End of training
Browse files- all_results.json +11 -11
- eval_results.json +6 -6
- train_results.json +6 -6
- trainer_state.json +1309 -403
all_results.json
CHANGED
@@ -1,16 +1,16 @@
|
|
1 |
{
|
2 |
"avg_token_length": 41.888755020080325,
|
3 |
-
"epoch":
|
4 |
-
"eval_accuracy": 0.
|
5 |
-
"eval_loss": 0.
|
6 |
-
"eval_runtime": 3.
|
7 |
"eval_samples": 2490,
|
8 |
-
"eval_samples_per_second":
|
9 |
-
"eval_steps_per_second":
|
10 |
-
"total_flos":
|
11 |
-
"train_loss": 0.
|
12 |
-
"train_runtime":
|
13 |
"train_samples": 392702,
|
14 |
-
"train_samples_per_second":
|
15 |
-
"train_steps_per_second":
|
16 |
}
|
|
|
1 |
{
|
2 |
"avg_token_length": 41.888755020080325,
|
3 |
+
"epoch": 10.0,
|
4 |
+
"eval_accuracy": 0.8357429504394531,
|
5 |
+
"eval_loss": 0.4303068220615387,
|
6 |
+
"eval_runtime": 3.5423,
|
7 |
"eval_samples": 2490,
|
8 |
+
"eval_samples_per_second": 702.935,
|
9 |
+
"eval_steps_per_second": 88.079,
|
10 |
+
"total_flos": 2.6187994846190592e+17,
|
11 |
+
"train_loss": 0.4432659422714054,
|
12 |
+
"train_runtime": 8098.9158,
|
13 |
"train_samples": 392702,
|
14 |
+
"train_samples_per_second": 484.882,
|
15 |
+
"train_steps_per_second": 15.153
|
16 |
}
|
eval_results.json
CHANGED
@@ -1,10 +1,10 @@
|
|
1 |
{
|
2 |
"avg_token_length": 41.888755020080325,
|
3 |
-
"epoch":
|
4 |
-
"eval_accuracy": 0.
|
5 |
-
"eval_loss": 0.
|
6 |
-
"eval_runtime": 3.
|
7 |
"eval_samples": 2490,
|
8 |
-
"eval_samples_per_second":
|
9 |
-
"eval_steps_per_second":
|
10 |
}
|
|
|
1 |
{
|
2 |
"avg_token_length": 41.888755020080325,
|
3 |
+
"epoch": 10.0,
|
4 |
+
"eval_accuracy": 0.8357429504394531,
|
5 |
+
"eval_loss": 0.4303068220615387,
|
6 |
+
"eval_runtime": 3.5423,
|
7 |
"eval_samples": 2490,
|
8 |
+
"eval_samples_per_second": 702.935,
|
9 |
+
"eval_steps_per_second": 88.079
|
10 |
}
|
train_results.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
-
"epoch":
|
3 |
-
"total_flos":
|
4 |
-
"train_loss": 0.
|
5 |
-
"train_runtime":
|
6 |
"train_samples": 392702,
|
7 |
-
"train_samples_per_second":
|
8 |
-
"train_steps_per_second":
|
9 |
}
|
|
|
1 |
{
|
2 |
+
"epoch": 10.0,
|
3 |
+
"total_flos": 2.6187994846190592e+17,
|
4 |
+
"train_loss": 0.4432659422714054,
|
5 |
+
"train_runtime": 8098.9158,
|
6 |
"train_samples": 392702,
|
7 |
+
"train_samples_per_second": 484.882,
|
8 |
+
"train_steps_per_second": 15.153
|
9 |
}
|
trainer_state.json
CHANGED
@@ -1,926 +1,1832 @@
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
-
"epoch":
|
5 |
"eval_steps": 500,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
10 |
"log_history": [
|
11 |
{
|
12 |
"epoch": 0.04074315514993481,
|
13 |
-
"grad_norm": 3.
|
14 |
-
"learning_rate":
|
15 |
-
"loss":
|
16 |
"step": 500
|
17 |
},
|
18 |
{
|
19 |
"epoch": 0.08148631029986962,
|
20 |
-
"grad_norm":
|
21 |
-
"learning_rate":
|
22 |
-
"loss": 0.
|
23 |
"step": 1000
|
24 |
},
|
25 |
{
|
26 |
"epoch": 0.12222946544980444,
|
27 |
-
"grad_norm":
|
28 |
-
"learning_rate":
|
29 |
-
"loss": 0.
|
30 |
"step": 1500
|
31 |
},
|
32 |
{
|
33 |
"epoch": 0.16297262059973924,
|
34 |
-
"grad_norm": 4.
|
35 |
-
"learning_rate":
|
36 |
-
"loss": 0.
|
37 |
"step": 2000
|
38 |
},
|
39 |
{
|
40 |
"epoch": 0.20371577574967406,
|
41 |
-
"grad_norm":
|
42 |
-
"learning_rate":
|
43 |
-
"loss": 0.
|
44 |
"step": 2500
|
45 |
},
|
46 |
{
|
47 |
"epoch": 0.24445893089960888,
|
48 |
-
"grad_norm": 3.
|
49 |
-
"learning_rate":
|
50 |
-
"loss": 0.
|
51 |
"step": 3000
|
52 |
},
|
53 |
{
|
54 |
"epoch": 0.28520208604954367,
|
55 |
-
"grad_norm":
|
56 |
-
"learning_rate":
|
57 |
-
"loss": 0.
|
58 |
"step": 3500
|
59 |
},
|
60 |
{
|
61 |
"epoch": 0.3259452411994785,
|
62 |
-
"grad_norm":
|
63 |
-
"learning_rate":
|
64 |
-
"loss": 0.
|
65 |
"step": 4000
|
66 |
},
|
67 |
{
|
68 |
"epoch": 0.3666883963494133,
|
69 |
-
"grad_norm":
|
70 |
-
"learning_rate":
|
71 |
-
"loss": 0.
|
72 |
"step": 4500
|
73 |
},
|
74 |
{
|
75 |
"epoch": 0.4074315514993481,
|
76 |
-
"grad_norm":
|
77 |
-
"learning_rate":
|
78 |
-
"loss": 0.
|
79 |
"step": 5000
|
80 |
},
|
81 |
{
|
82 |
"epoch": 0.44817470664928294,
|
83 |
-
"grad_norm":
|
84 |
-
"learning_rate":
|
85 |
-
"loss": 0.
|
86 |
"step": 5500
|
87 |
},
|
88 |
{
|
89 |
"epoch": 0.48891786179921776,
|
90 |
-
"grad_norm":
|
91 |
-
"learning_rate":
|
92 |
-
"loss": 0.
|
93 |
"step": 6000
|
94 |
},
|
95 |
{
|
96 |
"epoch": 0.5296610169491526,
|
97 |
-
"grad_norm": 3.
|
98 |
-
"learning_rate":
|
99 |
-
"loss": 0.
|
100 |
"step": 6500
|
101 |
},
|
102 |
{
|
103 |
"epoch": 0.5704041720990873,
|
104 |
-
"grad_norm":
|
105 |
-
"learning_rate":
|
106 |
-
"loss": 0.
|
107 |
"step": 7000
|
108 |
},
|
109 |
{
|
110 |
"epoch": 0.6111473272490222,
|
111 |
-
"grad_norm": 3.
|
112 |
-
"learning_rate":
|
113 |
-
"loss": 0.
|
114 |
"step": 7500
|
115 |
},
|
116 |
{
|
117 |
"epoch": 0.651890482398957,
|
118 |
-
"grad_norm":
|
119 |
-
"learning_rate":
|
120 |
-
"loss": 0.
|
121 |
"step": 8000
|
122 |
},
|
123 |
{
|
124 |
"epoch": 0.6926336375488917,
|
125 |
-
"grad_norm":
|
126 |
-
"learning_rate":
|
127 |
-
"loss": 0.
|
128 |
"step": 8500
|
129 |
},
|
130 |
{
|
131 |
"epoch": 0.7333767926988266,
|
132 |
-
"grad_norm":
|
133 |
-
"learning_rate":
|
134 |
-
"loss": 0.
|
135 |
"step": 9000
|
136 |
},
|
137 |
{
|
138 |
"epoch": 0.7741199478487614,
|
139 |
-
"grad_norm":
|
140 |
-
"learning_rate":
|
141 |
-
"loss": 0.
|
142 |
"step": 9500
|
143 |
},
|
144 |
{
|
145 |
"epoch": 0.8148631029986962,
|
146 |
-
"grad_norm": 4.
|
147 |
-
"learning_rate":
|
148 |
-
"loss": 0.
|
149 |
"step": 10000
|
150 |
},
|
151 |
{
|
152 |
"epoch": 0.855606258148631,
|
153 |
-
"grad_norm":
|
154 |
-
"learning_rate":
|
155 |
-
"loss": 0.
|
156 |
"step": 10500
|
157 |
},
|
158 |
{
|
159 |
"epoch": 0.8963494132985659,
|
160 |
-
"grad_norm": 3.
|
161 |
-
"learning_rate":
|
162 |
-
"loss": 0.
|
163 |
"step": 11000
|
164 |
},
|
165 |
{
|
166 |
"epoch": 0.9370925684485006,
|
167 |
-
"grad_norm": 5.
|
168 |
-
"learning_rate":
|
169 |
-
"loss": 0.
|
170 |
"step": 11500
|
171 |
},
|
172 |
{
|
173 |
"epoch": 0.9778357235984355,
|
174 |
-
"grad_norm":
|
175 |
-
"learning_rate":
|
176 |
-
"loss": 0.
|
177 |
"step": 12000
|
178 |
},
|
179 |
{
|
180 |
"epoch": 1.0,
|
181 |
-
"eval_accuracy": 0.
|
182 |
-
"eval_loss": 0.
|
183 |
-
"eval_runtime": 3.
|
184 |
-
"eval_samples_per_second":
|
185 |
-
"eval_steps_per_second":
|
186 |
"step": 12272
|
187 |
},
|
188 |
{
|
189 |
"epoch": 1.0185788787483703,
|
190 |
-
"grad_norm":
|
191 |
-
"learning_rate":
|
192 |
-
"loss": 0.
|
193 |
"step": 12500
|
194 |
},
|
195 |
{
|
196 |
"epoch": 1.0593220338983051,
|
197 |
-
"grad_norm":
|
198 |
-
"learning_rate":
|
199 |
-
"loss": 0.
|
200 |
"step": 13000
|
201 |
},
|
202 |
{
|
203 |
"epoch": 1.1000651890482398,
|
204 |
-
"grad_norm":
|
205 |
-
"learning_rate":
|
206 |
-
"loss": 0.
|
207 |
"step": 13500
|
208 |
},
|
209 |
{
|
210 |
"epoch": 1.1408083441981747,
|
211 |
-
"grad_norm":
|
212 |
-
"learning_rate":
|
213 |
-
"loss": 0.
