100
|
1527 |
-
},
|
1528 |
-
{
|
1529 |
-
"epoch": 15.24,
|
1530 |
-
"learning_rate": 1.2284210526315789e-09,
|
1531 |
-
"loss": 0.0002,
|
1532 |
-
"step": 6125
|
1533 |
-
},
|
1534 |
-
{
|
1535 |
-
"epoch": 15.3,
|
1536 |
-
"learning_rate": 1.2205263157894737e-09,
|
1537 |
-
"loss": 0.0002,
|
1538 |
-
"step": 6150
|
1539 |
-
},
|
1540 |
-
{
|
1541 |
-
"epoch": 15.36,
|
1542 |
-
"learning_rate": 1.2126315789473684e-09,
|
1543 |
-
"loss": 0.0002,
|
1544 |
-
"step": 6175
|
1545 |
-
},
|
1546 |
-
{
|
1547 |
-
"epoch": 15.42,
|
1548 |
-
"learning_rate": 1.204736842105263e-09,
|
1549 |
-
"loss": 0.0002,
|
1550 |
-
"step": 6200
|
1551 |
-
},
|
1552 |
-
{
|
1553 |
-
"epoch": 15.49,
|
1554 |
-
"learning_rate": 1.1968421052631577e-09,
|
1555 |
-
"loss": 0.0002,
|
1556 |
-
"step": 6225
|
1557 |
-
},
|
1558 |
-
{
|
1559 |
-
"epoch": 15.55,
|
1560 |
-
"learning_rate": 1.1889473684210526e-09,
|
1561 |
-
"loss": 0.0002,
|
1562 |
-
"step": 6250
|
1563 |
-
},
|
1564 |
-
{
|
1565 |
-
"epoch": 15.61,
|
1566 |
-
"learning_rate": 1.1810526315789475e-09,
|
1567 |
-
"loss": 0.0001,
|
1568 |
-
"step": 6275
|
1569 |
-
},
|
1570 |
-
{
|
1571 |
-
"epoch": 15.67,
|
1572 |
-
"learning_rate": 1.1731578947368421e-09,
|
1573 |
-
"loss": 0.0002,
|
1574 |
-
"step": 6300
|
1575 |
-
},
|
1576 |
-
{
|
1577 |
-
"epoch": 15.73,
|
1578 |
-
"learning_rate": 1.1652631578947368e-09,
|
1579 |
-
"loss": 0.0002,
|
1580 |
-
"step": 6325
|
1581 |
-
},
|
1582 |
-
{
|
1583 |
-
"epoch": 15.8,
|
1584 |
-
"learning_rate": 1.1573684210526315e-09,
|
1585 |
-
"loss": 0.0002,
|
1586 |
-
"step": 6350
|
1587 |
-
},
|
1588 |
-
{
|
1589 |
-
"epoch": 15.86,
|
1590 |
-
"learning_rate": 1.1494736842105264e-09,
|
1591 |
-
"loss": 0.0002,
|
1592 |
-
"step": 6375
|
1593 |
-
},
|
1594 |
-
{
|
1595 |
-
"epoch": 15.92,
|
1596 |
-
"learning_rate": 1.141578947368421e-09,
|
1597 |
-
"loss": 0.0002,
|
1598 |
-
"step": 6400
|
1599 |
-
},
|
1600 |
-
{
|
1601 |
-
"epoch": 15.98,
|
1602 |
-
"learning_rate": 1.1336842105263159e-09,
|
1603 |
-
"loss": 0.0002,
|
1604 |
-
"step": 6425
|
1605 |
-
},
|
1606 |
-
{
|
1607 |
-
"epoch": 16.04,
|
1608 |
-
"learning_rate": 1.1257894736842106e-09,
|
1609 |
-
"loss": 0.0002,
|
1610 |
-
"step": 6450
|
1611 |
-
},
|
1612 |
-
{
|
1613 |
-
"epoch": 16.11,
|
1614 |
-
"learning_rate": 1.1178947368421052e-09,
|
1615 |
-
"loss": 0.0002,
|
1616 |
-
"step": 6475
|
1617 |
-
},
|
1618 |
-
{
|
1619 |
-
"epoch": 16.17,
|
1620 |
-
"learning_rate": 1.1099999999999999e-09,
|
1621 |
-
"loss": 0.0002,
|
1622 |
-
"step": 6500
|
1623 |
-
},
|
1624 |
-
{
|
1625 |
-
"epoch": 16.23,
|
1626 |
-
"learning_rate": 1.1021052631578948e-09,
|
1627 |
-
"loss": 0.0002,
|
1628 |
-
"step": 6525
|
1629 |
-
},
|
1630 |
-
{
|
1631 |
-
"epoch": 16.29,
|
1632 |
-
"learning_rate": 1.0942105263157894e-09,
|
1633 |
-
"loss": 0.0002,
|
1634 |
-
"step": 6550
|
1635 |
-
},
|
1636 |
-
{
|
1637 |
-
"epoch": 16.36,
|
1638 |
-
"learning_rate": 1.0863157894736843e-09,
|
1639 |
-
"loss": 0.0002,
|
1640 |
-
"step": 6575
|
1641 |
-
},
|
1642 |
-
{
|
1643 |
-
"epoch": 16.42,
|
1644 |
-
"learning_rate": 1.078421052631579e-09,
|
1645 |
-
"loss": 0.0002,
|
1646 |
-
"step": 6600
|
1647 |
-
},
|
1648 |
-
{
|
1649 |
-
"epoch": 16.48,
|
1650 |
-
"learning_rate": 1.0705263157894736e-09,
|
1651 |
-
"loss": 0.0002,
|
1652 |
-
"step": 6625
|
1653 |
-
},
|
1654 |
-
{
|
1655 |
-
"epoch": 16.54,
|
1656 |
-
"learning_rate": 1.0626315789473685e-09,
|
1657 |
-
"loss": 0.0002,
|
1658 |
-
"step": 6650
|
1659 |
-
},
|
1660 |
-
{
|
1661 |
-
"epoch": 16.6,
|
1662 |
-
"learning_rate": 1.0547368421052632e-09,
|
1663 |
-
"loss": 0.0002,
|
1664 |
-
"step": 6675
|
1665 |
-
},
|
1666 |
-
{
|
1667 |
-
"epoch": 16.67,
|
1668 |
-
"learning_rate": 1.0468421052631578e-09,
|
1669 |
-
"loss": 0.0002,
|
1670 |
-
"step": 6700
|
1671 |
-
},
|
1672 |
-
{
|
1673 |
-
"epoch": 16.73,
|
1674 |
-
"learning_rate": 1.0389473684210525e-09,
|
1675 |
-
"loss": 0.0002,
|
1676 |
-
"step": 6725
|
1677 |
-
},
|
1678 |
-
{
|
1679 |
-
"epoch": 16.79,
|
1680 |
-
"learning_rate": 1.0310526315789474e-09,
|
1681 |
-
"loss": 0.0002,
|
1682 |
-
"step": 6750
|
1683 |
-
},
|
1684 |
-
{
|
1685 |
-
"epoch": 16.85,
|
1686 |
-
"learning_rate": 1.0231578947368423e-09,
|
1687 |
-
"loss": 0.0002,
|
1688 |
-
"step": 6775
|
1689 |
-
},
|
1690 |
-
{
|
1691 |
-
"epoch": 16.92,
|
1692 |
-
"learning_rate": 1.015263157894737e-09,
|
1693 |
-
"loss": 0.0002,
|
1694 |
-
"step": 6800
|
1695 |
-
},
|
1696 |
-
{
|
1697 |
-
"epoch": 16.98,
|
1698 |
-
"learning_rate": 1.0073684210526316e-09,
|
1699 |
-
"loss": 0.0002,
|
1700 |
-
"step": 6825
|
1701 |
-
},
|
1702 |
-
{
|
1703 |
-
"epoch": 17.04,
|
1704 |
-
"learning_rate": 9.994736842105262e-10,
|
1705 |
-
"loss": 0.0002,
|
1706 |
-
"step": 6850
|
1707 |
-
},
|
1708 |
-
{
|
1709 |
-
"epoch": 17.1,
|
1710 |
-
"learning_rate": 9.91578947368421e-10,
|
1711 |
-
"loss": 0.0002,
|
1712 |
-
"step": 6875
|
1713 |
-
},
|
1714 |
-
{
|
1715 |
-
"epoch": 17.16,
|
1716 |
-
"learning_rate": 9.836842105263158e-10,
|
1717 |
-
"loss": 0.0002,
|
1718 |
-
"step": 6900
|
1719 |
-
},
|
1720 |
-
{
|
1721 |
-
"epoch": 17.23,
|
1722 |
-
"learning_rate": 9.757894736842107e-10,
|
1723 |
-
"loss": 0.0002,
|
1724 |
-
"step": 6925
|
1725 |
-
},
|
1726 |
-
{
|
1727 |
-
"epoch": 17.29,
|
1728 |
-
"learning_rate": 9.678947368421053e-10,
|
1729 |
-
"loss": 0.0002,
|
1730 |
-
"step": 6950
|
1731 |
-
},
|
1732 |
-
{
|
1733 |
-
"epoch": 17.35,
|
1734 |
-
"learning_rate": 9.6e-10,
|
1735 |
-
"loss": 0.0002,
|
1736 |
-
"step": 6975
|
1737 |
-
},
|
1738 |
-
{
|
1739 |
-
"epoch": 17.41,
|
1740 |
-
"learning_rate": 9.521052631578947e-10,
|
1741 |
-
"loss": 0.0002,
|
1742 |
-
"step": 7000
|
1743 |
-
},
|
1744 |
-
{
|
1745 |
-
"epoch": 17.41,
|
1746 |
-
"eval_loss": 0.50146484375,
|
1747 |
-
"eval_runtime": 167.242,
|
1748 |
-
"eval_samples_per_second": 10.141,
|
1749 |
-
"eval_steps_per_second": 0.634,
|
1750 |
-
"eval_wer": 16.539747399702822,
|
1751 |
-
"step": 7000
|
1752 |
-
},
|
1753 |
-
{
|
1754 |
-
"epoch": 17.48,
|
1755 |
-
"learning_rate": 9.44842105263158e-10,
|
1756 |
-
"loss": 0.0002,
|
1757 |
-
"step": 7025
|
1758 |
-
},
|
1759 |
-
{
|
1760 |
-
"epoch": 17.54,
|
1761 |
-
"learning_rate": 9.369473684210527e-10,
|
1762 |
-
"loss": 0.0002,
|
1763 |
-
"step": 7050
|
1764 |
-
},
|
1765 |
-
{
|
1766 |
-
"epoch": 17.6,
|
1767 |
-
"learning_rate": 9.290526315789473e-10,
|
1768 |
-
"loss": 0.0002,
|
1769 |
-
"step": 7075
|
1770 |
-
},
|
1771 |
-
{
|
1772 |
-
"epoch": 17.66,
|
1773 |
-
"learning_rate": 9.211578947368421e-10,
|
1774 |
-
"loss": 0.0002,
|
1775 |
-
"step": 7100
|
1776 |
-
},
|
1777 |
-
{
|
1778 |
-
"epoch": 17.72,
|
1779 |
-
"learning_rate": 9.132631578947368e-10,
|
1780 |
-
"loss": 0.0002,
|
1781 |
-
"step": 7125
|
1782 |
-
},
|
1783 |
-
{
|
1784 |
-
"epoch": 17.79,
|
1785 |
-
"learning_rate": 9.053684210526316e-10,
|
1786 |
-
"loss": 0.0002,
|
1787 |
-
"step": 7150
|
1788 |
-
},
|
1789 |
-
{
|
1790 |
-
"epoch": 17.85,
|
1791 |
-
"learning_rate": 8.974736842105264e-10,
|
1792 |
-
"loss": 0.0001,
|
1793 |
-