|
214 |
"step": 14000
|
215 |
},
|
216 |
{
|
217 |
"epoch": 1.1815514993481095,
|
218 |
-
"grad_norm": 3.
|
219 |
-
"learning_rate":
|
220 |
-
"loss": 0.
|
221 |
"step": 14500
|
222 |
},
|
223 |
{
|
224 |
"epoch": 1.2222946544980444,
|
225 |
-
"grad_norm":
|
226 |
-
"learning_rate":
|
227 |
-
"loss": 0.
|
228 |
"step": 15000
|
229 |
},
|
230 |
{
|
231 |
"epoch": 1.263037809647979,
|
232 |
-
"grad_norm":
|
233 |
-
"learning_rate":
|
234 |
-
"loss": 0.
|
235 |
"step": 15500
|
236 |
},
|
237 |
{
|
238 |
"epoch": 1.303780964797914,
|
239 |
-
"grad_norm":
|
240 |
-
"learning_rate":
|
241 |
-
"loss": 0.
|
242 |
"step": 16000
|
243 |
},
|
244 |
{
|
245 |
"epoch": 1.3445241199478488,
|
246 |
-
"grad_norm": 3.
|
247 |
-
"learning_rate":
|
248 |
-
"loss": 0.
|
249 |
"step": 16500
|
250 |
},
|
251 |
{
|
252 |
"epoch": 1.3852672750977835,
|
253 |
-
"grad_norm":
|
254 |
-
"learning_rate":
|
255 |
-
"loss": 0.
|
256 |
"step": 17000
|
257 |
},
|
258 |
{
|
259 |
"epoch": 1.4260104302477183,
|
260 |
-
"grad_norm":
|
261 |
-
"learning_rate":
|
262 |
-
"loss": 0.
|
263 |
"step": 17500
|
264 |
},
|
265 |
{
|
266 |
"epoch": 1.4667535853976532,
|
267 |
-
"grad_norm": 3.
|
268 |
-
"learning_rate":
|
269 |
-
"loss": 0.
|
270 |
"step": 18000
|
271 |
},
|
272 |
{
|
273 |
"epoch": 1.5074967405475879,
|
274 |
-
"grad_norm": 3.
|
275 |
-
"learning_rate":
|
276 |
-
"loss": 0.
|
277 |
"step": 18500
|
278 |
},
|
279 |
{
|
280 |
"epoch": 1.548239895697523,
|
281 |
-
"grad_norm":
|
282 |
-
"learning_rate":
|
283 |
-
"loss": 0.
|
284 |
"step": 19000
|
285 |
},
|
286 |
{
|
287 |
"epoch": 1.5889830508474576,
|
288 |
-
"grad_norm":
|
289 |
-
"learning_rate":
|
290 |
-
"loss": 0.
|
291 |
"step": 19500
|
292 |
},
|
293 |
{
|
294 |
"epoch": 1.6297262059973925,
|
295 |
-
"grad_norm":
|
296 |
-
"learning_rate":
|
297 |
-
"loss": 0.
|
298 |
"step": 20000
|
299 |
},
|
300 |
{
|
301 |
"epoch": 1.6704693611473274,
|
302 |
-
"grad_norm":
|
303 |
-
"learning_rate":
|
304 |
-
"loss": 0.
|
305 |
"step": 20500
|
306 |
},
|
307 |
{
|
308 |
"epoch": 1.711212516297262,
|
309 |
-
"grad_norm":
|
310 |
-
"learning_rate":
|
311 |
-
"loss": 0.
|
312 |
"step": 21000
|
313 |
},
|
314 |
{
|
315 |
"epoch": 1.7519556714471969,
|
316 |
-
"grad_norm": 6.
|
317 |
-
"learning_rate":
|
318 |
-
"loss": 0.
|
319 |
"step": 21500
|
320 |
},
|
321 |
{
|
322 |
"epoch": 1.7926988265971318,
|
323 |
-
"grad_norm": 3.
|
324 |
-
"learning_rate":
|
325 |
-
"loss": 0.
|
326 |
"step": 22000
|
327 |
},
|
328 |
{
|
329 |
"epoch": 1.8334419817470664,
|
330 |
-
"grad_norm":
|
331 |
-
"learning_rate":
|
332 |
-
"loss": 0.
|
333 |
"step": 22500
|
334 |
},
|
335 |
{
|
336 |
"epoch": 1.8741851368970013,
|
337 |
-
"grad_norm": 4.
|
338 |
-
"learning_rate":
|
339 |
-
"loss": 0.
|
340 |
"step": 23000
|
341 |
},
|
342 |
{
|
343 |
"epoch": 1.9149282920469362,
|
344 |
-
"grad_norm": 3.
|
345 |
-
"learning_rate":
|
346 |
-
"loss": 0.
|
347 |
"step": 23500
|
348 |
},
|
349 |
{
|
350 |
"epoch": 1.9556714471968708,
|
351 |
-
"grad_norm": 5.
|
352 |
-
"learning_rate":
|
353 |
-
"loss": 0.
|
354 |
"step": 24000
|
355 |
},
|
356 |
{
|
357 |
"epoch": 1.996414602346806,
|
358 |
-
"grad_norm": 3.
|
359 |
-
"learning_rate":
|
360 |
-
"loss": 0.
|
361 |
"step": 24500
|
362 |
},
|
363 |
{
|
364 |
"epoch": 2.0,
|
365 |
-
"eval_accuracy": 0.
|
366 |
-
"eval_loss": 0.
|
367 |
-
"eval_runtime":
|
368 |
-
"eval_samples_per_second":
|
369 |
-
"eval_steps_per_second":
|
370 |
"step": 24544
|
371 |
},
|
372 |
{
|
373 |
"epoch": 2.0371577574967406,
|
374 |
-
"grad_norm":
|
375 |
-
"learning_rate":
|
376 |
-
"loss": 0.
|
377 |
"step": 25000
|
378 |
},
|
379 |
{
|
380 |
"epoch": 2.077900912646675,
|
381 |
-
"grad_norm": 3.
|
382 |
-
"learning_rate":
|
383 |
-
"loss": 0.
|
384 |
"step": 25500
|
385 |
},
|
386 |
{
|
387 |
"epoch": 2.1186440677966103,
|
388 |
-
"grad_norm": 3.
|
389 |
-
"learning_rate":
|
390 |
-
"loss": 0.
|
391 |
"step": 26000
|
392 |
},
|
393 |
{
|
394 |
"epoch": 2.159387222946545,
|
395 |
-
"grad_norm": 3.
|
396 |
-
"learning_rate":
|
397 |
-
"loss": 0.
|
398 |
"step": 26500
|
399 |
},
|
400 |
{
|
401 |
"epoch": 2.2001303780964796,
|
402 |
-
"grad_norm": 3.
|
403 |
-
"learning_rate":
|
404 |
-
"loss": 0.
|
405 |
"step": 27000
|
406 |
},
|
407 |
{
|
408 |
"epoch": 2.2408735332464147,
|
409 |
-
"grad_norm":
|
410 |
-
"learning_rate":
|
411 |
-
"loss": 0.
|
412 |
"step": 27500
|
413 |
},
|
414 |
{
|
415 |
"epoch": 2.2816166883963493,
|
416 |
-
"grad_norm":
|
417 |
-
"learning_rate":
|
418 |
-
"loss": 0.
|
419 |
"step": 28000
|
420 |
},
|
421 |
{
|
422 |
"epoch": 2.322359843546284,
|
423 |
-
"grad_norm": 2.
|
424 |
-
"learning_rate":
|
425 |
-
"loss": 0.
|
426 |
"step": 28500
|
427 |
},
|
428 |
{
|
429 |
"epoch": 2.363102998696219,
|
430 |
-
"grad_norm":
|
431 |
-
"learning_rate":
|
432 |
-
"loss": 0.
|
433 |
"step": 29000
|
434 |
},
|
435 |
{
|
436 |
"epoch": 2.4038461538461537,
|
437 |
-
"grad_norm":
|
438 |
-
"learning_rate":
|
439 |
-
"loss": 0.
|
440 |
"step": 29500
|
441 |
},
|
442 |
{
|
443 |
"epoch": 2.444589308996089,
|
444 |
-
"grad_norm": 4.
|
445 |
-
"learning_rate":
|
446 |
-
"loss": 0.
|
447 |
"step": 30000
|
448 |
},
|
449 |
{
|
450 |
"epoch": 2.4853324641460235,
|
451 |
-
"grad_norm":
|
452 |
-
"learning_rate":
|
453 |
-
"loss": 0.
|
454 |
"step": 30500
|
455 |
},
|
456 |
{
|
457 |
"epoch": 2.526075619295958,
|
458 |
-
"grad_norm":
|
459 |
-
"learning_rate":
|
460 |
-
"loss": 0.
|
461 |
"step": 31000
|
462 |
},
|
463 |
{
|
464 |
"epoch": 2.5668187744458932,
|
465 |
-
"grad_norm":
|
466 |
-
"learning_rate":
|
467 |
-
"loss": 0.
|
468 |
"step": 31500
|
469 |
},
|
470 |
{
|
471 |
"epoch": 2.607561929595828,
|
472 |
-
"grad_norm":
|
473 |
-
"learning_rate":
|
474 |
-
"loss": 0.
|
475 |
"step": 32000
|
476 |
},
|
477 |
{
|
478 |
"epoch": 2.648305084745763,
|
479 |
-
"grad_norm":
|
480 |
-
"learning_rate":
|
481 |
-
"loss": 0.
|
482 |
"step": 32500
|
483 |
},
|
484 |
{
|
485 |
"epoch": 2.6890482398956976,
|
486 |
-
"grad_norm":
|
487 |
-
"learning_rate":
|
488 |
-
"loss": 0.
|
489 |
"step": 33000
|
490 |
},
|
491 |
{
|
492 |
"epoch": 2.7297913950456323,
|
493 |
-
"grad_norm":
|
494 |
-
"learning_rate":
|
495 |
-
"loss": 0.
|
496 |
"step": 33500
|
497 |
},
|
498 |
{
|
499 |
"epoch": 2.770534550195567,
|
500 |
-
"grad_norm":
|
501 |
-
"learning_rate":
|
502 |
-
"loss": 0.
|
503 |
"step": 34000
|
504 |
},
|
505 |
{
|
506 |
"epoch": 2.811277705345502,
|
507 |
-
"grad_norm":
|
508 |
-
"learning_rate":
|
509 |
-
"loss": 0.
|
510 |
"step": 34500
|
511 |
},
|
512 |
{
|
513 |
"epoch": 2.8520208604954367,
|
514 |
-
"grad_norm":
|
515 |
-
"learning_rate":
|
516 |
-
"loss": 0.
|
517 |
"step": 35000
|
518 |
},
|
519 |
{
|
520 |
"epoch": 2.8927640156453718,
|
521 |
-
"grad_norm":
|
522 |
-
"learning_rate":
|
523 |
-
"loss": 0.