"step": 7175
|
1794 |
-
},
|
1795 |
-
{
|
1796 |
-
"epoch": 17.91,
|
1797 |
-
"learning_rate": 8.895789473684211e-10,
|
1798 |
-
"loss": 0.0002,
|
1799 |
-
"step": 7200
|
1800 |
-
},
|
1801 |
-
{
|
1802 |
-
"epoch": 17.97,
|
1803 |
-
"learning_rate": 8.816842105263158e-10,
|
1804 |
-
"loss": 0.0002,
|
1805 |
-
"step": 7225
|
1806 |
-
},
|
1807 |
-
{
|
1808 |
-
"epoch": 18.03,
|
1809 |
-
"learning_rate": 8.737894736842105e-10,
|
1810 |
-
"loss": 0.0002,
|
1811 |
-
"step": 7250
|
1812 |
-
},
|
1813 |
-
{
|
1814 |
-
"epoch": 18.1,
|
1815 |
-
"learning_rate": 8.658947368421052e-10,
|
1816 |
-
"loss": 0.0002,
|
1817 |
-
"step": 7275
|
1818 |
-
},
|
1819 |
-
{
|
1820 |
-
"epoch": 18.16,
|
1821 |
-
"learning_rate": 8.579999999999999e-10,
|
1822 |
-
"loss": 0.0002,
|
1823 |
-
"step": 7300
|
1824 |
-
},
|
1825 |
-
{
|
1826 |
-
"epoch": 18.22,
|
1827 |
-
"learning_rate": 8.501052631578948e-10,
|
1828 |
-
"loss": 0.0002,
|
1829 |
-
"step": 7325
|
1830 |
-
},
|
1831 |
-
{
|
1832 |
-
"epoch": 18.28,
|
1833 |
-
"learning_rate": 8.422105263157895e-10,
|
1834 |
-
"loss": 0.0002,
|
1835 |
-
"step": 7350
|
1836 |
-
},
|
1837 |
-
{
|
1838 |
-
"epoch": 18.35,
|
1839 |
-
"learning_rate": 8.343157894736842e-10,
|
1840 |
-
"loss": 0.0002,
|
1841 |
-
"step": 7375
|
1842 |
-
},
|
1843 |
-
{
|
1844 |
-
"epoch": 18.41,
|
1845 |
-
"learning_rate": 8.264210526315789e-10,
|
1846 |
-
"loss": 0.0002,
|
1847 |
-
"step": 7400
|
1848 |
-
},
|
1849 |
-
{
|
1850 |
-
"epoch": 18.47,
|
1851 |
-
"learning_rate": 8.185263157894737e-10,
|
1852 |
-
"loss": 0.0002,
|
1853 |
-
"step": 7425
|
1854 |
-
},
|
1855 |
-
{
|
1856 |
-
"epoch": 18.53,
|
1857 |
-
"learning_rate": 8.106315789473683e-10,
|
1858 |
-
"loss": 0.0002,
|
1859 |
-
"step": 7450
|
1860 |
-
},
|
1861 |
-
{
|
1862 |
-
"epoch": 18.59,
|
1863 |
-
"learning_rate": 8.027368421052631e-10,
|
1864 |
-
"loss": 0.0002,
|
1865 |
-
"step": 7475
|
1866 |
-
},
|
1867 |
-
{
|
1868 |
-
"epoch": 18.66,
|
1869 |
-
"learning_rate": 7.94842105263158e-10,
|
1870 |
-
"loss": 0.0002,
|
1871 |
-
"step": 7500
|
1872 |
-
},
|
1873 |
-
{
|
1874 |
-
"epoch": 18.72,
|
1875 |
-
"learning_rate": 7.869473684210527e-10,
|
1876 |
-
"loss": 0.0002,
|
1877 |
-
"step": 7525
|
1878 |
-
},
|
1879 |
-
{
|
1880 |
-
"epoch": 18.78,
|
1881 |
-
"learning_rate": 7.790526315789474e-10,
|
1882 |
-
"loss": 0.0002,
|
1883 |
-
"step": 7550
|
1884 |
-
},
|
1885 |
-
{
|
1886 |
-
"epoch": 18.84,
|
1887 |
-
"learning_rate": 7.711578947368421e-10,
|
1888 |
-
"loss": 0.0002,
|
1889 |
-
"step": 7575
|
1890 |
-
},
|
1891 |
-
{
|
1892 |
-
"epoch": 18.91,
|
1893 |
-
"learning_rate": 7.632631578947369e-10,
|
1894 |
-
"loss": 0.0002,
|
1895 |
-
"step": 7600
|
1896 |
-
},
|
1897 |
-
{
|
1898 |
-
"epoch": 18.97,
|
1899 |
-
"learning_rate": 7.553684210526315e-10,
|
1900 |
-
"loss": 0.0002,
|
1901 |
-
"step": 7625
|
1902 |
-
},
|
1903 |
-
{
|
1904 |
-
"epoch": 19.03,
|
1905 |
-
"learning_rate": 7.474736842105263e-10,
|
1906 |
-
"loss": 0.0002,
|
1907 |
-
"step": 7650
|
1908 |
-
},
|
1909 |
-
{
|
1910 |
-
"epoch": 19.09,
|
1911 |
-
"learning_rate": 7.395789473684211e-10,
|
1912 |
-
"loss": 0.0002,
|
1913 |
-
"step": 7675
|
1914 |
-
},
|
1915 |
-
{
|
1916 |
-
"epoch": 19.15,
|
1917 |
-
"learning_rate": 7.316842105263158e-10,
|
1918 |
-
"loss": 0.0002,
|
1919 |
-
"step": 7700
|
1920 |
-
},
|
1921 |
-
{
|
1922 |
-
"epoch": 19.22,
|
1923 |
-
"learning_rate": 7.237894736842105e-10,
|
1924 |
-
"loss": 0.0002,
|
1925 |
-
"step": 7725
|
1926 |
-
},
|
1927 |
-
{
|
1928 |
-
"epoch": 19.28,
|
1929 |
-
"learning_rate": 7.158947368421053e-10,
|
1930 |
-
"loss": 0.0002,
|
1931 |
-
"step": 7750
|
1932 |
-
},
|
1933 |
-
{
|
1934 |
-
"epoch": 19.34,
|
1935 |
-
"learning_rate": 7.079999999999999e-10,
|
1936 |
-
"loss": 0.0002,
|
1937 |
-
"step": 7775
|
1938 |
-
},
|
1939 |
-
{
|
1940 |
-
"epoch": 19.4,
|
1941 |
-
"learning_rate": 7.001052631578947e-10,
|
1942 |
-
"loss": 0.0002,
|
1943 |
-
"step": 7800
|
1944 |
-
},
|
1945 |
-
{
|
1946 |
-
"epoch": 19.47,
|
1947 |
-
"learning_rate": 6.922105263157895e-10,
|
1948 |
-
"loss": 0.0002,
|
1949 |
-
"step": 7825
|
1950 |
-
},
|
1951 |
-
{
|
1952 |
-
"epoch": 19.53,
|
1953 |
-
"learning_rate": 6.843157894736841e-10,
|
1954 |
-
"loss": 0.0002,
|
1955 |
-
"step": 7850
|
1956 |
-
},
|
1957 |
-
{
|
1958 |
-
"epoch": 19.59,
|
1959 |
-
"learning_rate": 6.76421052631579e-10,
|
1960 |
-
"loss": 0.0002,
|
1961 |
-
"step": 7875
|
1962 |
-
},
|
1963 |
-
{
|
1964 |
-
"epoch": 19.65,
|
1965 |
-
"learning_rate": 6.685263157894737e-10,
|
1966 |
-
"loss": 0.0002,
|
1967 |
-
"step": 7900
|
1968 |
-
},
|
1969 |
-
{
|
1970 |
-
"epoch": 19.71,
|
1971 |
-
"learning_rate": 6.606315789473684e-10,
|
1972 |
-
"loss": 0.0002,
|
1973 |
-
"step": 7925
|
1974 |
-
},
|
1975 |
-
{
|
1976 |
-
"epoch": 19.78,
|
1977 |
-
"learning_rate": 6.527368421052632e-10,
|
1978 |
-
"loss": 0.0002,
|
1979 |
-
"step": 7950
|
1980 |
-
},
|
1981 |
-
{
|
1982 |
-
"epoch": 19.84,
|
1983 |
-
"learning_rate": 6.448421052631579e-10,
|
1984 |
-
"loss": 0.0002,
|
1985 |
-
"step": 7975
|
1986 |
-
},
|
1987 |
-
{
|
1988 |
-
"epoch": 19.9,
|
1989 |
-
"learning_rate": 6.369473684210527e-10,
|
1990 |
-
"loss": 0.0002,
|
1991 |
-
"step": 8000
|
1992 |
-
},
|
1993 |
-
{
|
1994 |
-
"epoch": 19.9,
|
1995 |
-
"eval_loss": 0.50146484375,
|
1996 |
-
"eval_runtime": 167.9653,
|
1997 |
-
"eval_samples_per_second": 10.097,
|
1998 |
-
"eval_steps_per_second": 0.631,
|
1999 |
-
"eval_wer": 16.50260029717682,
|
2000 |
-
"step": 8000
|
2001 |
-
},
|
2002 |
-
{
|
2003 |
-
"epoch": 19.96,
|
2004 |
-
"learning_rate": 6.296842105263158e-10,
|
2005 |
-
"loss": 0.0002,
|
2006 |
-
"step": 8025
|
2007 |
-
},
|
2008 |
-
{
|
2009 |
-
"epoch": 20.02,
|
2010 |
-
"learning_rate": 6.217894736842105e-10,
|
2011 |
-
"loss": 0.0002,
|
2012 |
-
"step": 8050
|
2013 |
-
},
|
2014 |
-
{
|
2015 |
-
"epoch": 20.09,
|
2016 |
-
"learning_rate": 6.138947368421052e-10,
|
2017 |
-
"loss": 0.0002,
|
2018 |
-
"step": 8075
|
2019 |
-
},
|
2020 |
-
{
|
2021 |
-
"epoch": 20.15,
|
2022 |
-
"learning_rate": 6.060000000000001e-10,
|
2023 |
-
"loss": 0.0002,
|
2024 |
-
"step": 8100
|
2025 |
-
},
|
2026 |
-
{
|
2027 |
-
"epoch": 20.21,
|
2028 |
-
"learning_rate": 5.981052631578947e-10,
|
2029 |
-
"loss": 0.0002,
|
2030 |
-
"step": 8125
|
2031 |
-
},
|
2032 |
-
{
|
2033 |
-
"epoch": 20.27,
|
2034 |
-
"learning_rate": 5.902105263157894e-10,
|
2035 |
-
"loss": 0.0002,
|
2036 |
-
"step": 8150
|
2037 |
-
},
|
2038 |
-
{
|
2039 |
-
"epoch": 20.34,
|
2040 |
-
"learning_rate": 5.823157894736842e-10,
|
2041 |
-
"loss": 0.0002,
|
2042 |
-
"step": 8175
|
2043 |
-
},
|
2044 |
-
{
|
2045 |
-
"epoch": 20.4,
|
2046 |
-
"learning_rate": 5.74421052631579e-10,
|
2047 |
-
"loss": 0.0002,
|
2048 |
-
"step": 8200
|
2049 |
-
},
|
2050 |
-
{
|
2051 |
-
"epoch": 20.46,
|
2052 |
-
"learning_rate": 5.665263157894737e-10,
|
2053 |
-
"loss": 0.0002,
|
2054 |
-
"step": 8225
|
2055 |
-
},
|
2056 |
-
{
|
2057 |
-
"epoch": 20.52,
|
2058 |
-
"learning_rate": 5.586315789473684e-10,
|
2059 |
-
"loss": 0.0002,
|
2060 |
-
"step": 8250
|
2061 |
-
},
|
2062 |
-
{
|
2063 |
-
"epoch": 20.58,
|
2064 |
-
"learning_rate": 5.507368421052632e-10,
|
2065 |
-
"loss": 0.0002,
|
2066 |
-
"step": 8275
|
2067 |
-
},
|
2068 |
-
{
|
2069 |
-
"epoch": 20.65,
|
2070 |
-
"learning_rate": 5.428421052631579e-10,
|
2071 |
-
"loss": 0.0002,
|
2072 |
-
"step": 8300
|
2073 |
-
},
|
2074 |
-
{
|
2075 |
-
"epoch": 20.71,
|
2076 |
-
"learning_rate": 5.349473684210526e-10,
|
2077 |
-
"loss": 0.0002,
|
2078 |
-
"step": 8325
|
2079 |
-
},
|
2080 |
-
{
|
2081 |
-
"epoch": 20.77,
|
2082 |
-
"learning_rate": 5.270526315789474e-10,
|
2083 |
-
"loss": 0.0002,
|
2084 |
-
"step": 8350
|
2085 |
-
},
|
2086 |
-
{
|
2087 |
-
"epoch": 20.83,
|
2088 |
-
"learning_rate": 5.191578947368421e-10,
|
2089 |
-
"loss": 0.0002,
|
2090 |
-
"step": 8375
|
2091 |
-
},
|
2092 |
-
{
|
2093 |
-
"epoch": 20.9,
|
2094 |
-
"learning_rate": 5.112631578947368e-10,
|
2095 |
-
"loss": 0.0002,
|
2096 |
-
"step": 8400
|
2097 |
-
},
|
2098 |
-
{
|
2099 |
-
"epoch": 20.96,
|
2100 |
-
"learning_rate": 5.033684210526316e-10,
|
2101 |
-
"loss": 0.0002,
|
2102 |
-
"step": 8425
|
2103 |
-
},
|
2104 |
-
{
|
2105 |
-
"epoch": 21.02,
|
2106 |
-
"learning_rate": 4.954736842105263e-10,
|
2107 |
-
"loss": 0.0002,
|
2108 |
-
"step": 8450
|
2109 |
-
},
|
2110 |
-
{
|
2111 |
-
"epoch": 21.08,
|
2112 |
-
"learning_rate": 4.87578947368421e-10,
|
2113 |
-
"loss": 0.0002,
|
2114 |
-
"step": 8475
|
2115 |
-
},
|
2116 |
-
{
|
2117 |
-
"epoch": 21.14,
|
2118 |
-
"learning_rate": 4.796842105263158e-10,
|
2119 |
-
"loss": 0.0002,
|
2120 |
-
"step": 8500
|
2121 |
-
},
|
2122 |
-
{
|
2123 |
-
"epoch": 21.21,
|
2124 |
-
"learning_rate": 4.717894736842105e-10,
|
2125 |
-
"loss": 0.0002,
|
2126 |
-
"step": 8525
|
2127 |
-
},
|
2128 |
-
{
|
2129 |
-
"epoch": 21.27,
|
2130 |
-
"learning_rate": 4.6389473684210526e-10,
|
2131 |
-
"loss": 0.0002,
|
2132 |
-
"step": 8550
|
2133 |
-
},
|
2134 |
-
{
|
2135 |
-
"epoch": 21.33,
|
2136 |
-
"learning_rate": 4.56e-10,
|
2137 |
-
"loss": 0.0002,
|
2138 |
-
"step": 8575
|
2139 |
-
},
|
2140 |
-
{
|
2141 |
-
"epoch": 21.39,
|
2142 |
-
"learning_rate": 4.481052631578947e-10,
|
2143 |
-
"loss": 0.0002,
|
2144 |
-
"step": 8600
|
2145 |
-
},
|
2146 |
-
{
|
2147 |
-
"epoch": 21.46,
|
2148 |
-
"learning_rate": 4.402105263157895e-10,
|
2149 |
-
"loss": 0.0002,
|
2150 |
-
"step": 8625
|
2151 |
-
},
|
2152 |
-
{
|
2153 |
-
"epoch": 21.52,
|
2154 |
-
"learning_rate": 4.323157894736842e-10,
|
2155 |
-
"loss": 0.0002,
|
2156 |
-
"step": 8650
|
2157 |
-
},
|
2158 |
-
{
|
2159 |
-
"epoch": 21.58,
|
2160 |
-
"learning_rate": 4.244210526315789e-10,
|
2161 |
-
"loss": 0.0002,
|
2162 |
-
"step": 8675
|
2163 |
-
},
|
2164 |
-
{
|
2165 |
-
"epoch": 21.64,
|
2166 |
-
"learning_rate": 4.165263157894737e-10,
|
2167 |
-
"loss": 0.0002,
|
2168 |
-
"step": 8700
|
2169 |
-
},
|
2170 |
-
{
|
2171 |
-
"epoch": 21.7,
|
2172 |
-
"learning_rate": 4.0863157894736844e-10,
|
2173 |
-
"loss": 0.0002,
|
2174 |
-
"step": 8725
|
2175 |
-
},
|
2176 |
-
{
|
2177 |
-
"epoch": 21.77,
|
2178 |
-
"learning_rate": 4.0073684210526316e-10,
|
2179 |
-
"loss": 0.0002,
|
2180 |
-
"step": 8750
|
2181 |
-
},
|
2182 |
-
{
|
2183 |
-
"epoch": 21.83,
|
2184 |
-
"learning_rate": 3.928421052631579e-10,
|
2185 |
-
"loss": 0.0002,
|
2186 |
-
"step": 8775
|
2187 |
-
},
|
2188 |
-
{
|
2189 |
-
"epoch": 21.89,
|
2190 |
-
"learning_rate": 3.8494736842105264e-10,
|
2191 |
-
"loss": 0.0002,
|
2192 |
-
"step": 8800
|
2193 |
-
},
|
2194 |
-
{
|
2195 |
-
"epoch": 21.95,
|
2196 |
-
"learning_rate": 3.7705263157894736e-10,
|
2197 |
-
"loss": 0.0002,
|
2198 |
-
"step": 8825
|
2199 |
-
},
|
2200 |
-
{
|
2201 |
-
"epoch": 22.01,
|
2202 |
-
"learning_rate": 3.6915789473684213e-10,
|
2203 |
-
"loss": 0.0002,
|
2204 |
-
"step": 8850
|
2205 |
-
},
|
2206 |
-
{
|
2207 |
-
"epoch": 22.08,
|
2208 |
-
"learning_rate": 3.6126315789473685e-10,
|
2209 |
-
"loss": 0.0002,
|
2210 |
-
"step": 8875
|
2211 |
-
},
|
2212 |
-
{
|
2213 |
-
"epoch": 22.14,
|
2214 |
-
"learning_rate": 3.5336842105263157e-10,
|
2215 |
-
"loss": 0.0002,
|
2216 |
-
"step": 8900
|
2217 |
-
},
|
2218 |
-
{
|
2219 |
-
"epoch": 22.2,
|
2220 |
-
"learning_rate": 3.4547368421052634e-10,
|
2221 |
-
"loss": 0.0002,
|
2222 |
-
"step": 8925
|
2223 |
-
},
|
2224 |
-
{
|
2225 |
-
"epoch": 22.26,
|
2226 |
-
"learning_rate": 3.3757894736842105e-10,
|
2227 |
-
"loss": 0.0002,
|
2228 |
-
"step": 8950
|
2229 |
-
},
|
2230 |
-
{
|
2231 |
-
"epoch": 22.33,
|
2232 |
-
"learning_rate": 3.296842105263158e-10,
|
2233 |
-
"loss": 0.0002,
|
2234 |
-
"step": 8975
|
2235 |
-
},
|
2236 |
-
{
|
2237 |
-
"epoch": 22.39,
|
2238 |
-
"learning_rate": 3.217894736842105e-10,
|
2239 |
-
"loss": 0.0002,
|
2240 |
-
"step": 9000
|
2241 |
-
},
|
2242 |
-
{
|
2243 |
-
"epoch": 22.39,
|
2244 |
-
"eval_loss": 0.50146484375,
|
2245 |
-
"eval_runtime": 167.9301,
|
2246 |
-
"eval_samples_per_second": 10.099,
|
2247 |
-
"eval_steps_per_second": 0.631,
|
2248 |
-
"eval_wer": 16.521173848439823,
|
2249 |
-
"step": 9000
|
2250 |
-
},
|
2251 |
-
{
|
2252 |
-
"epoch": 22.45,
|
2253 |
-
"learning_rate": 3.1452631578947366e-10,
|
2254 |
-
"loss": 0.0002,
|
2255 |
-
"step": 9025
|
2256 |
-
},
|
2257 |
-
{
|
2258 |
-
"epoch": 22.51,
|
2259 |
-
"learning_rate": 3.0663157894736843e-10,
|
2260 |
-
"loss": 0.0002,
|
2261 |
-
"step": 9050
|
2262 |
-
},
|
2263 |
-
{
|
2264 |
-
"epoch": 22.57,
|
2265 |
-
"learning_rate": 2.987368421052632e-10,
|
2266 |
-
"loss": 0.0002,
|
2267 |
-
"step": 9075
|
2268 |
-
},
|
2269 |
-
{
|
2270 |
-
"epoch": 22.64,
|
2271 |
-
"learning_rate": 2.9084210526315786e-10,
|
2272 |
-
"loss": 0.0002,
|
2273 |
-
"step": 9100
|
2274 |
-
},
|
2275 |
-
{
|
2276 |
-
"epoch": 22.7,
|
2277 |
-
"learning_rate": 2.8294736842105263e-10,
|
2278 |
-
"loss": 0.0002,
|
2279 |
-
"step": 9125
|
2280 |
-
},
|
2281 |
-
{
|
2282 |
-
"epoch": 22.76,
|
2283 |
-
"learning_rate": 2.7505263157894735e-10,
|
2284 |
-
"loss": 0.0001,
|
2285 |
-
"step": 9150
|
2286 |
-
},
|
2287 |
-
{
|
2288 |
-
"epoch": 22.82,
|
2289 |
-
"learning_rate": 2.671578947368421e-10,
|
2290 |
-
"loss": 0.0002,
|
2291 |
-
"step": 9175
|
2292 |
-
},
|
2293 |
-
{
|
2294 |
-
"epoch": 22.89,
|
2295 |
-
"learning_rate": 2.5926315789473684e-10,
|
2296 |
-
"loss": 0.0002,
|
2297 |
-
"step": 9200
|
2298 |
-
},
|
2299 |
-
{
|
2300 |
-
"epoch": 22.95,
|
2301 |
-
"learning_rate": 2.5136842105263156e-10,
|
2302 |
-
"loss": 0.0002,
|
2303 |
-
"step": 9225
|
2304 |
-
},
|
2305 |
-
{
|
2306 |
-
"epoch": 23.01,
|
2307 |
-
"learning_rate": 2.4347368421052633e-10,
|
2308 |
-
"loss": 0.0002,
|
2309 |
-
"step": 9250
|
2310 |
-
},
|
2311 |
-
{
|
2312 |
-
"epoch": 23.07,
|
2313 |
-
"learning_rate": 2.3557894736842104e-10,
|
2314 |
-
"loss": 0.0002,
|
2315 |
-
"step": 9275
|
2316 |
-
},
|
2317 |
-
{
|
2318 |
-
"epoch": 23.13,
|
2319 |
-
"learning_rate": 2.276842105263158e-10,
|
2320 |
-
"loss": 0.0002,
|
2321 |
-
"step": 9300
|
2322 |
-
},
|
2323 |
-
{
|
2324 |
-
"epoch": 23.2,
|
2325 |
-
"learning_rate": 2.197894736842105e-10,
|
2326 |
-
"loss": 0.0002,
|
2327 |
-
"step": 9325
|