|
524 |
"step": 35500
|
525 |
},
|
526 |
{
|
527 |
"epoch": 2.9335071707953064,
|
528 |
-
"grad_norm": 4.
|
529 |
-
"learning_rate":
|
530 |
-
"loss": 0.
|
531 |
"step": 36000
|
532 |
},
|
533 |
{
|
534 |
"epoch": 2.974250325945241,
|
535 |
-
"grad_norm": 3.
|
536 |
-
"learning_rate":
|
537 |
-
"loss": 0.
|
538 |
"step": 36500
|
539 |
},
|
540 |
{
|
541 |
"epoch": 3.0,
|
542 |
-
"eval_accuracy": 0.
|
543 |
-
"eval_loss": 0.
|
544 |
-
"eval_runtime": 3.
|
545 |
-
"eval_samples_per_second":
|
546 |
-
"eval_steps_per_second":
|
547 |
"step": 36816
|
548 |
},
|
549 |
{
|
550 |
"epoch": 3.014993481095176,
|
551 |
-
"grad_norm":
|
552 |
-
"learning_rate":
|
553 |
-
"loss": 0.
|
554 |
"step": 37000
|
555 |
},
|
556 |
{
|
557 |
"epoch": 3.055736636245111,
|
558 |
-
"grad_norm":
|
559 |
-
"learning_rate":
|
560 |
-
"loss": 0.
|
561 |
"step": 37500
|
562 |
},
|
563 |
{
|
564 |
"epoch": 3.0964797913950455,
|
565 |
-
"grad_norm":
|
566 |
-
"learning_rate":
|
567 |
-
"loss": 0.
|
568 |
"step": 38000
|
569 |
},
|
570 |
{
|
571 |
"epoch": 3.1372229465449806,
|
572 |
-
"grad_norm":
|
573 |
-
"learning_rate":
|
574 |
-
"loss": 0.
|
575 |
"step": 38500
|
576 |
},
|
577 |
{
|
578 |
"epoch": 3.1779661016949152,
|
579 |
-
"grad_norm":
|
580 |
-
"learning_rate":
|
581 |
-
"loss": 0.
|
582 |
"step": 39000
|
583 |
},
|
584 |
{
|
585 |
"epoch": 3.21870925684485,
|
586 |
-
"grad_norm":
|
587 |
-
"learning_rate":
|
588 |
-
"loss": 0.
|
589 |
"step": 39500
|
590 |
},
|
591 |
{
|
592 |
"epoch": 3.259452411994785,
|
593 |
-
"grad_norm": 6.
|
594 |
-
"learning_rate":
|
595 |
-
"loss": 0.
|
596 |
"step": 40000
|
597 |
},
|
598 |
{
|
599 |
"epoch": 3.3001955671447196,
|
600 |
-
"grad_norm":
|
601 |
-
"learning_rate":
|
602 |
-
"loss": 0.
|
603 |
"step": 40500
|
604 |
},
|
605 |
{
|
606 |
"epoch": 3.3409387222946547,
|
607 |
-
"grad_norm": 3.
|
608 |
-
"learning_rate":
|
609 |
-
"loss": 0.
|
610 |
"step": 41000
|
611 |
},
|
612 |
{
|
613 |
"epoch": 3.3816818774445894,
|
614 |
-
"grad_norm":
|
615 |
-
"learning_rate":
|
616 |
-
"loss": 0.
|
617 |
"step": 41500
|
618 |
},
|
619 |
{
|
620 |
"epoch": 3.422425032594524,
|
621 |
-
"grad_norm":
|
622 |
-
"learning_rate":
|
623 |
-
"loss": 0.
|
624 |
"step": 42000
|
625 |
},
|
626 |
{
|
627 |
"epoch": 3.463168187744459,
|
628 |
-
"grad_norm":
|
629 |
-
"learning_rate":
|
630 |
-
"loss": 0.
|
631 |
"step": 42500
|
632 |
},
|
633 |
{
|
634 |
"epoch": 3.5039113428943938,
|
635 |
-
"grad_norm":
|
636 |
-
"learning_rate":
|
637 |
-
"loss": 0.
|
638 |
"step": 43000
|
639 |
},
|
640 |
{
|
641 |
"epoch": 3.5446544980443284,
|
642 |
-
"grad_norm":
|
643 |
-
"learning_rate":
|
644 |
-
"loss": 0.
|
645 |
"step": 43500
|
646 |
},
|
647 |
{
|
648 |
"epoch": 3.5853976531942635,
|
649 |
-
"grad_norm":
|
650 |
-
"learning_rate":
|
651 |
-
"loss": 0.
|
652 |
"step": 44000
|
653 |
},
|
654 |
{
|
655 |
"epoch": 3.626140808344198,
|
656 |
-
"grad_norm":
|
657 |
-
"learning_rate":
|
658 |
-
"loss": 0.
|
659 |
"step": 44500
|
660 |
},
|
661 |
{
|
662 |
"epoch": 3.666883963494133,
|
663 |
-
"grad_norm":
|
664 |
-
"learning_rate":
|
665 |
-
"loss": 0.
|
666 |
"step": 45000
|
667 |
},
|
668 |
{
|
669 |
"epoch": 3.707627118644068,
|
670 |
-
"grad_norm":
|
671 |
-
"learning_rate":
|
672 |
-
"loss": 0.
|
673 |
"step": 45500
|
674 |
},
|
675 |
{
|
676 |
"epoch": 3.7483702737940026,
|
677 |
-
"grad_norm": 4.
|
678 |
-
"learning_rate":
|
679 |
-
"loss": 0.
|
680 |
"step": 46000
|
681 |
},
|
682 |
{
|
683 |
"epoch": 3.7891134289439377,
|
684 |
-
"grad_norm":
|
685 |
-
"learning_rate":
|
686 |
-
"loss": 0.
|
687 |
"step": 46500
|
688 |
},
|
689 |
{
|
690 |
"epoch": 3.8298565840938723,
|
691 |
-
"grad_norm":
|
692 |
-
"learning_rate":
|
693 |
-
"loss": 0.
|
694 |
"step": 47000
|
695 |
},
|
696 |
{
|
697 |
"epoch": 3.870599739243807,
|
698 |
-
"grad_norm":
|
699 |
-
"learning_rate":
|
700 |
-
"loss": 0.
|
701 |
"step": 47500
|
702 |
},
|
703 |
{
|
704 |
"epoch": 3.9113428943937416,
|
705 |
-
"grad_norm": 3.
|
706 |
-
"learning_rate":
|
707 |
-
"loss": 0.
|
708 |
"step": 48000
|
709 |
},
|
710 |
{
|
711 |
"epoch": 3.9520860495436767,
|
712 |
-
"grad_norm":
|
713 |
-
"learning_rate":
|
714 |
-
"loss": 0.
|
715 |
"step": 48500
|
716 |
},
|
717 |
{
|
718 |
"epoch": 3.9928292046936114,
|
719 |
-
"grad_norm":
|
720 |
-
"learning_rate":
|
721 |
-
"loss": 0.
|
722 |
"step": 49000
|
723 |
},
|
724 |
{
|
725 |
"epoch": 4.0,
|
726 |
-
"eval_accuracy": 0.
|
727 |
-
"eval_loss": 0.
|
728 |
-
"eval_runtime": 3.
|
729 |
-
"eval_samples_per_second":
|
730 |
-
"eval_steps_per_second":
|
731 |
"step": 49088
|
732 |
},
|
733 |
{
|
734 |
"epoch": 4.0335723598435465,
|
735 |
-
"grad_norm":
|
736 |
-
"learning_rate":
|
737 |
-
"loss": 0.
|
738 |
"step": 49500
|
739 |
},
|
740 |
{
|
741 |
"epoch": 4.074315514993481,
|
742 |
-
"grad_norm":
|
743 |
-
"learning_rate":
|
744 |
-
"loss": 0.
|
745 |
"step": 50000
|
746 |
},
|
747 |
{
|
748 |
"epoch": 4.115058670143416,
|
749 |
-
"grad_norm":
|
750 |
-
"learning_rate":
|
751 |
-
"loss": 0.
|
752 |
"step": 50500
|
753 |
},
|
754 |
{
|
755 |
"epoch": 4.15580182529335,
|
756 |
-
"grad_norm":
|
757 |
-
"learning_rate":
|
758 |
-
"loss": 0.
|
759 |
"step": 51000
|
760 |
},
|
761 |
{
|
762 |
"epoch": 4.196544980443286,
|
763 |
-
"grad_norm": 5.
|
764 |
-
"learning_rate":
|
765 |
-
"loss": 0.
|
766 |
"step": 51500
|
767 |
},
|
768 |
{
|
769 |
"epoch": 4.237288135593221,
|
770 |
-
"grad_norm": 5.
|
771 |
-
"learning_rate":
|
772 |
-
"loss": 0.
|
773 |
"step": 52000
|
774 |
},
|
775 |
{
|
776 |
"epoch": 4.278031290743155,
|
777 |
-
"grad_norm":
|
778 |
-
"learning_rate":
|
779 |
-
"loss": 0.
|
780 |
"step": 52500
|
781 |
},
|
782 |
{
|
783 |
"epoch": 4.31877444589309,
|
784 |
-
"grad_norm":
|
785 |
-
"learning_rate":
|
786 |
-
"loss": 0.
|
787 |
"step": 53000
|
788 |
},
|
789 |
{
|
790 |
"epoch": 4.3595176010430245,
|
791 |
-
"grad_norm": 4.
|
792 |
-
"learning_rate":
|
793 |
-
"loss": 0.
|
794 |
"step": 53500
|
795 |
},
|
796 |
{
|
797 |
"epoch": 4.400260756192959,
|
798 |
-
"grad_norm":
|
799 |
-
"learning_rate":
|
800 |
-
"loss": 0.
|
801 |
"step": 54000
|
802 |
},
|
803 |
{
|
804 |
"epoch": 4.441003911342895,
|
805 |
-
"grad_norm":
|
806 |
-
"learning_rate":
|
807 |
-
"loss": 0.
|
808 |
"step": 54500
|
809 |
},
|
810 |
{
|
811 |
"epoch": 4.481747066492829,
|
812 |
-
"grad_norm":
|
813 |
-
"learning_rate":
|
814 |
-
"loss": 0.
|
815 |
"step": 55000
|
816 |
},
|
817 |
{
|
818 |
"epoch": 4.522490221642764,
|
819 |
-
"grad_norm": 4.
|
820 |
-
"learning_rate":
|
821 |
-
"loss": 0.
|
822 |
"step": 55500
|
823 |
},
|
824 |
{
|
825 |
"epoch": 4.563233376792699,
|
826 |
-
"grad_norm":
|
827 |
-
"learning_rate":
|
828 |
-
"loss": 0.
|
829 |
"step": 56000
|
830 |
},
|
831 |
{
|
832 |
"epoch": 4.603976531942633,
|
833 |
-
"grad_norm":
|
834 |
-
"learning_rate":
|
835 |
-
"loss": 0.