2328 |
-
},
|
2329 |
-
{
|
2330 |
-
"epoch": 23.26,
|
2331 |
-
"learning_rate": 2.1189473684210525e-10,
|
2332 |
-
"loss": 0.0002,
|
2333 |
-
"step": 9350
|
2334 |
-
},
|
2335 |
-
{
|
2336 |
-
"epoch": 23.32,
|
2337 |
-
"learning_rate": 2.0400000000000002e-10,
|
2338 |
-
"loss": 0.0002,
|
2339 |
-
"step": 9375
|
2340 |
-
},
|
2341 |
-
{
|
2342 |
-
"epoch": 23.38,
|
2343 |
-
"learning_rate": 1.9610526315789474e-10,
|
2344 |
-
"loss": 0.0002,
|
2345 |
-
"step": 9400
|
2346 |
-
},
|
2347 |
-
{
|
2348 |
-
"epoch": 23.45,
|
2349 |
-
"learning_rate": 1.8821052631578948e-10,
|
2350 |
-
"loss": 0.0002,
|
2351 |
-
"step": 9425
|
2352 |
-
},
|
2353 |
-
{
|
2354 |
-
"epoch": 23.51,
|
2355 |
-
"learning_rate": 1.8031578947368422e-10,
|
2356 |
-
"loss": 0.0002,
|
2357 |
-
"step": 9450
|
2358 |
-
},
|
2359 |
-
{
|
2360 |
-
"epoch": 23.57,
|
2361 |
-
"learning_rate": 1.7242105263157894e-10,
|
2362 |
-
"loss": 0.0002,
|
2363 |
-
"step": 9475
|
2364 |
-
},
|
2365 |
-
{
|
2366 |
-
"epoch": 23.63,
|
2367 |
-
"learning_rate": 1.6452631578947369e-10,
|
2368 |
-
"loss": 0.0002,
|
2369 |
-
"step": 9500
|
2370 |
-
},
|
2371 |
-
{
|
2372 |
-
"epoch": 23.69,
|
2373 |
-
"learning_rate": 1.566315789473684e-10,
|
2374 |
-
"loss": 0.0002,
|
2375 |
-
"step": 9525
|
2376 |
-
},
|
2377 |
-
{
|
2378 |
-
"epoch": 23.76,
|
2379 |
-
"learning_rate": 1.4873684210526317e-10,
|
2380 |
-
"loss": 0.0002,
|
2381 |
-
"step": 9550
|
2382 |
-
},
|
2383 |
-
{
|
2384 |
-
"epoch": 23.82,
|
2385 |
-
"learning_rate": 1.408421052631579e-10,
|
2386 |
-
"loss": 0.0002,
|
2387 |
-
"step": 9575
|
2388 |
-
},
|
2389 |
-
{
|
2390 |
-
"epoch": 23.88,
|
2391 |
-
"learning_rate": 1.3294736842105264e-10,
|
2392 |
-
"loss": 0.0002,
|
2393 |
-
"step": 9600
|
2394 |
-
},
|
2395 |
-
{
|
2396 |
-
"epoch": 23.94,
|
2397 |
-
"learning_rate": 1.2505263157894735e-10,
|
2398 |
-
"loss": 0.0002,
|
2399 |
-
"step": 9625
|
2400 |
-
},
|
2401 |
-
{
|
2402 |
-
"epoch": 24.0,
|
2403 |
-
"learning_rate": 1.171578947368421e-10,
|
2404 |
-
"loss": 0.0002,
|
2405 |
-
"step": 9650
|
2406 |
-
},
|
2407 |
-
{
|
2408 |
-
"epoch": 24.07,
|
2409 |
-
"learning_rate": 1.0926315789473683e-10,
|
2410 |
-
"loss": 0.0002,
|
2411 |
-
"step": 9675
|
2412 |
-
},
|
2413 |
-
{
|
2414 |
-
"epoch": 24.13,
|
2415 |
-
"learning_rate": 1.0136842105263158e-10,
|
2416 |
-
"loss": 0.0002,
|
2417 |
-
"step": 9700
|
2418 |
-
},
|
2419 |
-
{
|
2420 |
-
"epoch": 24.19,
|
2421 |
-
"learning_rate": 9.347368421052631e-11,
|
2422 |
-
"loss": 0.0002,
|
2423 |
-
"step": 9725
|
2424 |
-
},
|
2425 |
-
{
|
2426 |
-
"epoch": 24.25,
|
2427 |
-
"learning_rate": 8.557894736842106e-11,
|
2428 |
-
"loss": 0.0002,
|
2429 |
-
"step": 9750
|
2430 |
-
},
|
2431 |
-
{
|
2432 |
-
"epoch": 24.32,
|
2433 |
-
"learning_rate": 7.768421052631579e-11,
|
2434 |
-
"loss": 0.0002,
|
2435 |
-
"step": 9775
|
2436 |
-
},
|
2437 |
-
{
|
2438 |
-
"epoch": 24.38,
|
2439 |
-
"learning_rate": 6.978947368421053e-11,
|
2440 |
-
"loss": 0.0002,
|
2441 |
-
"step": 9800
|
2442 |
-
},
|
2443 |
-
{
|
2444 |
-
"epoch": 24.44,
|
2445 |
-
"learning_rate": 6.189473684210526e-11,
|
2446 |
-
"loss": 0.0002,
|
2447 |
-
"step": 9825
|
2448 |
-
},
|
2449 |
-
{
|
2450 |
-
"epoch": 24.5,
|
2451 |
-
"learning_rate": 5.3999999999999994e-11,
|
2452 |
-
"loss": 0.0002,
|
2453 |
-
"step": 9850
|
2454 |
-
},
|
2455 |
-
{
|
2456 |
-
"epoch": 24.56,
|
2457 |
-
"learning_rate": 4.610526315789473e-11,
|
2458 |
-
"loss": 0.0002,
|
2459 |
-
"step": 9875
|
2460 |
-
},
|
2461 |
-
{
|
2462 |
-
"epoch": 24.63,
|
2463 |
-
"learning_rate": 3.8210526315789475e-11,
|
2464 |
-
"loss": 0.0002,
|
2465 |
-
"step": 9900
|
2466 |
-
},
|
2467 |
-
{
|
2468 |
-
"epoch": 24.69,
|
2469 |
-
"learning_rate": 3.031578947368421e-11,
|
2470 |
-
"loss": 0.0002,
|
2471 |
-
"step": 9925
|
2472 |
-
},
|
2473 |
-
{
|
2474 |
-
"epoch": 24.75,
|
2475 |
-
"learning_rate": 2.2421052631578947e-11,
|
2476 |
-
"loss": 0.0002,
|
2477 |
-
"step": 9950
|
2478 |
-
},
|
2479 |
-
{
|
2480 |
-
"epoch": 24.81,
|
2481 |
-
"learning_rate": 1.4526315789473685e-11,
|
2482 |
-
"loss": 0.0002,
|
2483 |
-
"step": 9975
|
2484 |
-
},
|
2485 |
-
{
|
2486 |
-
"epoch": 24.88,
|
2487 |
-
"learning_rate": 6.63157894736842e-12,
|
2488 |
-
"loss": 0.0002,
|
2489 |
-
"step": 10000
|
2490 |
-
},
|
2491 |
-
{
|
2492 |
-
"epoch": 24.88,
|
2493 |
-
"eval_loss": 0.501953125,
|
2494 |
-
"eval_runtime": 167.4253,
|
2495 |
-
"eval_samples_per_second": 10.13,
|
2496 |
-
"eval_steps_per_second": 0.633,
|
2497 |
-
"eval_wer": 16.558320950965825,
|
2498 |
-
"step": 10000
|
2499 |
-
},
|
2500 |
-
{
|
2501 |
-
"epoch": 24.88,
|
2502 |
-
"step": 10000,
|
2503 |
-
"total_flos": 9.23057726024665e+19,
|
2504 |
-
"train_loss": 0.00017381784915924072,
|
2505 |
-
"train_runtime": 19905.1566,
|
2506 |
-
"train_samples_per_second": 16.076,
|
2507 |
-
"train_steps_per_second": 0.502
|
2508 |
}
|
2509 |
],
|
2510 |
-
"max_steps":
|
2511 |
-
"num_train_epochs":
|
2512 |
-
"total_flos":
|
2513 |
"trial_name": null,
|
2514 |
"trial_params": null
|
2515 |
}
|
|
|
1 |
{
|
2 |
+
"best_metric": 16.465453194650816,
|
3 |
+
"best_model_checkpoint": "./checkpoint-3000",
|
4 |
+
"epoch": 12.437810945273633,
|
5 |
+
"global_step": 5000,
|
6 |
"is_hyper_param_search": false,
|
7 |
"is_local_process_zero": true,
|
8 |
"is_world_process_zero": true,
|
9 |
"log_history": [
|
10 |
{
|
11 |
"epoch": 0.06,
|
12 |
+
"learning_rate": 5.045361133432068e-10,
|
13 |
"loss": 0.0002,
|
14 |
"step": 25
|
15 |
},
|
16 |
{
|
17 |
"epoch": 0.12,
|
18 |
+
"learning_rate": 6.229195710491768e-10,
|
19 |
"loss": 0.0002,
|
20 |
"step": 50
|
21 |
},
|
22 |
{
|
23 |
"epoch": 0.19,
|
24 |
+
"learning_rate": 6.903829450223392e-10,
|
25 |
"loss": 0.0002,
|
26 |
"step": 75
|
27 |
},
|
28 |
{
|
29 |
"epoch": 0.25,
|
30 |
+
"learning_rate": 7.377725845391017e-10,
|
31 |
"loss": 0.0002,
|
32 |
"step": 100
|
33 |
},
|
34 |
{
|
35 |
"epoch": 0.31,
|
36 |
+
"learning_rate": 7.743343231239583e-10,
|
37 |
"loss": 0.0002,
|
38 |
"step": 125
|
39 |
},
|
40 |
{
|
41 |
"epoch": 0.37,
|
42 |
+
"learning_rate": 8.041073861170494e-10,
|
43 |
"loss": 0.0002,
|
44 |
"step": 150
|
45 |
},
|
46 |
{
|
47 |
"epoch": 0.44,
|
48 |
+
"learning_rate": 8.292222957399573e-10,
|
49 |
"loss": 0.0002,
|
50 |
"step": 175
|
51 |
},
|
52 |
{
|
53 |
"epoch": 0.5,
|
54 |
+
"learning_rate": 8.509413541357753e-10,
|
55 |
"loss": 0.0002,
|
56 |
"step": 200
|
57 |
},
|
58 |
{
|
59 |
"epoch": 0.56,
|
60 |
+
"learning_rate": 8.700744577655556e-10,
|
61 |
"loss": 0.0002,
|
62 |
"step": 225
|
63 |
},
|
64 |
{
|
65 |
"epoch": 0.62,
|
66 |
+
"learning_rate": 8.871723942761204e-10,
|
67 |
"loss": 0.0002,
|
68 |
"step": 250
|
69 |
},
|
70 |
{
|
71 |
"epoch": 0.68,
|
72 |
+
"learning_rate": 9.026267958246848e-10,
|
73 |
"loss": 0.0002,
|
74 |
"step": 275
|
75 |
},
|
76 |
{
|
77 |
"epoch": 0.75,
|
78 |
+
"learning_rate": 9.16726106663399e-10,
|
79 |
"loss": 0.0002,
|
80 |
"step": 300
|
81 |
},
|
82 |
{
|
83 |
"epoch": 0.81,
|
84 |
+
"learning_rate": 9.296889251455016e-10,
|
85 |
"loss": 0.0002,
|
86 |
"step": 325
|
87 |
},
|
88 |
{
|
89 |
"epoch": 0.87,
|
90 |
+
"learning_rate": 9.416848797368693e-10,
|
91 |
"loss": 0.0002,
|
92 |
"step": 350
|
93 |
},
|
94 |
{
|
95 |
"epoch": 0.93,
|
96 |
+
"learning_rate": 9.528482449516372e-10,
|
97 |
"loss": 0.0002,
|
98 |
"step": 375
|
99 |
},
|
100 |
{
|
101 |
"epoch": 1.0,
|
102 |
+
"learning_rate": 9.632871309784313e-10,
|
103 |
"loss": 0.0002,
|
104 |
"step": 400
|
105 |
},
|
106 |
{
|
107 |
"epoch": 1.06,
|
108 |
+
"learning_rate": 9.73089868785391e-10,
|
109 |
"loss": 0.0002,
|
110 |
"step": 425
|
111 |
},
|
112 |
{
|
113 |
"epoch": 1.12,
|
114 |
+
"learning_rate": 9.823295589572114e-10,
|
115 |
"loss": 0.0002,
|
116 |
"step": 450
|
117 |
},
|
118 |
{
|
119 |
"epoch": 1.18,
|
120 |
+
"learning_rate": 9.910673836465485e-10,
|
121 |
"loss": 0.0002,
|
122 |
"step": 475
|
123 |
},
|
124 |
{
|
125 |
"epoch": 1.24,
|
126 |
+
"learning_rate": 9.993550644973806e-10,
|
127 |
"loss": 0.0002,
|
128 |
"step": 500
|
129 |
},
|
130 |
{
|
131 |
"epoch": 1.31,
|
132 |
+
"learning_rate": 9.95111111111111e-10,
|
133 |
"loss": 0.0002,
|
134 |
"step": 525
|
135 |
},
|
136 |
{
|
137 |
"epoch": 1.37,
|
138 |
+
"learning_rate": 9.895555555555556e-10,
|
139 |
"loss": 0.0002,
|
140 |
"step": 550
|
141 |
},
|
142 |
{
|
143 |
"epoch": 1.43,
|
144 |
+
"learning_rate": 9.840000000000001e-10,
|
145 |
"loss": 0.0002,
|
146 |
"step": 575
|
147 |
},
|
148 |
{
|
149 |
"epoch": 1.49,
|
150 |
+
"learning_rate": 9.784444444444444e-10,
|
151 |
"loss": 0.0002,
|
152 |
"step": 600
|
153 |
},
|
154 |
{
|
155 |
"epoch": 1.55,
|
156 |
+
"learning_rate": 9.72888888888889e-10,
|
157 |
"loss": 0.0002,
|
158 |
"step": 625
|
159 |
},
|
160 |
{
|
161 |
"epoch": 1.62,
|
162 |
+
"learning_rate": 9.673333333333335e-10,
|
163 |
"loss": 0.0002,
|
164 |
"step": 650
|
165 |
},
|
166 |
{
|
167 |
"epoch": 1.68,
|
168 |
+
"learning_rate": 9.617777777777778e-10,
|
169 |
"loss": 0.0002,
|
170 |
"step": 675
|
171 |
},
|
172 |
{
|
173 |
"epoch": 1.74,
|
174 |
+
"learning_rate": 9.562222222222222e-10,
|
175 |
"loss": 0.0002,
|
176 |
"step": 700
|
177 |
},
|
178 |
{
|
179 |
"epoch": 1.8,
|
180 |
+
"learning_rate": 9.506666666666667e-10,
|
181 |
"loss": 0.0002,
|
182 |
"step": 725
|
183 |
},
|
184 |
{
|
185 |
"epoch": 1.87,
|
186 |
+
"learning_rate": 9.451111111111112e-10,
|
187 |
"loss": 0.0002,
|
188 |
"step": 750
|
189 |
},
|
190 |
{
|
191 |
"epoch": 1.93,
|
192 |
+
"learning_rate": 9.395555555555556e-10,
|
193 |
"loss": 0.0002,
|
194 |
"step": 775
|
195 |
},
|
196 |
{
|
197 |
"epoch": 1.99,
|
198 |
+
"learning_rate": 9.34e-10,
|
199 |
"loss": 0.0002,
|
200 |
"step": 800
|
201 |
},
|
202 |
{
|
203 |
"epoch": 2.05,
|
204 |
+
"learning_rate": 9.284444444444445e-10,
|
205 |
"loss": 0.0002,
|
206 |
"step": 825
|
207 |
},
|
208 |
{
|
209 |
"epoch": 2.11,
|
210 |
+
"learning_rate": 9.228888888888889e-10,
|
211 |
"loss": 0.0002,
|
212 |
"step": 850
|
213 |
},
|
214 |
{
|
215 |
"epoch": 2.18,
|
216 |
+
"learning_rate": 9.173333333333334e-10,
|
217 |
"loss": 0.0002,
|
218 |
"step": 875
|
219 |
},
|
220 |
{
|
221 |
"epoch": 2.24,
|
222 |
+
"learning_rate": 9.117777777777779e-10,
|
223 |
"loss": 0.0002,
|
224 |
"step": 900
|
225 |
},
|
226 |
{
|
227 |
"epoch": 2.3,
|
228 |
+
"learning_rate": 9.062222222222223e-10,
|
229 |
"loss": 0.0002,
|
230 |
"step": 925
|
231 |
},
|
232 |
{
|
233 |
"epoch": 2.36,
|
234 |
+
"learning_rate": 9.006666666666667e-10,
|
235 |
"loss": 0.0002,
|
236 |
"step": 950
|
237 |
},
|
238 |
{
|
239 |
"epoch": 2.43,
|
240 |
+
"learning_rate": 8.951111111111111e-10,
|
241 |
"loss": 0.0002,
|
242 |
"step": 975
|
243 |
},
|
244 |
{
|
245 |
"epoch": 2.49,
|
246 |
+
"learning_rate": 8.895555555555556e-10,
|
247 |
"loss": 0.0002,
|
248 |
"step": 1000
|
249 |
},
|
250 |
{
|
251 |
"epoch": 2.49,
|
252 |
"eval_loss": 0.50146484375,
|
253 |
+
"eval_runtime": 168.0976,
|
254 |
+
"eval_samples_per_second": 10.089,
|
255 |
+
"eval_steps_per_second": 0.631,
|
256 |
+
"eval_wer": 16.51188707280832,
|
257 |
"step": 1000
|
258 |
},
|
259 |
{
|
260 |
"epoch": 2.55,
|
261 |
+
"learning_rate": 8.844444444444446e-10,
|
262 |
"loss": 0.0002,
|
263 |
"step": 1025
|
264 |
},
|
265 |
{
|
266 |
"epoch": 2.61,
|
267 |
+
"learning_rate": 8.78888888888889e-10,
|
268 |
"loss": 0.0002,
|
269 |
"step": 1050
|
270 |
},
|
271 |
{
|
272 |
"epoch": 2.67,
|
273 |
+
"learning_rate": 8.733333333333333e-10,
|
274 |
"loss": 0.0002,
|
275 |
"step": 1075
|
276 |
},
|
277 |
{
|
278 |
"epoch": 2.74,
|
279 |
+
"learning_rate": 8.677777777777778e-10,
|
280 |
"loss": 0.0002,
|
281 |
"step": 1100
|
282 |
},
|
283 |
{
|
284 |
"epoch": 2.8,
|
285 |
+
"learning_rate": 8.622222222222223e-10,
|
286 |
"loss": 0.0002,
|
287 |
"step": 1125
|
288 |
},
|
289 |
{
|
290 |
"epoch": 2.86,
|
291 |
+
"learning_rate": 8.566666666666667e-10,
|
292 |
"loss": 0.0002,
|
293 |
"step": 1150
|
294 |
},
|
295 |
{
|
296 |
"epoch": 2.92,
|
297 |
+
"learning_rate": 8.511111111111112e-10,
|
298 |
"loss": 0.0002,
|
299 |
"step": 1175
|
300 |
},
|
301 |
{
|
302 |
"epoch": 2.99,
|
303 |
+
"learning_rate": 8.455555555555556e-10,
|
304 |
"loss": 0.0002,
|
305 |
"step": 1200
|
306 |
},
|
307 |
{
|
308 |
"epoch": 3.05,
|
309 |
+
"learning_rate": 8.4e-10,
|
310 |
"loss": 0.0002,
|
311 |
"step": 1225
|
312 |
},
|
313 |
{
|
314 |
"epoch": 3.11,
|
315 |
+
"learning_rate": 8.344444444444445e-10,
|
316 |
"loss": 0.0002,
|
317 |
"step": 1250
|
318 |
},
|
319 |
{
|
320 |
"epoch": 3.17,
|
321 |
+
"learning_rate": 8.28888888888889e-10,
|
322 |
"loss": 0.0002,
|
323 |
"step": 1275
|
324 |
},
|
325 |
{
|
326 |
"epoch": 3.23,
|
327 |
+
"learning_rate": 8.233333333333334e-10,
|
328 |
"loss": 0.0002,
|
329 |
"step": 1300
|
330 |
},
|
331 |
{
|
332 |
"epoch": 3.3,
|
333 |
+
"learning_rate": 8.177777777777779e-10,
|
334 |
"loss": 0.0002,
|
335 |
"step": 1325
|
336 |
},
|
337 |
{
|
338 |
"epoch": 3.36,
|
339 |
+
"learning_rate": 8.122222222222222e-10,
|
340 |
"loss": 0.0002,
|
341 |
"step": 1350
|
342 |
},
|
343 |
{
|
344 |
"epoch": 3.42,
|
345 |
+
"learning_rate": 8.066666666666667e-10,
|
346 |
"loss": 0.0002,
|
347 |
"step": 1375
|
348 |
},
|
349 |
{
|
350 |
"epoch": 3.48,
|
351 |
+
"learning_rate": 8.011111111111112e-10,
|
352 |
"loss": 0.0002,
|
353 |
"step": 1400
|
354 |
},
|
355 |
{
|
356 |
"epoch": 3.54,
|
357 |
+
"learning_rate": 7.955555555555556e-10,
|
358 |
"loss": 0.0002,
|
359 |
"step": 1425
|
360 |
},
|
361 |
{
|
362 |
"epoch": 3.61,
|
363 |
+
"learning_rate": 7.900000000000001e-10,
|
364 |
"loss": 0.0002,
|
365 |
"step": 1450
|
366 |
},
|
367 |
{
|
368 |
"epoch": 3.67,
|
369 |
+
"learning_rate": 7.844444444444446e-10,
|
370 |
"loss": 0.0002,
|
371 |
"step": 1475
|
372 |
},
|
373 |
{
|
374 |
"epoch": 3.73,
|
375 |
+
"learning_rate": 7.788888888888889e-10,
|
376 |
"loss": 0.0002,
|
377 |
"step": 1500
|
378 |
},
|
379 |
{
|
380 |
"epoch": 3.79,
|
381 |
+
"learning_rate": 7.733333333333334e-10,
|
382 |
"loss": 0.0002,
|
383 |
"step": 1525
|
384 |
},
|
385 |
{
|
386 |
"epoch": 3.86,
|
387 |
+
"learning_rate": 7.677777777777778e-10,
|
388 |
"loss": 0.0002,
|
389 |