|
836 |
"step": 56500
|
837 |
},
|
838 |
{
|
839 |
"epoch": 4.644719687092568,
|
840 |
-
"grad_norm":
|
841 |
-
"learning_rate":
|
842 |
-
"loss": 0.
|
843 |
"step": 57000
|
844 |
},
|
845 |
{
|
846 |
"epoch": 4.6854628422425035,
|
847 |
-
"grad_norm":
|
848 |
-
"learning_rate":
|
849 |
-
"loss": 0.
|
850 |
"step": 57500
|
851 |
},
|
852 |
{
|
853 |
"epoch": 4.726205997392438,
|
854 |
-
"grad_norm":
|
855 |
-
"learning_rate":
|
856 |
-
"loss": 0.
|
857 |
"step": 58000
|
858 |
},
|
859 |
{
|
860 |
"epoch": 4.766949152542373,
|
861 |
-
"grad_norm":
|
862 |
-
"learning_rate":
|
863 |
-
"loss": 0.
|
864 |
"step": 58500
|
865 |
},
|
866 |
{
|
867 |
"epoch": 4.8076923076923075,
|
868 |
-
"grad_norm": 3.
|
869 |
-
"learning_rate":
|
870 |
-
"loss": 0.
|
871 |
"step": 59000
|
872 |
},
|
873 |
{
|
874 |
"epoch": 4.848435462842242,
|
875 |
-
"grad_norm": 4.
|
876 |
-
"learning_rate":
|
877 |
-
"loss": 0.
|
878 |
"step": 59500
|
879 |
},
|
880 |
{
|
881 |
"epoch": 4.889178617992178,
|
882 |
-
"grad_norm":
|
883 |
-
"learning_rate":
|
884 |
-
"loss": 0.
|
885 |
"step": 60000
|
886 |
},
|
887 |
{
|
888 |
"epoch": 4.929921773142112,
|
889 |
-
"grad_norm":
|
890 |
-
"learning_rate":
|
891 |
-
"loss": 0.
|
892 |
"step": 60500
|
893 |
},
|
894 |
{
|
895 |
"epoch": 4.970664928292047,
|
896 |
-
"grad_norm":
|
897 |
-
"learning_rate":
|
898 |
-
"loss": 0.
|
899 |
"step": 61000
|
900 |
},
|
901 |
{
|
902 |
"epoch": 5.0,
|
903 |
-
"eval_accuracy": 0.
|
904 |
-
"eval_loss": 0.
|
905 |
-
"eval_runtime": 3.
|
906 |
-
"eval_samples_per_second":
|
907 |
-
"eval_steps_per_second":
|
908 |
"step": 61360
|
909 |
},
|
910 |
{
|
911 |
-
"epoch": 5.
|
912 |
-
"
|
913 |
-
"
|
914 |
-
"
|
915 |
-
"
|
916 |
-
|
917 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
918 |
}
|
919 |
],
|
920 |
"logging_steps": 500,
|
921 |
-
"max_steps":
|
922 |
"num_input_tokens_seen": 0,
|
923 |
-
"num_train_epochs":
|
924 |
"save_steps": 500,
|
925 |
"stateful_callbacks": {
|
926 |
"TrainerControl": {
|
@@ -934,7 +1840,7 @@
|
|
934 |
"attributes": {}
|
935 |
}
|
936 |
},
|
937 |
-
"total_flos":
|
938 |
"train_batch_size": 32,
|
939 |
"trial_name": null,
|
940 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 10.0,
|
5 |
"eval_steps": 500,
|
6 |
+
"global_step": 122720,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
10 |
"log_history": [
|
11 |
{
|
12 |
"epoch": 0.04074315514993481,
|
13 |
+
"grad_norm": 3.142120361328125,
|
14 |
+
"learning_rate": 0.0002987777053455019,
|
15 |
+
"loss": 0.9832,
|
16 |
"step": 500
|
17 |
},
|
18 |
{
|
19 |
"epoch": 0.08148631029986962,
|
20 |
+
"grad_norm": 2.901562213897705,
|
21 |
+
"learning_rate": 0.0002975554106910039,
|
22 |
+
"loss": 0.7704,
|
23 |
"step": 1000
|
24 |
},
|
25 |
{
|
26 |
"epoch": 0.12222946544980444,
|
27 |
+
"grad_norm": 4.939942836761475,
|
28 |
+
"learning_rate": 0.00029633311603650584,
|
29 |
+
"loss": 0.7042,
|
30 |
"step": 1500
|
31 |
},
|
32 |
{
|
33 |
"epoch": 0.16297262059973924,
|
34 |
+
"grad_norm": 4.045433044433594,
|
35 |
+
"learning_rate": 0.0002951108213820078,
|
36 |
+
"loss": 0.6688,
|
37 |
"step": 2000
|
38 |
},
|
39 |
{
|
40 |
"epoch": 0.20371577574967406,
|
41 |
+
"grad_norm": 4.076220989227295,
|
42 |
+
"learning_rate": 0.00029388852672750977,
|
43 |
+
"loss": 0.6432,
|
44 |
"step": 2500
|
45 |
},
|
46 |
{
|
47 |
"epoch": 0.24445893089960888,
|
48 |
+
"grad_norm": 3.716000556945801,
|
49 |
+
"learning_rate": 0.0002926662320730117,
|
50 |
+
"loss": 0.6165,
|
51 |
"step": 3000
|
52 |
},
|
53 |
{
|
54 |
"epoch": 0.28520208604954367,
|
55 |
+
"grad_norm": 3.549494981765747,
|
56 |
+
"learning_rate": 0.00029144393741851364,
|
57 |
+
"loss": 0.6124,
|
58 |
"step": 3500
|
59 |
},
|
60 |
{
|
61 |
"epoch": 0.3259452411994785,
|
62 |
+
"grad_norm": 4.904158592224121,
|
63 |
+
"learning_rate": 0.0002902216427640156,
|
64 |
+
"loss": 0.587,
|
65 |
"step": 4000
|
66 |
},
|
67 |
{
|
68 |
"epoch": 0.3666883963494133,
|
69 |
+
"grad_norm": 3.506251573562622,
|
70 |
+
"learning_rate": 0.0002889993481095176,
|
71 |
+
"loss": 0.5961,
|
72 |
"step": 4500
|
73 |
},
|
74 |
{
|
75 |
"epoch": 0.4074315514993481,
|
76 |
+
"grad_norm": 3.886331558227539,
|
77 |
+
"learning_rate": 0.00028777705345501956,
|
78 |
+
"loss": 0.5834,
|
79 |
"step": 5000
|
80 |
},
|
81 |
{
|
82 |
"epoch": 0.44817470664928294,
|
83 |
+
"grad_norm": 5.864737510681152,
|
84 |
+
"learning_rate": 0.0002865547588005215,
|
85 |
+
"loss": 0.5782,
|
86 |
"step": 5500
|
87 |
},
|
88 |
{
|
89 |
"epoch": 0.48891786179921776,
|
90 |
+
"grad_norm": 2.608105421066284,
|
91 |
+
"learning_rate": 0.00028533246414602344,
|
92 |
+
"loss": 0.5744,
|
93 |
"step": 6000
|
94 |
},
|
95 |
{
|
96 |
"epoch": 0.5296610169491526,
|
97 |
+
"grad_norm": 3.3634021282196045,
|
98 |
+
"learning_rate": 0.00028411016949152543,
|
99 |
+
"loss": 0.5754,
|
100 |
"step": 6500
|
101 |
},
|
102 |
{
|
103 |
"epoch": 0.5704041720990873,
|
104 |
+
"grad_norm": 4.730838775634766,
|
105 |
+
"learning_rate": 0.00028288787483702737,
|
106 |
+
"loss": 0.5568,
|
107 |
"step": 7000
|
108 |
},
|
109 |
{
|
110 |
"epoch": 0.6111473272490222,
|
111 |
+
"grad_norm": 3.3426826000213623,
|
112 |
+
"learning_rate": 0.0002816655801825293,
|
113 |
+
"loss": 0.5591,
|
114 |
"step": 7500
|
115 |
},
|
116 |
{
|
117 |
"epoch": 0.651890482398957,
|
118 |
+
"grad_norm": 4.444048881530762,
|
119 |
+
"learning_rate": 0.00028044328552803124,
|
120 |
+
"loss": 0.5623,
|
121 |
"step": 8000
|
122 |
},
|
123 |
{
|
124 |
"epoch": 0.6926336375488917,
|
125 |
+
"grad_norm": 2.7828869819641113,
|
126 |
+
"learning_rate": 0.00027922099087353323,
|
127 |
+
"loss": 0.5518,
|
128 |
"step": 8500
|
129 |
},
|
130 |
{
|
131 |
"epoch": 0.7333767926988266,
|
132 |
+
"grad_norm": 4.194392681121826,
|
133 |
+
"learning_rate": 0.00027799869621903517,
|
134 |
+
"loss": 0.5591,
|
135 |
"step": 9000
|
136 |
},
|
137 |
{
|
138 |
"epoch": 0.7741199478487614,
|
139 |
+
"grad_norm": 5.856390476226807,
|
140 |
+
"learning_rate": 0.0002767764015645371,
|
141 |
+
"loss": 0.5526,
|
142 |
"step": 9500
|
143 |
},
|
144 |
{
|
145 |
"epoch": 0.8148631029986962,
|
146 |
+
"grad_norm": 4.61172342300415,
|
147 |
+
"learning_rate": 0.0002755541069100391,
|
148 |
+
"loss": 0.54,
|
149 |
"step": 10000
|
150 |
},
|
151 |
{
|
152 |
"epoch": 0.855606258148631,
|
153 |
+
"grad_norm": 4.518310070037842,
|
154 |
+
"learning_rate": 0.00027433181225554104,
|
155 |
+
"loss": 0.5432,
|
156 |
"step": 10500
|
157 |
},
|
158 |
{
|
159 |
"epoch": 0.8963494132985659,
|
160 |
+
"grad_norm": 3.190687417984009,
|
161 |
+
"learning_rate": 0.000273109517601043,
|
162 |
+
"loss": 0.5424,
|
163 |
"step": 11000
|
164 |
},
|
165 |
{
|
166 |
"epoch": 0.9370925684485006,
|
167 |
+
"grad_norm": 5.003259181976318,
|
168 |
+
"learning_rate": 0.00027188722294654497,
|
169 |
+
"loss": 0.5408,
|
170 |
"step": 11500
|
171 |
},
|
172 |
{
|
173 |
"epoch": 0.9778357235984355,
|
174 |
+
"grad_norm": 5.045399188995361,
|
175 |
+
"learning_rate": 0.0002706649282920469,
|
176 |
+
"loss": 0.5345,
|
177 |
"step": 12000
|
178 |
},
|
179 |
{
|
180 |
"epoch": 1.0,
|
181 |
+
"eval_accuracy": 0.810441792011261,
|
182 |
+
"eval_loss": 0.476459264755249,
|
183 |
+
"eval_runtime": 3.5774,
|
184 |
+
"eval_samples_per_second": 696.032,
|
185 |
+
"eval_steps_per_second": 87.214,
|
186 |
"step": 12272
|
187 |
},
|
188 |
{
|
189 |
"epoch": 1.0185788787483703,
|
190 |
+
"grad_norm": 3.0707671642303467,
|
191 |
+
"learning_rate": 0.00026944263363754884,
|
192 |
+
"loss": 0.5299,
|
193 |
"step": 12500
|
194 |
},
|
195 |
{
|
196 |
"epoch": 1.0593220338983051,
|
197 |
+
"grad_norm": 3.0093071460723877,
|
198 |
+
"learning_rate": 0.00026822033898305083,
|
199 |
+
"loss": 0.5162,
|
200 |
"step": 13000
|
201 |
},
|
202 |
{
|
203 |
"epoch": 1.1000651890482398,
|
204 |
+
"grad_norm": 5.579827785491943,
|
205 |
+
"learning_rate": 0.00026699804432855277,
|
206 |
+
"loss": 0.5202,
|
207 |
"step": 13500
|
208 |
},
|
209 |
{
|
210 |
"epoch": 1.1408083441981747,
|
211 |
+
"grad_norm": 4.641988754272461,
|
212 |
+
"learning_rate": 0.00026577574967405476,
|
213 |
+
"loss": 0.5164,
|
214 |
"step": 14000
|
215 |
},
|
216 |
{
|
217 |
"epoch": 1.1815514993481095,
|
218 |
+
"grad_norm": 3.9216480255126953,
|
219 |
+
"learning_rate": 0.0002645534550195567,
|
220 |
+
"loss": 0.5289,
|
221 |
"step": 14500
|
222 |
},
|
223 |
{
|
224 |
"epoch": 1.2222946544980444,
|
225 |
+
"grad_norm": 4.214970588684082,
|
226 |
+
"learning_rate": 0.00026333116036505864,
|
227 |
+
"loss": 0.5185,
|
228 |
"step": 15000
|
229 |
},
|
230 |
{
|
231 |
"epoch": 1.263037809647979,
|
232 |
+
"grad_norm": 5.768319606781006,
|
233 |
+
"learning_rate": 0.00026210886571056063,
|
234 |
+
"loss": 0.5232,
|
235 |
"step": 15500
|
236 |
},
|
237 |
{
|
238 |
"epoch": 1.303780964797914,
|
239 |
+
"grad_norm": 5.26653528213501,
|
240 |
+
"learning_rate": 0.00026088657105606257,
|
241 |
+
"loss": 0.5182,
|
242 |
"step": 16000
|
243 |
},
|
244 |
{
|
245 |
"epoch": 1.3445241199478488,
|
246 |
+
"grad_norm": 3.733856201171875,
|
247 |
+
"learning_rate": 0.0002596642764015645,
|
248 |
+
"loss": 0.5092,
|
249 |
"step": 16500
|
250 |
},
|
251 |
{
|
252 |
"epoch": 1.3852672750977835,
|
253 |
+
"grad_norm": 5.003189563751221,
|
254 |
+
"learning_rate": 0.0002584419817470665,
|
255 |
+
"loss": 0.525,
|
256 |
"step": 17000
|
257 |
},
|
258 |
{
|
259 |
"epoch": 1.4260104302477183,
|
260 |
+
"grad_norm": 5.187269687652588,
|
261 |
+
"learning_rate": 0.00025721968709256843,
|
262 |
+
"loss": 0.5141,
|
263 |
"step": 17500
|
264 |
},
|
265 |
{
|
266 |
"epoch": 1.4667535853976532,
|
267 |
+
"grad_norm": 3.338898181915283,
|
268 |
+
"learning_rate": 0.00025599739243807037,
|
269 |
+
"loss": 0.5059,
|
270 |
"step": 18000
|
271 |
},
|
272 |
{
|
273 |
"epoch": 1.5074967405475879,
|
274 |
+
"grad_norm": 3.167060136795044,
|
275 |
+
"learning_rate": 0.0002547750977835723,
|
276 |
+
"loss": 0.5113,
|
277 |
"step": 18500
|
278 |
},
|
279 |
{
|
280 |
"epoch": 1.548239895697523,
|
281 |
+
"grad_norm": 5.613772392272949,
|
282 |
+
"learning_rate": 0.0002535528031290743,
|
283 |
+
"loss": 0.5002,
|
284 |
"step": 19000
|
285 |
},
|
286 |
{
|
287 |
"epoch": 1.5889830508474576,
|
288 |
+
"grad_norm": 3.6054158210754395,
|
289 |
+
"learning_rate": 0.00025233050847457624,
|
290 |
+
"loss": 0.5108,
|
291 |
"step": 19500
|
292 |
},
|
293 |
{
|
294 |
"epoch": 1.6297262059973925,
|
295 |
+
"grad_norm": 3.535848617553711,
|
296 |
+
"learning_rate": 0.0002511082138200782,
|
297 |
+
"loss": 0.5134,
|
298 |
"step": 20000
|
299 |
},
|
300 |
{
|
301 |
"epoch": 1.6704693611473274,
|
302 |
+
"grad_norm": 2.494443655014038,
|
303 |
+
"learning_rate": 0.00024988591916558017,
|
304 |
+
"loss": 0.5104,
|
305 |
"step": 20500
|
306 |
},
|
307 |
{
|
308 |
"epoch": 1.711212516297262,
|
309 |
+
"grad_norm": 4.686164855957031,
|
310 |
+
"learning_rate": 0.0002486636245110821,
|
311 |
+
"loss": 0.4987,
|
312 |
"step": 21000
|
313 |
},
|
314 |
{
|
315 |
"epoch": 1.7519556714471969,
|
316 |
+
"grad_norm": 6.178253173828125,
|
317 |
+
"learning_rate": 0.00024744132985658404,
|
318 |
+
"loss": 0.501,
|
319 |
"step": 21500
|
320 |
},
|
321 |
{
|
322 |
"epoch": 1.7926988265971318,
|
323 |
+
"grad_norm": 3.4898393154144287,
|
324 |
+
"learning_rate": 0.00024621903520208603,
|
325 |
+
"loss": 0.504,
|
326 |
"step": 22000
|
327 |
},
|
328 |
{
|
329 |
"epoch": 1.8334419817470664,
|
330 |
+
"grad_norm": 5.974262237548828,
|
331 |
+
"learning_rate": 0.000244996740547588,
|
332 |
+
"loss": 0.5112,
|
333 |
"step": 22500
|
334 |
},
|
335 |
{
|
336 |
"epoch": 1.8741851368970013,
|
337 |
+
"grad_norm": 4.1161980628967285,
|
338 |
+
"learning_rate": 0.00024377444589308996,
|
339 |
+
"loss": 0.4934,
|
340 |
"step": 23000
|
341 |
},
|
342 |
{
|
343 |
"epoch": 1.9149282920469362,
|
344 |
+
"grad_norm": 3.786954164505005,
|
345 |
+
"learning_rate": 0.0002425521512385919,
|
346 |
+
"loss": 0.5051,
|
347 |
"step": 23500
|
348 |
},
|
349 |
{
|
350 |
"epoch": 1.9556714471968708,
|
351 |
+
"grad_norm": 5.365992546081543,
|
352 |
+
"learning_rate": 0.00024132985658409386,
|
353 |
+
"loss": 0.5011,
|
354 |
"step": 24000
|
355 |
},
|
356 |
{
|
357 |
"epoch": 1.996414602346806,
|
358 |
+
"grad_norm": 3.0433201789855957,
|
359 |
+
"learning_rate": 0.00024010756192959583,
|
360 |
+
"loss": 0.4965,
|
361 |
"step": 24500
|
362 |
},
|
363 |
{
|
364 |
"epoch": 2.0,
|
365 |
+
"eval_accuracy": 0.8253012299537659,
|
366 |
+
"eval_loss": 0.4476940929889679,
|
367 |
+
"eval_runtime": 3.5371,
|
368 |
+
"eval_samples_per_second": 703.959,
|
369 |
+
"eval_steps_per_second": 88.207,
|
370 |
"step": 24544
|
371 |
},
|
372 |
{
|
373 |
"epoch": 2.0371577574967406,
|
374 |
+
"grad_norm": 5.630396366119385,
|
375 |
+
"learning_rate": 0.00023888526727509777,
|
376 |
+
"loss": 0.4787,
|
377 |
"step": 25000
|
378 |
},
|
379 |
{
|
380 |
"epoch": 2.077900912646675,
|
381 |
+
"grad_norm": 3.847637891769409,
|
382 |
+
"learning_rate": 0.00023766297262059973,
|
383 |
+
"loss": 0.4754,
|
384 |
"step": 25500
|
385 |
},
|
386 |
{
|
387 |
"epoch": 2.1186440677966103,
|
388 |
+
"grad_norm": 3.24627685546875,
|
389 |
+
"learning_rate": 0.00023644067796610167,
|
390 |
+
"loss": 0.4802,
|
391 |
"step": 26000
|
392 |
},
|
393 |
{
|
394 |
"epoch": 2.159387222946545,
|
395 |
+
"grad_norm": 3.7128448486328125,
|
396 |
+
"learning_rate": 0.00023521838331160363,
|
397 |
+
"loss": 0.483,
|
398 |
"step": 26500
|
399 |
},
|
400 |
{
|
401 |
"epoch": 2.2001303780964796,
|
402 |
+
"grad_norm": 3.6413943767547607,
|
403 |
+
"learning_rate": 0.0002339960886571056,
|
404 |
+
"loss": 0.48,
|
405 |
"step": 27000
|
406 |
},
|
407 |
{
|
408 |
"epoch": 2.2408735332464147,
|
409 |
+
"grad_norm": 5.281874179840088,
|
410 |
+
"learning_rate": 0.00023277379400260753,
|
411 |
+
"loss": 0.4739,
|
412 |
"step": 27500
|
413 |
},
|
414 |
{
|
415 |
"epoch": 2.2816166883963493,
|
416 |
+
"grad_norm": 3.1675126552581787,
|
417 |
+
"learning_rate": 0.0002315514993481095,
|
418 |
+
"loss": 0.4868,
|
419 |
"step": 28000
|
420 |
},
|
421 |
{
|
422 |
"epoch": 2.322359843546284,
|
423 |
+
"grad_norm": 2.996647596359253,
|
424 |
+
"learning_rate": 0.00023032920469361144,
|
425 |
+
"loss": 0.47,
|
426 |
"step": 28500
|
427 |
},
|
428 |
{
|
429 |
"epoch": 2.363102998696219,
|
430 |
+
"grad_norm": 6.074376583099365,
|
431 |
+
"learning_rate": 0.0002291069100391134,
|
432 |
+
"loss": 0.4747,