"step": 1550
|
390 |
},
|
391 |
{
|
392 |
"epoch": 3.92,
|
393 |
+
"learning_rate": 7.622222222222223e-10,
|
394 |
"loss": 0.0002,
|
395 |
"step": 1575
|
396 |
},
|
397 |
{
|
398 |
"epoch": 3.98,
|
399 |
+
"learning_rate": 7.566666666666667e-10,
|
400 |
"loss": 0.0002,
|
401 |
"step": 1600
|
402 |
},
|
403 |
{
|
404 |
"epoch": 4.04,
|
405 |
+
"learning_rate": 7.511111111111111e-10,
|
406 |
"loss": 0.0002,
|
407 |
"step": 1625
|
408 |
},
|
409 |
{
|
410 |
"epoch": 4.1,
|
411 |
+
"learning_rate": 7.455555555555556e-10,
|
412 |
"loss": 0.0002,
|
413 |
"step": 1650
|
414 |
},
|
415 |
{
|
416 |
"epoch": 4.17,
|
417 |
+
"learning_rate": 7.4e-10,
|
418 |
"loss": 0.0002,
|
419 |
"step": 1675
|
420 |
},
|
421 |
{
|
422 |
"epoch": 4.23,
|
423 |
+
"learning_rate": 7.344444444444445e-10,
|
424 |
"loss": 0.0002,
|
425 |
"step": 1700
|
426 |
},
|
427 |
{
|
428 |
"epoch": 4.29,
|
429 |
+
"learning_rate": 7.28888888888889e-10,
|
430 |
"loss": 0.0002,
|
431 |
"step": 1725
|
432 |
},
|
433 |
{
|
434 |
"epoch": 4.35,
|
435 |
+
"learning_rate": 7.233333333333334e-10,
|
436 |
"loss": 0.0002,
|
437 |
"step": 1750
|
438 |
},
|
439 |
{
|
440 |
"epoch": 4.42,
|
441 |
+
"learning_rate": 7.177777777777777e-10,
|
442 |
"loss": 0.0002,
|
443 |
"step": 1775
|
444 |
},
|
445 |
{
|
446 |
"epoch": 4.48,
|
447 |
+
"learning_rate": 7.122222222222223e-10,
|
448 |
"loss": 0.0002,
|
449 |
"step": 1800
|
450 |
},
|
451 |
{
|
452 |
"epoch": 4.54,
|
453 |
+
"learning_rate": 7.066666666666667e-10,
|
454 |
"loss": 0.0002,
|
455 |
"step": 1825
|
456 |
},
|
457 |
{
|
458 |
"epoch": 4.6,
|
459 |
+
"learning_rate": 7.011111111111111e-10,
|
460 |
"loss": 0.0002,
|
461 |
"step": 1850
|
462 |
},
|
463 |
{
|
464 |
"epoch": 4.66,
|
465 |
+
"learning_rate": 6.955555555555557e-10,
|
466 |
"loss": 0.0002,
|
467 |
"step": 1875
|
468 |
},
|
469 |
{
|
470 |
"epoch": 4.73,
|
471 |
+
"learning_rate": 6.9e-10,
|
472 |
"loss": 0.0002,
|
473 |
"step": 1900
|
474 |
},
|
475 |
{
|
476 |
"epoch": 4.79,
|
477 |
+
"learning_rate": 6.844444444444444e-10,
|
478 |
"loss": 0.0002,
|
479 |
"step": 1925
|
480 |
},
|
481 |
{
|
482 |
"epoch": 4.85,
|
483 |
+
"learning_rate": 6.78888888888889e-10,
|
484 |
"loss": 0.0002,
|
485 |
"step": 1950
|
486 |
},
|
487 |
{
|
488 |
"epoch": 4.91,
|
489 |
+
"learning_rate": 6.733333333333334e-10,
|
490 |
"loss": 0.0002,
|
491 |
"step": 1975
|
492 |
},
|
493 |
{
|
494 |
"epoch": 4.98,
|
495 |
+
"learning_rate": 6.677777777777778e-10,
|
496 |
"loss": 0.0002,
|
497 |
"step": 2000
|
498 |
},
|
499 |
{
|
500 |
"epoch": 4.98,
|
501 |
"eval_loss": 0.50146484375,
|
502 |
+
"eval_runtime": 168.0899,
|
503 |
+
"eval_samples_per_second": 10.09,
|
504 |
+
"eval_steps_per_second": 0.631,
|
505 |
+
"eval_wer": 16.558320950965825,
|
506 |
"step": 2000
|
507 |
},
|
508 |
{
|
509 |
"epoch": 5.04,
|
510 |
+
"learning_rate": 6.626666666666667e-10,
|
511 |
"loss": 0.0002,
|
512 |
"step": 2025
|
513 |
},
|
514 |
{
|
515 |
"epoch": 5.1,
|
516 |
+
"learning_rate": 6.571111111111111e-10,
|
517 |
"loss": 0.0002,
|
518 |
"step": 2050
|
519 |
},
|
520 |
{
|
521 |
"epoch": 5.16,
|
522 |
+
"learning_rate": 6.515555555555556e-10,
|
523 |
"loss": 0.0002,
|
524 |
"step": 2075
|
525 |
},
|
526 |
{
|
527 |
"epoch": 5.22,
|
528 |
+
"learning_rate": 6.46e-10,
|
529 |
"loss": 0.0002,
|
530 |
"step": 2100
|
531 |
},
|
532 |
{
|
533 |
"epoch": 5.29,
|
534 |
+
"learning_rate": 6.404444444444445e-10,
|
535 |
"loss": 0.0002,
|
536 |
"step": 2125
|
537 |
},
|
538 |
{
|
539 |
"epoch": 5.35,
|
540 |
+
"learning_rate": 6.348888888888889e-10,
|
541 |
"loss": 0.0002,
|
542 |
"step": 2150
|
543 |
},
|
544 |
{
|
545 |
"epoch": 5.41,
|
546 |
+
"learning_rate": 6.293333333333333e-10,
|
547 |
"loss": 0.0002,
|
548 |
"step": 2175
|
549 |
},
|
550 |
{
|
551 |
"epoch": 5.47,
|
552 |
+
"learning_rate": 6.237777777777778e-10,
|
553 |
"loss": 0.0002,
|
554 |
"step": 2200
|
555 |
},
|
556 |
{
|
557 |
"epoch": 5.53,
|
558 |
+
"learning_rate": 6.182222222222223e-10,
|
559 |
"loss": 0.0002,
|
560 |
"step": 2225
|
561 |
},
|
562 |
{
|
563 |
"epoch": 5.6,
|
564 |
+
"learning_rate": 6.126666666666667e-10,
|
565 |
"loss": 0.0002,
|
566 |
"step": 2250
|
567 |
},
|
568 |
{
|
569 |
"epoch": 5.66,
|
570 |
+
"learning_rate": 6.071111111111112e-10,
|
571 |
"loss": 0.0002,
|
572 |
"step": 2275
|
573 |
},
|
574 |
{
|
575 |
"epoch": 5.72,
|
576 |
+
"learning_rate": 6.015555555555556e-10,
|
577 |
"loss": 0.0002,
|
578 |
"step": 2300
|
579 |
},
|
580 |
{
|
581 |
"epoch": 5.78,
|
582 |
+
"learning_rate": 5.96e-10,
|
583 |
"loss": 0.0002,
|
584 |
"step": 2325
|
585 |
},
|
586 |
{
|
587 |
"epoch": 5.85,
|
588 |
+
"learning_rate": 5.904444444444444e-10,
|
589 |
"loss": 0.0002,
|
590 |
"step": 2350
|
591 |
},
|
592 |
{
|
593 |
"epoch": 5.91,
|
594 |
+
"learning_rate": 5.84888888888889e-10,
|
595 |
"loss": 0.0002,
|
596 |
"step": 2375
|
597 |
},
|
598 |
{
|
599 |
"epoch": 5.97,
|
600 |
+
"learning_rate": 5.793333333333334e-10,
|
601 |
"loss": 0.0002,
|
602 |
"step": 2400
|
603 |
},
|
604 |
{
|
605 |
"epoch": 6.03,
|
606 |
+
"learning_rate": 5.737777777777778e-10,
|
607 |
"loss": 0.0002,
|
608 |
"step": 2425
|
609 |
},
|
610 |
{
|
611 |
"epoch": 6.09,
|
612 |
+
"learning_rate": 5.682222222222223e-10,
|
613 |
"loss": 0.0002,
|
614 |
"step": 2450
|
615 |
},
|
616 |
{
|
617 |
"epoch": 6.16,
|
618 |
+
"learning_rate": 5.626666666666667e-10,
|
619 |
"loss": 0.0002,
|
620 |
"step": 2475
|
621 |
},
|
622 |
{
|
623 |
"epoch": 6.22,
|
624 |
+
"learning_rate": 5.571111111111111e-10,
|
625 |
"loss": 0.0002,
|
626 |
"step": 2500
|
627 |
},
|
628 |
{
|
629 |
"epoch": 6.28,
|
630 |
+
"learning_rate": 5.515555555555556e-10,
|
631 |
"loss": 0.0002,
|
632 |
"step": 2525
|
633 |
},
|
634 |
{
|
635 |
"epoch": 6.34,
|
636 |
+
"learning_rate": 5.460000000000001e-10,
|
637 |
"loss": 0.0002,
|
638 |
"step": 2550
|
639 |
},
|
640 |
{
|
641 |
"epoch": 6.41,
|
642 |
+
"learning_rate": 5.404444444444444e-10,
|
643 |
"loss": 0.0002,
|
644 |
"step": 2575
|
645 |
},
|
646 |
{
|
647 |
"epoch": 6.47,
|
648 |
+
"learning_rate": 5.348888888888889e-10,
|
649 |
"loss": 0.0002,
|
650 |
"step": 2600
|
651 |
},
|
652 |
{
|
653 |
"epoch": 6.53,
|
654 |
+
"learning_rate": 5.293333333333334e-10,
|
655 |
"loss": 0.0002,
|
656 |
"step": 2625
|
657 |
},
|
658 |
{
|
659 |
"epoch": 6.59,
|
660 |
+
"learning_rate": 5.237777777777778e-10,
|
661 |
"loss": 0.0002,
|
662 |
"step": 2650
|
663 |
},
|
664 |
{
|
665 |
"epoch": 6.65,
|
666 |
+
"learning_rate": 5.182222222222223e-10,
|
667 |
"loss": 0.0002,
|
668 |
"step": 2675
|
669 |
},
|
670 |
{
|
671 |
"epoch": 6.72,
|
672 |
+
"learning_rate": 5.126666666666668e-10,
|
673 |
"loss": 0.0002,
|
674 |
"step": 2700
|
675 |
},
|
676 |
{
|
677 |
"epoch": 6.78,
|
678 |
+
"learning_rate": 5.071111111111111e-10,
|
679 |
"loss": 0.0002,
|
680 |
"step": 2725
|
681 |
},
|
682 |
{
|
683 |
"epoch": 6.84,
|
684 |
+
"learning_rate": 5.015555555555555e-10,
|
685 |
"loss": 0.0002,
|
686 |
"step": 2750
|
687 |
},
|
688 |
{
|
689 |
"epoch": 6.9,
|
690 |
+
"learning_rate": 4.96e-10,
|
691 |
"loss": 0.0002,
|
692 |