|
433 |
"step": 29000
|
434 |
},
|
435 |
{
|
436 |
"epoch": 2.4038461538461537,
|
437 |
+
"grad_norm": 4.998762607574463,
|
438 |
+
"learning_rate": 0.00022788461538461537,
|
439 |
+
"loss": 0.4801,
|
440 |
"step": 29500
|
441 |
},
|
442 |
{
|
443 |
"epoch": 2.444589308996089,
|
444 |
+
"grad_norm": 4.5490031242370605,
|
445 |
+
"learning_rate": 0.0002266623207301173,
|
446 |
+
"loss": 0.4852,
|
447 |
"step": 30000
|
448 |
},
|
449 |
{
|
450 |
"epoch": 2.4853324641460235,
|
451 |
+
"grad_norm": 4.78202486038208,
|
452 |
+
"learning_rate": 0.00022544002607561927,
|
453 |
+
"loss": 0.4871,
|
454 |
"step": 30500
|
455 |
},
|
456 |
{
|
457 |
"epoch": 2.526075619295958,
|
458 |
+
"grad_norm": 4.470094680786133,
|
459 |
+
"learning_rate": 0.0002242177314211212,
|
460 |
+
"loss": 0.4913,
|
461 |
"step": 31000
|
462 |
},
|
463 |
{
|
464 |
"epoch": 2.5668187744458932,
|
465 |
+
"grad_norm": 8.678927421569824,
|
466 |
+
"learning_rate": 0.0002229954367666232,
|
467 |
+
"loss": 0.4852,
|
468 |
"step": 31500
|
469 |
},
|
470 |
{
|
471 |
"epoch": 2.607561929595828,
|
472 |
+
"grad_norm": 4.836784362792969,
|
473 |
+
"learning_rate": 0.00022177314211212516,
|
474 |
+
"loss": 0.4706,
|
475 |
"step": 32000
|
476 |
},
|
477 |
{
|
478 |
"epoch": 2.648305084745763,
|
479 |
+
"grad_norm": 3.676816940307617,
|
480 |
+
"learning_rate": 0.0002205508474576271,
|
481 |
+
"loss": 0.4759,
|
482 |
"step": 32500
|
483 |
},
|
484 |
{
|
485 |
"epoch": 2.6890482398956976,
|
486 |
+
"grad_norm": 4.985140800476074,
|
487 |
+
"learning_rate": 0.00021932855280312906,
|
488 |
+
"loss": 0.478,
|
489 |
"step": 33000
|
490 |
},
|
491 |
{
|
492 |
"epoch": 2.7297913950456323,
|
493 |
+
"grad_norm": 4.114302635192871,
|
494 |
+
"learning_rate": 0.00021810625814863103,
|
495 |
+
"loss": 0.4802,
|
496 |
"step": 33500
|
497 |
},
|
498 |
{
|
499 |
"epoch": 2.770534550195567,
|
500 |
+
"grad_norm": 3.370340585708618,
|
501 |
+
"learning_rate": 0.00021688396349413296,
|
502 |
+
"loss": 0.488,
|
503 |
"step": 34000
|
504 |
},
|
505 |
{
|
506 |
"epoch": 2.811277705345502,
|
507 |
+
"grad_norm": 4.248477935791016,
|
508 |
+
"learning_rate": 0.00021566166883963493,
|
509 |
+
"loss": 0.4807,
|
510 |
"step": 34500
|
511 |
},
|
512 |
{
|
513 |
"epoch": 2.8520208604954367,
|
514 |
+
"grad_norm": 4.306929588317871,
|
515 |
+
"learning_rate": 0.0002144393741851369,
|
516 |
+
"loss": 0.4647,
|
517 |
"step": 35000
|
518 |
},
|
519 |
{
|
520 |
"epoch": 2.8927640156453718,
|
521 |
+
"grad_norm": 4.3554253578186035,
|
522 |
+
"learning_rate": 0.00021321707953063883,
|
523 |
+
"loss": 0.47,
|
524 |
"step": 35500
|
525 |
},
|
526 |
{
|
527 |
"epoch": 2.9335071707953064,
|
528 |
+
"grad_norm": 4.758439064025879,
|
529 |
+
"learning_rate": 0.0002119947848761408,
|
530 |
+
"loss": 0.4796,
|
531 |
"step": 36000
|
532 |
},
|
533 |
{
|
534 |
"epoch": 2.974250325945241,
|
535 |
+
"grad_norm": 3.889279842376709,
|
536 |
+
"learning_rate": 0.00021077249022164273,
|
537 |
+
"loss": 0.465,
|
538 |
"step": 36500
|
539 |
},
|
540 |
{
|
541 |
"epoch": 3.0,
|
542 |
+
"eval_accuracy": 0.7819277048110962,
|
543 |
+
"eval_loss": 0.5279985070228577,
|
544 |
+
"eval_runtime": 3.5081,
|
545 |
+
"eval_samples_per_second": 709.784,
|
546 |
+
"eval_steps_per_second": 88.937,
|
547 |
"step": 36816
|
548 |
},
|
549 |
{
|
550 |
"epoch": 3.014993481095176,
|
551 |
+
"grad_norm": 2.9451563358306885,
|
552 |
+
"learning_rate": 0.0002095501955671447,
|
553 |
+
"loss": 0.4777,
|
554 |
"step": 37000
|
555 |
},
|
556 |
{
|
557 |
"epoch": 3.055736636245111,
|
558 |
+
"grad_norm": 4.705958843231201,
|
559 |
+
"learning_rate": 0.00020832790091264666,
|
560 |
+
"loss": 0.4567,
|
561 |
"step": 37500
|
562 |
},
|
563 |
{
|
564 |
"epoch": 3.0964797913950455,
|
565 |
+
"grad_norm": 4.704124450683594,
|
566 |
+
"learning_rate": 0.0002071056062581486,
|
567 |
+
"loss": 0.451,
|
568 |
"step": 38000
|
569 |
},
|
570 |
{
|
571 |
"epoch": 3.1372229465449806,
|
572 |
+
"grad_norm": 3.8129329681396484,
|
573 |
+
"learning_rate": 0.00020588331160365056,
|
574 |
+
"loss": 0.4615,
|
575 |
"step": 38500
|
576 |
},
|
577 |
{
|
578 |
"epoch": 3.1779661016949152,
|
579 |
+
"grad_norm": 4.089268207550049,
|
580 |
+
"learning_rate": 0.0002046610169491525,
|
581 |
+
"loss": 0.45,
|
582 |
"step": 39000
|
583 |
},
|
584 |
{
|
585 |
"epoch": 3.21870925684485,
|
586 |
+
"grad_norm": 4.320204257965088,
|
587 |
+
"learning_rate": 0.00020343872229465447,
|
588 |
+
"loss": 0.4502,
|
589 |
"step": 39500
|
590 |
},
|
591 |
{
|
592 |
"epoch": 3.259452411994785,
|
593 |
+
"grad_norm": 6.090836048126221,
|
594 |
+
"learning_rate": 0.00020221642764015643,
|
595 |
+
"loss": 0.4557,
|
596 |
"step": 40000
|
597 |
},
|
598 |
{
|
599 |
"epoch": 3.3001955671447196,
|
600 |
+
"grad_norm": 3.542620897293091,
|
601 |
+
"learning_rate": 0.0002009941329856584,
|
602 |
+
"loss": 0.4595,
|
603 |
"step": 40500
|
604 |
},
|
605 |
{
|
606 |
"epoch": 3.3409387222946547,
|
607 |
+
"grad_norm": 3.8268799781799316,
|
608 |
+
"learning_rate": 0.00019977183833116036,
|
609 |
+
"loss": 0.4514,
|
610 |
"step": 41000
|
611 |
},
|
612 |
{
|
613 |
"epoch": 3.3816818774445894,
|
614 |
+
"grad_norm": 3.9095098972320557,
|
615 |
+
"learning_rate": 0.00019854954367666232,
|
616 |
+
"loss": 0.4637,
|
617 |
"step": 41500
|
618 |
},
|
619 |
{
|
620 |
"epoch": 3.422425032594524,
|
621 |
+
"grad_norm": 4.160515785217285,
|
622 |
+
"learning_rate": 0.00019732724902216426,
|
623 |
+
"loss": 0.4458,
|
624 |
"step": 42000
|
625 |
},
|
626 |
{
|
627 |
"epoch": 3.463168187744459,
|
628 |
+
"grad_norm": 3.263516426086426,
|
629 |
+
"learning_rate": 0.00019610495436766623,
|
630 |
+
"loss": 0.4592,
|
631 |
"step": 42500
|
632 |
},
|
633 |
{
|
634 |
"epoch": 3.5039113428943938,
|
635 |
+
"grad_norm": 4.349822044372559,
|
636 |
+
"learning_rate": 0.00019488265971316816,
|
637 |
+
"loss": 0.463,
|
638 |
"step": 43000
|
639 |
},
|
640 |
{
|
641 |
"epoch": 3.5446544980443284,
|
642 |
+
"grad_norm": 4.780073642730713,
|
643 |
+
"learning_rate": 0.00019366036505867013,
|
644 |
+
"loss": 0.4579,
|
645 |
"step": 43500
|
646 |
},
|
647 |
{
|
648 |
"epoch": 3.5853976531942635,
|
649 |
+
"grad_norm": 5.213197708129883,
|
650 |
+
"learning_rate": 0.0001924380704041721,
|
651 |
+
"loss": 0.4537,
|
652 |
"step": 44000
|
653 |
},
|
654 |
{
|
655 |
"epoch": 3.626140808344198,
|
656 |
+
"grad_norm": 5.990653038024902,
|
657 |
+
"learning_rate": 0.00019121577574967403,
|
658 |
+
"loss": 0.4495,
|
659 |
"step": 44500
|
660 |
},
|
661 |
{
|
662 |
"epoch": 3.666883963494133,
|
663 |
+
"grad_norm": 8.249653816223145,
|
664 |
+
"learning_rate": 0.000189993481095176,
|
665 |
+
"loss": 0.4628,
|
666 |
"step": 45000
|
667 |
},
|
668 |
{
|
669 |
"epoch": 3.707627118644068,
|
670 |
+
"grad_norm": 7.788904666900635,
|
671 |
+
"learning_rate": 0.00018877118644067796,
|
672 |
+
"loss": 0.4583,
|
673 |
"step": 45500
|
674 |
},
|
675 |
{
|
676 |
"epoch": 3.7483702737940026,
|
677 |
+
"grad_norm": 4.790676116943359,
|
678 |
+
"learning_rate": 0.0001875488917861799,
|
679 |
+
"loss": 0.4479,
|
680 |
"step": 46000
|
681 |
},
|
682 |
{
|
683 |
"epoch": 3.7891134289439377,
|
684 |
+
"grad_norm": 6.014172077178955,
|
685 |
+
"learning_rate": 0.00018632659713168186,
|
686 |
+
"loss": 0.4482,
|
687 |
"step": 46500
|
688 |
},
|
689 |
{
|
690 |
"epoch": 3.8298565840938723,
|
691 |
+
"grad_norm": 4.927757263183594,
|
692 |
+
"learning_rate": 0.0001851043024771838,
|
693 |
+
"loss": 0.4614,
|
694 |
"step": 47000
|
695 |
},
|
696 |
{
|
697 |
"epoch": 3.870599739243807,
|
698 |
+
"grad_norm": 3.723923444747925,
|
699 |
+
"learning_rate": 0.00018388200782268576,
|
700 |
+
"loss": 0.459,
|
701 |
"step": 47500
|
702 |
},
|
703 |
{
|
704 |
"epoch": 3.9113428943937416,
|
705 |
+
"grad_norm": 3.931187391281128,
|
706 |
+
"learning_rate": 0.00018265971316818773,
|
707 |
+
"loss": 0.4635,
|
708 |
"step": 48000
|
709 |
},
|
710 |
{
|
711 |
"epoch": 3.9520860495436767,
|
712 |
+
"grad_norm": 3.8654162883758545,
|
713 |
+
"learning_rate": 0.00018143741851368966,
|
714 |
+
"loss": 0.4427,
|
715 |
"step": 48500
|
716 |
},
|
717 |
{
|
718 |
"epoch": 3.9928292046936114,
|
719 |
+
"grad_norm": 5.012121677398682,
|
720 |
+
"learning_rate": 0.00018021512385919163,
|
721 |
+
"loss": 0.4634,
|
722 |
"step": 49000
|
723 |
},
|
724 |
{
|
725 |
"epoch": 4.0,
|
726 |
+
"eval_accuracy": 0.8244979977607727,
|
727 |
+
"eval_loss": 0.4430766999721527,
|
728 |
+
"eval_runtime": 3.5278,
|
729 |
+
"eval_samples_per_second": 705.821,
|
730 |
+
"eval_steps_per_second": 88.44,
|
731 |
"step": 49088
|
732 |
},
|
733 |
{
|
734 |
"epoch": 4.0335723598435465,
|
735 |
+
"grad_norm": 4.3685173988342285,
|
736 |
+
"learning_rate": 0.00017899282920469362,
|
737 |
+
"loss": 0.4362,
|
738 |
"step": 49500
|
739 |
},
|
740 |
{
|
741 |
"epoch": 4.074315514993481,
|
742 |
+
"grad_norm": 7.587422847747803,
|
743 |
+
"learning_rate": 0.00017777053455019556,
|
744 |
+
"loss": 0.4204,
|
745 |
"step": 50000
|
746 |
},
|
747 |
{
|
748 |
"epoch": 4.115058670143416,
|
749 |
+
"grad_norm": 6.1443963050842285,
|
750 |
+
"learning_rate": 0.00017654823989569752,
|
751 |
+
"loss": 0.4345,
|
752 |
"step": 50500
|
753 |
},
|
754 |
{
|
755 |
"epoch": 4.15580182529335,
|
756 |
+
"grad_norm": 5.416019439697266,
|
757 |
+
"learning_rate": 0.00017532594524119946,
|
758 |
+
"loss": 0.4295,
|
759 |
"step": 51000
|
760 |
},
|
761 |
{
|
762 |
"epoch": 4.196544980443286,
|
763 |
+
"grad_norm": 5.976702690124512,
|
764 |
+
"learning_rate": 0.00017410365058670142,
|
765 |
+
"loss": 0.431,
|
766 |
"step": 51500
|
767 |
},
|
768 |
{
|
769 |
"epoch": 4.237288135593221,
|
770 |
+
"grad_norm": 5.283466815948486,
|
771 |
+
"learning_rate": 0.0001728813559322034,
|
772 |
+
"loss": 0.434,
|
773 |
"step": 52000
|
774 |
},
|
775 |
{
|
776 |
"epoch": 4.278031290743155,
|
777 |
+
"grad_norm": 6.41757345199585,
|
778 |
+
"learning_rate": 0.00017165906127770533,
|
779 |
+
"loss": 0.4283,
|
780 |
"step": 52500
|
781 |
},
|
782 |
{
|
783 |
"epoch": 4.31877444589309,
|
784 |
+
"grad_norm": 6.220736980438232,
|
785 |
+
"learning_rate": 0.0001704367666232073,
|
786 |
+
"loss": 0.4407,
|
787 |
"step": 53000
|
788 |
},
|
789 |
{
|
790 |
"epoch": 4.3595176010430245,
|
791 |
+
"grad_norm": 4.802826881408691,
|
792 |
+
"learning_rate": 0.00016921447196870926,
|
793 |
+
"loss": 0.4345,
|
794 |
"step": 53500
|
795 |
},
|
796 |
{
|
797 |
"epoch": 4.400260756192959,
|
798 |
+
"grad_norm": 3.7236640453338623,
|
799 |
+
"learning_rate": 0.0001679921773142112,
|
800 |
+
"loss": 0.4421,
|
801 |
"step": 54000
|
802 |
},
|
803 |
{
|
804 |
"epoch": 4.441003911342895,
|
805 |
+
"grad_norm": 5.5577712059021,
|
806 |
+
"learning_rate": 0.00016676988265971316,
|
807 |
+
"loss": 0.4391,
|
808 |
"step": 54500
|
809 |
},
|
810 |
{
|
811 |
"epoch": 4.481747066492829,
|
812 |
+
"grad_norm": 4.09696626663208,
|
813 |
+
"learning_rate": 0.0001655475880052151,
|
814 |
+
"loss": 0.4376,
|
815 |
"step": 55000
|
816 |
},
|
817 |
{
|
818 |
"epoch": 4.522490221642764,
|
819 |
+
"grad_norm": 4.795173645019531,
|
820 |
+
"learning_rate": 0.00016432529335071706,
|
821 |
+
"loss": 0.4316,
|
822 |
"step": 55500
|
823 |
},
|
824 |
{
|
825 |
"epoch": 4.563233376792699,
|
826 |
+
"grad_norm": 2.8243749141693115,
|
827 |
+
"learning_rate": 0.00016310299869621902,
|
828 |
+
"loss": 0.4341,
|
829 |
"step": 56000
|
830 |
},
|
831 |
{
|
832 |
"epoch": 4.603976531942633,
|
833 |
+
"grad_norm": 5.1735029220581055,
|
834 |
+
"learning_rate": 0.00016188070404172096,
|
835 |
+
"loss": 0.4402,
|
836 |
"step": 56500
|
837 |
},
|
838 |
{
|
839 |
"epoch": 4.644719687092568,
|
840 |
+
"grad_norm": 5.092055320739746,
|
841 |
+
"learning_rate": 0.00016065840938722293,
|
842 |
+
"loss": 0.4432,
|
843 |
"step": 57000
|
844 |
},
|
845 |
{
|
846 |
"epoch": 4.6854628422425035,
|
847 |
+
"grad_norm": 4.940371990203857,
|
848 |
+
"learning_rate": 0.00015943611473272486,
|
849 |
+
"loss": 0.4393,
|
850 |
"step": 57500
|
851 |
},
|
852 |
{
|
853 |
"epoch": 4.726205997392438,
|
854 |
+
"grad_norm": 5.340673446655273,
|
855 |
+
"learning_rate": 0.00015821382007822685,
|
856 |
+
"loss": 0.4337,
|
857 |
"step": 58000
|
858 |
},
|
859 |
{
|
860 |
"epoch": 4.766949152542373,
|
861 |
+
"grad_norm": 4.699270248413086,
|
862 |
+
"learning_rate": 0.00015699152542372882,
|
863 |
+
"loss": 0.4385,
|
864 |
"step": 58500
|
865 |
},
|
866 |
{
|
867 |
"epoch": 4.8076923076923075,
|
868 |
+
"grad_norm": 3.0422141551971436,
|
869 |
+
"learning_rate": 0.00015576923076923076,
|
870 |
+
"loss": 0.4451,
|
871 |
"step": 59000
|
872 |
},
|
873 |
{
|
874 |
"epoch": 4.848435462842242,
|
875 |
+
"grad_norm": 4.510776996612549,
|
876 |
+
"learning_rate": 0.00015454693611473272,
|
877 |
+
"loss": 0.4322,
|
878 |
"step": 59500
|
879 |
},
|
880 |
{
|
881 |
"epoch": 4.889178617992178,
|
882 |
+
"grad_norm": 5.1884565353393555,
|
883 |
+
"learning_rate": 0.00015332464146023469,
|
884 |
+
"loss": 0.4346,
|
885 |
"step": 60000
|
886 |
},
|
887 |
{
|
888 |
"epoch": 4.929921773142112,
|
889 |
+
"grad_norm": 5.881906509399414,
|
890 |
+
"learning_rate": 0.00015210234680573662,
|
891 |
+
"loss": 0.435,
|
892 |
"step": 60500
|
893 |
},
|
894 |
{
|
895 |
"epoch": 4.970664928292047,
|
896 |
+
"grad_norm": 6.772756099700928,
|
897 |
+
"learning_rate": 0.0001508800521512386,
|
898 |
+
"loss": 0.436,
|
899 |
"step": 61000
|
900 |
},
|
901 |
{
|
902 |
"epoch": 5.0,
|
903 |
+
"eval_accuracy": 0.8345381617546082,
|
904 |
+
"eval_loss": 0.42114126682281494,
|
905 |
+
"eval_runtime": 3.525,
|
906 |
+
"eval_samples_per_second": 706.383,
|
907 |
+
"eval_steps_per_second": 88.511,
|
908 |
"step": 61360
|
909 |
},
|
910 |
{
|
911 |
+
"epoch": 5.011408083441982,
|
912 |
+
"grad_norm": 6.8924736976623535,
|
913 |
+
"learning_rate": 0.00014965775749674052,
|
914 |
+
"loss": 0.4388,
|
915 |
+
"step": 61500
|
916 |
+
},
|
917 |
+
{
|
918 |
+
"epoch": 5.052151238591916,
|
919 |
+
"grad_norm": 7.1412882804870605,
|
920 |
+
"learning_rate": 0.0001484354628422425,
|
921 |
+
"loss": 0.4125,
|
922 |
+
"step": 62000
|
923 |
+
},
|
924 |
+
{
|
925 |
+
"epoch": 5.092894393741851,
|
926 |
+
"grad_norm": 7.67710542678833,
|
927 |
+
"learning_rate": 0.00014721316818774445,
|
928 |
+
"loss": 0.4091,
|
929 |
+
"step": 62500
|
930 |
+
},
|
931 |
+
{
|
932 |
+
"epoch": 5.1336375488917865,
|
933 |
+
"grad_norm": 5.673561096191406,
|
934 |
+