"step": 2775
|
693 |
},
|
694 |
{
|
695 |
"epoch": 6.97,
|
696 |
+
"learning_rate": 4.904444444444445e-10,
|
697 |
"loss": 0.0002,
|
698 |
"step": 2800
|
699 |
},
|
700 |
{
|
701 |
"epoch": 7.03,
|
702 |
+
"learning_rate": 4.848888888888889e-10,
|
703 |
"loss": 0.0002,
|
704 |
"step": 2825
|
705 |
},
|
706 |
{
|
707 |
"epoch": 7.09,
|
708 |
+
"learning_rate": 4.793333333333333e-10,
|
709 |
"loss": 0.0002,
|
710 |
"step": 2850
|
711 |
},
|
712 |
{
|
713 |
"epoch": 7.15,
|
714 |
+
"learning_rate": 4.737777777777779e-10,
|
715 |
"loss": 0.0002,
|
716 |
"step": 2875
|
717 |
},
|
718 |
{
|
719 |
"epoch": 7.21,
|
720 |
+
"learning_rate": 4.682222222222222e-10,
|
721 |
"loss": 0.0002,
|
722 |
"step": 2900
|
723 |
},
|
724 |
{
|
725 |
"epoch": 7.28,
|
726 |
+
"learning_rate": 4.626666666666667e-10,
|
727 |
"loss": 0.0002,
|
728 |
"step": 2925
|
729 |
},
|
730 |
{
|
731 |
"epoch": 7.34,
|
732 |
+
"learning_rate": 4.5711111111111115e-10,
|
733 |
"loss": 0.0002,
|
734 |
"step": 2950
|
735 |
},
|
736 |
{
|
737 |
"epoch": 7.4,
|
738 |
+
"learning_rate": 4.515555555555556e-10,
|
739 |
"loss": 0.0002,
|
740 |
"step": 2975
|
741 |
},
|
742 |
{
|
743 |
"epoch": 7.46,
|
744 |
+
"learning_rate": 4.46e-10,
|
745 |
"loss": 0.0002,
|
746 |
"step": 3000
|
747 |
},
|
748 |
{
|
749 |
"epoch": 7.46,
|
750 |
"eval_loss": 0.50146484375,
|
751 |
+
"eval_runtime": 167.3892,
|
752 |
+
"eval_samples_per_second": 10.132,
|
753 |
"eval_steps_per_second": 0.633,
|
754 |
+
"eval_wer": 16.465453194650816,
|
755 |
"step": 3000
|
756 |
},
|
757 |
{
|
758 |
"epoch": 7.52,
|
759 |
+
"learning_rate": 4.408888888888889e-10,
|
760 |
"loss": 0.0002,
|
761 |
"step": 3025
|
762 |
},
|
763 |
{
|
764 |
"epoch": 7.59,
|
765 |
+
"learning_rate": 4.353333333333334e-10,
|
766 |
"loss": 0.0001,
|
767 |
"step": 3050
|
768 |
},
|
769 |
{
|
770 |
"epoch": 7.65,
|
771 |
+
"learning_rate": 4.297777777777778e-10,
|
772 |
"loss": 0.0002,
|
773 |
"step": 3075
|
774 |
},
|
775 |
{
|
776 |
"epoch": 7.71,
|
777 |
+
"learning_rate": 4.2422222222222224e-10,
|
778 |
"loss": 0.0002,
|
779 |
"step": 3100
|
780 |
},
|
781 |
{
|
782 |
"epoch": 7.77,
|
783 |
+
"learning_rate": 4.186666666666667e-10,
|
784 |
"loss": 0.0002,
|
785 |
"step": 3125
|
786 |
},
|
787 |
{
|
788 |
"epoch": 7.84,
|
789 |
+
"learning_rate": 4.131111111111111e-10,
|
790 |
"loss": 0.0002,
|
791 |
"step": 3150
|
792 |
},
|
793 |
{
|
794 |
"epoch": 7.9,
|
795 |
+
"learning_rate": 4.075555555555556e-10,
|
796 |
"loss": 0.0002,
|
797 |
"step": 3175
|
798 |
},
|
799 |
{
|
800 |
"epoch": 7.96,
|
801 |
+
"learning_rate": 4.0200000000000006e-10,
|
802 |
"loss": 0.0002,
|
803 |
"step": 3200
|
804 |
},
|
805 |
{
|
806 |
"epoch": 8.02,
|
807 |
+
"learning_rate": 3.9644444444444444e-10,
|
808 |
"loss": 0.0002,
|
809 |
"step": 3225
|
810 |
},
|
811 |
{
|
812 |
"epoch": 8.08,
|
813 |
+
"learning_rate": 3.908888888888889e-10,
|
814 |
"loss": 0.0002,
|
815 |
"step": 3250
|
816 |
},
|
817 |
{
|
818 |
"epoch": 8.15,
|
819 |
+
"learning_rate": 3.853333333333334e-10,
|
820 |
"loss": 0.0002,
|
821 |
"step": 3275
|
822 |
},
|
823 |
{
|
824 |
"epoch": 8.21,
|
825 |
+
"learning_rate": 3.797777777777778e-10,
|
826 |
"loss": 0.0002,
|
827 |
"step": 3300
|
828 |
},
|
829 |
{
|
830 |
"epoch": 8.27,
|
831 |
+
"learning_rate": 3.7422222222222226e-10,
|
832 |
"loss": 0.0002,
|
833 |
"step": 3325
|
834 |
},
|
835 |
{
|
836 |
"epoch": 8.33,
|
837 |
+
"learning_rate": 3.686666666666667e-10,
|
838 |
"loss": 0.0002,
|
839 |
"step": 3350
|
840 |
},
|
841 |
{
|
842 |
"epoch": 8.4,
|
843 |
+
"learning_rate": 3.631111111111111e-10,
|
844 |
"loss": 0.0002,
|
845 |
"step": 3375
|
846 |
},
|
847 |
{
|
848 |
"epoch": 8.46,
|
849 |
+
"learning_rate": 3.575555555555556e-10,
|
850 |
"loss": 0.0002,
|
851 |
"step": 3400
|
852 |
},
|
853 |
{
|
854 |
"epoch": 8.52,
|
855 |
+
"learning_rate": 3.52e-10,
|
856 |
"loss": 0.0002,
|
857 |
"step": 3425
|
858 |
},
|
859 |
{
|
860 |
"epoch": 8.58,
|
861 |
+
"learning_rate": 3.4644444444444445e-10,
|
862 |
"loss": 0.0002,
|
863 |
"step": 3450
|
864 |
},
|
865 |
{
|
866 |
"epoch": 8.64,
|
867 |
+
"learning_rate": 3.4088888888888894e-10,
|
868 |
"loss": 0.0002,
|
869 |
"step": 3475
|
870 |
},
|
871 |
{
|
872 |
"epoch": 8.71,
|
873 |
+
"learning_rate": 3.353333333333333e-10,
|
874 |
"loss": 0.0002,
|
875 |
"step": 3500
|
876 |
},
|
877 |
{
|
878 |
"epoch": 8.77,
|
879 |
+
"learning_rate": 3.297777777777778e-10,
|
880 |
+
"loss": 0.0002,
|
881 |
"step": 3525
|
882 |
},
|
883 |
{
|
884 |
"epoch": 8.83,
|
885 |
+
"learning_rate": 3.242222222222223e-10,
|
886 |
"loss": 0.0002,
|
887 |
"step": 3550
|
888 |
},
|
889 |
{
|
890 |
"epoch": 8.89,
|
891 |
+
"learning_rate": 3.1866666666666665e-10,
|
892 |
"loss": 0.0002,
|
893 |
"step": 3575
|
894 |
},
|
895 |
{
|
896 |
"epoch": 8.96,
|
897 |
+
"learning_rate": 3.1311111111111113e-10,
|
898 |
"loss": 0.0002,
|
899 |
"step": 3600
|
900 |
},
|
901 |
{
|
902 |
"epoch": 9.02,
|
903 |
+
"learning_rate": 3.075555555555556e-10,
|
904 |
"loss": 0.0002,
|
905 |
"step": 3625
|
906 |
},
|
907 |
{
|
908 |
"epoch": 9.08,
|
909 |
+
"learning_rate": 3.02e-10,
|
910 |
"loss": 0.0002,
|
911 |
"step": 3650
|
912 |
},
|
913 |
{
|
914 |
"epoch": 9.14,
|
915 |
+
"learning_rate": 2.9644444444444447e-10,
|
916 |
"loss": 0.0002,
|
917 |
"step": 3675
|
918 |
},
|
919 |
{
|
920 |
"epoch": 9.2,
|
921 |
+
"learning_rate": 2.908888888888889e-10,
|
922 |
"loss": 0.0002,
|
923 |
"step": 3700
|
924 |
},
|
925 |
{
|
926 |
"epoch": 9.27,
|
927 |
+
"learning_rate": 2.8533333333333333e-10,
|
928 |
"loss": 0.0002,
|
929 |
"step": 3725
|
930 |
},
|
931 |
{
|
932 |
"epoch": 9.33,
|
933 |
+
"learning_rate": 2.797777777777778e-10,
|
934 |
"loss": 0.0002,
|
935 |
"step": 3750
|
936 |
},
|
937 |
{
|
938 |
"epoch": 9.39,
|
939 |
+
"learning_rate": 2.7422222222222224e-10,
|
940 |
"loss": 0.0002,
|
941 |
"step": 3775
|
942 |
},
|
943 |
{
|
944 |
"epoch": 9.45,
|
945 |
+
"learning_rate": 2.6866666666666667e-10,
|
946 |
"loss": 0.0002,
|
947 |
"step": 3800
|
948 |
},
|
949 |
{
|
950 |
"epoch": 9.51,
|
951 |
+
"learning_rate": 2.6311111111111115e-10,
|
952 |
"loss": 0.0002,
|
953 |
"step": 3825
|
954 |
},
|
955 |
{
|
956 |
"epoch": 9.58,
|
957 |
+
"learning_rate": 2.5755555555555553e-10,
|
958 |
"loss": 0.0002,
|
959 |
"step": 3850
|
960 |
},
|
961 |
{
|
962 |
"epoch": 9.64,
|
963 |
+
"learning_rate": 2.52e-10,
|
964 |
"loss": 0.0002,
|
965 |
"step": 3875
|
966 |
},
|
967 |
{
|
968 |
"epoch": 9.7,
|
969 |
+
"learning_rate": 2.4644444444444444e-10,
|
970 |
"loss": 0.0002,
|
971 |
"step": 3900
|
972 |
},
|
973 |
{
|
974 |
"epoch": 9.76,
|
975 |
+
"learning_rate": 2.408888888888889e-10,
|
976 |
"loss": 0.0002,
|
977 |
"step": 3925
|
978 |
},
|
979 |
{
|
980 |
"epoch": 9.83,
|
981 |
+
"learning_rate": 2.3533333333333335e-10,
|
982 |
"loss": 0.0002,
|
983 |
"step": 3950
|
984 |
},
|
985 |
{
|
986 |
"epoch": 9.89,
|
987 |
+
"learning_rate": 2.2977777777777778e-10,
|
988 |
"loss": 0.0002,
|
989 |
"step": 3975
|
990 |
},
|
991 |
{
|
992 |
"epoch": 9.95,
|
993 |
+
"learning_rate": 2.2422222222222223e-10,
|
994 |
"loss": 0.0002,
|
995 |
"step": 4000
|
996 |
},
|
997 |
{
|
998 |
"epoch": 9.95,
|
999 |
"eval_loss": 0.50146484375,
|
1000 |
+
"eval_runtime": 167.9097,
|
1001 |
+
"eval_samples_per_second": 10.101,
|
1002 |
+
"eval_steps_per_second": 0.631,
|
1003 |
+
"eval_wer": 16.51188707280832,
|
1004 |
"step": 4000
|
1005 |
},
|
1006 |
{
|
1007 |
"epoch": 10.01,
|
1008 |
+
"learning_rate": 2.1911111111111112e-10,
|
1009 |
"loss": 0.0002,
|
1010 |
"step": 4025
|
1011 |
},
|
1012 |
{
|
1013 |
"epoch": 10.07,
|
1014 |
+
"learning_rate": 2.1355555555555558e-10,
|
1015 |
"loss": 0.0002,
|
1016 |
"step": 4050
|
1017 |
},
|
1018 |
{
|
1019 |
"epoch": 10.14,
|
1020 |
+
"learning_rate": 2.08e-10,
|
1021 |
"loss": 0.0002,
|
1022 |
"step": 4075
|
1023 |
},
|
1024 |
{
|
1025 |
"epoch": 10.2,
|
1026 |
+
"learning_rate": 2.0244444444444446e-10,
|
1027 |
"loss": 0.0002,
|
1028 |
"step": 4100
|
1029 |
},
|
1030 |
{
|
1031 |
"epoch": 10.26,
|
1032 |
+
"learning_rate": 1.9688888888888892e-10,
|
1033 |
"loss": 0.0002,
|
1034 |
"step": 4125
|
1035 |
},
|
1036 |
{
|
1037 |
"epoch": 10.32,
|
1038 |
+
"learning_rate": 1.9133333333333334e-10,
|
1039 |
"loss": 0.0002,
|
1040 |
"step": 4150
|
1041 |
},
|
1042 |
{
|
1043 |
"epoch": 10.39,
|
1044 |
+
"learning_rate": 1.8577777777777777e-10,
|
1045 |
"loss": 0.0002,
|
1046 |
"step": 4175
|
1047 |
},
|
1048 |
{
|
1049 |
"epoch": 10.45,
|
1050 |
+
"learning_rate": 1.8022222222222223e-10,
|
1051 |
"loss": 0.0002,
|
1052 |
"step": 4200
|
1053 |
},
|
1054 |
{
|
1055 |
"epoch": 10.51,
|
1056 |
+
"learning_rate": 1.7466666666666668e-10,
|
1057 |
"loss": 0.0002,
|
1058 |
"step": 4225
|
1059 |
},
|
1060 |
{
|
1061 |
"epoch": 10.57,
|
1062 |
+
"learning_rate": 1.6911111111111111e-10,
|
1063 |
"loss": 0.0002,
|
1064 |
"step": 4250
|
1065 |
},
|
1066 |
{
|
1067 |
"epoch": 10.63,
|
1068 |
+
"learning_rate": 1.6355555555555557e-10,
|
1069 |
"loss": 0.0002,
|
1070 |
"step": 4275
|
1071 |
},
|
1072 |
{
|
1073 |
"epoch": 10.7,
|
1074 |
+
"learning_rate": 1.5800000000000002e-10,
|
1075 |
"loss": 0.0002,
|
1076 |
"step": 4300
|
1077 |
},
|
1078 |
{
|
1079 |
"epoch": 10.76,
|
1080 |
+
"learning_rate": 1.5244444444444445e-10,
|
1081 |
"loss": 0.0002,
|
1082 |
"step": 4325
|
1083 |
},
|
1084 |
{
|
1085 |
"epoch": 10.82,
|
1086 |
+
"learning_rate": 1.468888888888889e-10,
|
1087 |
"loss": 0.0002,
|
1088 |
"step": 4350
|
1089 |
},
|
1090 |
{
|
1091 |
"epoch": 10.88,
|
1092 |
+
"learning_rate": 1.4133333333333334e-10,
|
1093 |
"loss": 0.0002,
|
1094 |
"step": 4375
|
1095 |
},
|
1096 |
{
|
1097 |
"epoch": 10.95,
|
1098 |
+
"learning_rate": 1.357777777777778e-10,
|
1099 |
"loss": 0.0002,
|
1100 |
"step": 4400
|
1101 |
},
|
1102 |
{
|
1103 |
"epoch": 11.01,
|
1104 |
+
"learning_rate": 1.3022222222222222e-10,
|
1105 |
"loss": 0.0002,
|
1106 |
"step": 4425
|
1107 |
},
|
1108 |
{
|
1109 |
"epoch": 11.07,
|
1110 |
+
"learning_rate": 1.2466666666666668e-10,
|
1111 |
"loss": 0.0002,
|
1112 |
"step": 4450
|
1113 |
},
|
1114 |
{
|
1115 |
"epoch": 11.13,
|
1116 |
+
"learning_rate": 1.1911111111111113e-10,
|
1117 |
"loss": 0.0002,
|
1118 |
"step": 4475
|
1119 |
},
|
1120 |
{
|
1121 |
"epoch": 11.19,
|
1122 |
+
"learning_rate": 1.1355555555555557e-10,
|
1123 |
"loss": 0.0002,
|
1124 |
"step": 4500
|
1125 |
},
|
1126 |
{
|
1127 |
"epoch": 11.26,
|
1128 |
+
"learning_rate": 1.08e-10,
|
1129 |
"loss": 0.0002,
|
1130 |
"step": 4525
|
1131 |
},
|
1132 |
{
|
1133 |
"epoch": 11.32,
|
1134 |
+
"learning_rate": 1.0244444444444446e-10,
|
1135 |
"loss": 0.0002,
|
1136 |
"step": 4550
|
1137 |
},
|
1138 |
{
|
1139 |
"epoch": 11.38,
|
1140 |
+
"learning_rate": 9.688888888888889e-11,
|
1141 |
"loss": 0.0002,
|
1142 |
"step": 4575
|
1143 |
},
|
1144 |
{
|
1145 |
"epoch": 11.44,
|
1146 |
+
"learning_rate": 9.133333333333334e-11,
|
1147 |
"loss": 0.0002,
|
1148 |
"step": 4600
|
1149 |
},
|
1150 |
{
|
1151 |
"epoch": 11.5,
|
1152 |
+
"learning_rate": 8.577777777777778e-11,
|
1153 |
"loss": 0.0002,
|
1154 |
"step": 4625
|
1155 |
},
|
1156 |
{
|
1157 |
"epoch": 11.57,
|
1158 |
+
"learning_rate": 8.022222222222223e-11,
|
1159 |
"loss": 0.0002,
|
1160 |
"step": 4650
|
1161 |
},
|
1162 |
{
|
1163 |
"epoch": 11.63,
|
1164 |
+
"learning_rate": 7.466666666666668e-11,
|
1165 |
"loss": 0.0002,
|
1166 |
"step": 4675
|
1167 |
},
|
1168 |
{
|
1169 |
"epoch": 11.69,
|
1170 |
+
"learning_rate": 6.911111111111111e-11,
|
1171 |
"loss": 0.0002,
|
1172 |
"step": 4700
|
1173 |
},
|
1174 |
{
|
1175 |
"epoch": 11.75,
|
1176 |
+
"learning_rate": 6.355555555555556e-11,
|
1177 |
"loss": 0.0002,
|
1178 |
"step": 4725
|
1179 |
},
|
1180 |
{
|
1181 |
"epoch": 11.82,
|
1182 |
+
"learning_rate": 5.8000000000000007e-11,
|
1183 |
"loss": 0.0002,
|
1184 |
"step": 4750
|
1185 |
},
|
1186 |
{
|
1187 |
"epoch": 11.88,
|
1188 |
+
"learning_rate": 5.244444444444445e-11,
|
1189 |
"loss": 0.0002,
|
1190 |
"step": 4775
|
1191 |
},
|
1192 |
{
|
1193 |
"epoch": 11.94,
|
1194 |
+
"learning_rate": 4.688888888888889e-11,
|
1195 |
"loss": 0.0002,
|
1196 |
"step": 4800
|
1197 |
},
|
1198 |
{
|
1199 |
"epoch": 12.0,
|
1200 |
+
"learning_rate": 4.133333333333333e-11,
|
1201 |
"loss": 0.0002,
|
1202 |
"step": 4825
|
1203 |
},
|
1204 |
{
|
1205 |
"epoch": 12.06,
|
1206 |
+
"learning_rate": 3.577777777777778e-11,
|
1207 |
"loss": 0.0002,
|
1208 |
"step": 4850
|
1209 |
},
|
1210 |
{
|
1211 |
"epoch": 12.13,
|
1212 |
+
"learning_rate": 3.0222222222222223e-11,
|
1213 |
"loss": 0.0002,
|
1214 |
"step": 4875
|
1215 |
},
|
1216 |
{
|
1217 |
"epoch": 12.19,
|
1218 |
+
"learning_rate": 2.466666666666667e-11,
|
1219 |
"loss": 0.0002,
|
1220 |
"step": 4900
|
1221 |
},
|
1222 |
{
|
1223 |
"epoch": 12.25,
|
1224 |
+
"learning_rate": 1.911111111111111e-11,
|
1225 |
"loss": 0.0002,
|
1226 |
"step": 4925
|
1227 |
},
|
1228 |
{
|
1229 |
"epoch": 12.31,
|
1230 |
+
"learning_rate": 1.3555555555555556e-11,
|
1231 |
"loss": 0.0002,
|
1232 |
"step": 4950
|
1233 |
},
|
1234 |
{
|
1235 |
"epoch": 12.38,
|
1236 |
+
"learning_rate": 8.000000000000001e-12,
|
1237 |
"loss": 0.0002,
|
1238 |
"step": 4975
|
1239 |
},
|
1240 |
{
|
1241 |
"epoch": 12.44,
|
1242 |
+
"learning_rate": 2.4444444444444444e-12,
|
1243 |
"loss": 0.0002,
|
1244 |
"step": 5000
|
1245 |
},
|
1246 |
{
|
1247 |
"epoch": 12.44,
|
1248 |
"eval_loss": 0.50146484375,
|
1249 |
+
"eval_runtime": 168.0048,
|
1250 |
+
"eval_samples_per_second": 10.095,
|
1251 |
+
"eval_steps_per_second": 0.631,
|
1252 |
"eval_wer": 16.558320950965825,
|
1253 |
"step": 5000
|
1254 |
},
|
1255 |
{
|
1256 |
+
"epoch": 12.44,
|
1257 |
+
"step": 5000,
|
1258 |
+
"total_flos": 4.615288630123325e+19,
|
1259 |
+
"train_loss": 0.00017441691160202026,
|
1260 |
+
"train_runtime": 9920.3408,
|
1261 |
+
"train_samples_per_second": 16.128,
|
1262 |
+
"train_steps_per_second": 0.504
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1263 |
}
|
1264 |
],
|
1265 |
+
"max_steps": 5000,
|
1266 |
+
"num_train_epochs": 13,
|
1267 |
+
"total_flos": 4.615288630123325e+19,
|
1268 |
"trial_name": null,
|
1269 |
"trial_params": null
|
1270 |
}
|