"learning_rate": 0.0001459908735332464,
|
935 |
+
"loss": 0.4138,
|
936 |
+
"step": 63000
|
937 |
+
},
|
938 |
+
{
|
939 |
+
"epoch": 5.174380704041721,
|
940 |
+
"grad_norm": 5.589047908782959,
|
941 |
+
"learning_rate": 0.00014476857887874836,
|
942 |
+
"loss": 0.4174,
|
943 |
+
"step": 63500
|
944 |
+
},
|
945 |
+
{
|
946 |
+
"epoch": 5.215123859191656,
|
947 |
+
"grad_norm": 3.591366767883301,
|
948 |
+
"learning_rate": 0.00014354628422425032,
|
949 |
+
"loss": 0.4116,
|
950 |
+
"step": 64000
|
951 |
+
},
|
952 |
+
{
|
953 |
+
"epoch": 5.25586701434159,
|
954 |
+
"grad_norm": 5.160641193389893,
|
955 |
+
"learning_rate": 0.00014232398956975226,
|
956 |
+
"loss": 0.4168,
|
957 |
+
"step": 64500
|
958 |
+
},
|
959 |
+
{
|
960 |
+
"epoch": 5.296610169491525,
|
961 |
+
"grad_norm": 7.442717552185059,
|
962 |
+
"learning_rate": 0.00014110169491525422,
|
963 |
+
"loss": 0.4137,
|
964 |
+
"step": 65000
|
965 |
+
},
|
966 |
+
{
|
967 |
+
"epoch": 5.337353324641461,
|
968 |
+
"grad_norm": 5.819495677947998,
|
969 |
+
"learning_rate": 0.0001398794002607562,
|
970 |
+
"loss": 0.4226,
|
971 |
+
"step": 65500
|
972 |
+
},
|
973 |
+
{
|
974 |
+
"epoch": 5.378096479791395,
|
975 |
+
"grad_norm": 5.86204719543457,
|
976 |
+
"learning_rate": 0.00013865710560625815,
|
977 |
+
"loss": 0.4158,
|
978 |
+
"step": 66000
|
979 |
+
},
|
980 |
+
{
|
981 |
+
"epoch": 5.41883963494133,
|
982 |
+
"grad_norm": 5.929037094116211,
|
983 |
+
"learning_rate": 0.0001374348109517601,
|
984 |
+
"loss": 0.4139,
|
985 |
+
"step": 66500
|
986 |
+
},
|
987 |
+
{
|
988 |
+
"epoch": 5.459582790091265,
|
989 |
+
"grad_norm": 6.2193427085876465,
|
990 |
+
"learning_rate": 0.00013621251629726205,
|
991 |
+
"loss": 0.4278,
|
992 |
+
"step": 67000
|
993 |
+
},
|
994 |
+
{
|
995 |
+
"epoch": 5.500325945241199,
|
996 |
+
"grad_norm": 3.6446192264556885,
|
997 |
+
"learning_rate": 0.000134990221642764,
|
998 |
+
"loss": 0.4263,
|
999 |
+
"step": 67500
|
1000 |
+
},
|
1001 |
+
{
|
1002 |
+
"epoch": 5.541069100391134,
|
1003 |
+
"grad_norm": 4.629395961761475,
|
1004 |
+
"learning_rate": 0.00013376792698826596,
|
1005 |
+
"loss": 0.4272,
|
1006 |
+
"step": 68000
|
1007 |
+
},
|
1008 |
+
{
|
1009 |
+
"epoch": 5.581812255541069,
|
1010 |
+
"grad_norm": 4.330788612365723,
|
1011 |
+
"learning_rate": 0.00013254563233376792,
|
1012 |
+
"loss": 0.4189,
|
1013 |
+
"step": 68500
|
1014 |
+
},
|
1015 |
+
{
|
1016 |
+
"epoch": 5.622555410691004,
|
1017 |
+
"grad_norm": 6.034902572631836,
|
1018 |
+
"learning_rate": 0.00013132333767926986,
|
1019 |
+
"loss": 0.4214,
|
1020 |
+
"step": 69000
|
1021 |
+
},
|
1022 |
+
{
|
1023 |
+
"epoch": 5.663298565840939,
|
1024 |
+
"grad_norm": 4.091671943664551,
|
1025 |
+
"learning_rate": 0.00013010104302477182,
|
1026 |
+
"loss": 0.4124,
|
1027 |
+
"step": 69500
|
1028 |
+
},
|
1029 |
+
{
|
1030 |
+
"epoch": 5.704041720990873,
|
1031 |
+
"grad_norm": 5.487693786621094,
|
1032 |
+
"learning_rate": 0.00012887874837027379,
|
1033 |
+
"loss": 0.4242,
|
1034 |
+
"step": 70000
|
1035 |
+
},
|
1036 |
+
{
|
1037 |
+
"epoch": 5.744784876140808,
|
1038 |
+
"grad_norm": 5.3786115646362305,
|
1039 |
+
"learning_rate": 0.00012765645371577575,
|
1040 |
+
"loss": 0.4298,
|
1041 |
+
"step": 70500
|
1042 |
+
},
|
1043 |
+
{
|
1044 |
+
"epoch": 5.7855280312907436,
|
1045 |
+
"grad_norm": 4.010681629180908,
|
1046 |
+
"learning_rate": 0.0001264341590612777,
|
1047 |
+
"loss": 0.4253,
|
1048 |
+
"step": 71000
|
1049 |
+
},
|
1050 |
+
{
|
1051 |
+
"epoch": 5.826271186440678,
|
1052 |
+
"grad_norm": 5.500887870788574,
|
1053 |
+
"learning_rate": 0.00012521186440677965,
|
1054 |
+
"loss": 0.4245,
|
1055 |
+
"step": 71500
|
1056 |
+
},
|
1057 |
+
{
|
1058 |
+
"epoch": 5.867014341590613,
|
1059 |
+
"grad_norm": 5.5760931968688965,
|
1060 |
+
"learning_rate": 0.0001239895697522816,
|
1061 |
+
"loss": 0.4102,
|
1062 |
+
"step": 72000
|
1063 |
+
},
|
1064 |
+
{
|
1065 |
+
"epoch": 5.9077574967405475,
|
1066 |
+
"grad_norm": 5.079644203186035,
|
1067 |
+
"learning_rate": 0.00012276727509778355,
|
1068 |
+
"loss": 0.4201,
|
1069 |
+
"step": 72500
|
1070 |
+
},
|
1071 |
+
{
|
1072 |
+
"epoch": 5.948500651890482,
|
1073 |
+
"grad_norm": 5.812402248382568,
|
1074 |
+
"learning_rate": 0.0001215449804432855,
|
1075 |
+
"loss": 0.4122,
|
1076 |
+
"step": 73000
|
1077 |
+
},
|
1078 |
+
{
|
1079 |
+
"epoch": 5.989243807040417,
|
1080 |
+
"grad_norm": 19.5577335357666,
|
1081 |
+
"learning_rate": 0.00012032268578878747,
|
1082 |
+
"loss": 0.4143,
|
1083 |
+
"step": 73500
|
1084 |
+
},
|
1085 |
+
{
|
1086 |
+
"epoch": 6.0,
|
1087 |
+
"eval_accuracy": 0.8341365456581116,
|
1088 |
+
"eval_loss": 0.4439743757247925,
|
1089 |
+
"eval_runtime": 3.5304,
|
1090 |
+
"eval_samples_per_second": 705.293,
|
1091 |
+
"eval_steps_per_second": 88.374,
|
1092 |
+
"step": 73632
|
1093 |
+
},
|
1094 |
+
{
|
1095 |
+
"epoch": 6.029986962190352,
|
1096 |
+
"grad_norm": 5.7537946701049805,
|
1097 |
+
"learning_rate": 0.00011910039113428943,
|
1098 |
+
"loss": 0.4095,
|
1099 |
+
"step": 74000
|
1100 |
+
},
|
1101 |
+
{
|
1102 |
+
"epoch": 6.070730117340287,
|
1103 |
+
"grad_norm": 4.046775817871094,
|
1104 |
+
"learning_rate": 0.00011787809647979139,
|
1105 |
+
"loss": 0.404,
|
1106 |
+
"step": 74500
|
1107 |
+
},
|
1108 |
+
{
|
1109 |
+
"epoch": 6.111473272490222,
|
1110 |
+
"grad_norm": 5.266561508178711,
|
1111 |
+
"learning_rate": 0.00011665580182529335,
|
1112 |
+
"loss": 0.3959,
|
1113 |
+
"step": 75000
|
1114 |
+
},
|
1115 |
+
{
|
1116 |
+
"epoch": 6.152216427640156,
|
1117 |
+
"grad_norm": 4.223605632781982,
|
1118 |
+
"learning_rate": 0.0001154335071707953,
|
1119 |
+
"loss": 0.3998,
|
1120 |
+
"step": 75500
|
1121 |
+
},
|
1122 |
+
{
|
1123 |
+
"epoch": 6.192959582790091,
|
1124 |
+
"grad_norm": 4.90071964263916,
|
1125 |
+
"learning_rate": 0.00011421121251629725,
|
1126 |
+
"loss": 0.398,
|
1127 |
+
"step": 76000
|
1128 |
+
},
|
1129 |
+
{
|
1130 |
+
"epoch": 6.2337027379400265,
|
1131 |
+
"grad_norm": 4.251009464263916,
|
1132 |
+
"learning_rate": 0.0001129889178617992,
|
1133 |
+
"loss": 0.4121,
|
1134 |
+
"step": 76500
|
1135 |
+
},
|
1136 |
+
{
|
1137 |
+
"epoch": 6.274445893089961,
|
1138 |
+
"grad_norm": 4.648316383361816,
|
1139 |
+
"learning_rate": 0.00011176662320730115,
|
1140 |
+
"loss": 0.4081,
|
1141 |
+
"step": 77000
|
1142 |
+
},
|
1143 |
+
{
|
1144 |
+
"epoch": 6.315189048239896,
|
1145 |
+
"grad_norm": 3.8864428997039795,
|
1146 |
+
"learning_rate": 0.00011054432855280312,
|
1147 |
+
"loss": 0.3964,
|
1148 |
+
"step": 77500
|
1149 |
+
},
|
1150 |
+
{
|
1151 |
+
"epoch": 6.3559322033898304,
|
1152 |
+
"grad_norm": 6.131734371185303,
|
1153 |
+
"learning_rate": 0.00010932203389830507,
|
1154 |
+
"loss": 0.4089,
|
1155 |
+
"step": 78000
|
1156 |
+
},
|
1157 |
+
{
|
1158 |
+
"epoch": 6.396675358539765,
|
1159 |
+
"grad_norm": 6.840878009796143,
|
1160 |
+
"learning_rate": 0.00010809973924380703,
|
1161 |
+
"loss": 0.4138,
|
1162 |
+
"step": 78500
|
1163 |
+
},
|
1164 |
+
{
|
1165 |
+
"epoch": 6.4374185136897,
|
1166 |
+
"grad_norm": 4.359363079071045,
|
1167 |
+
"learning_rate": 0.00010687744458930898,
|
1168 |
+
"loss": 0.4031,
|
1169 |
+
"step": 79000
|
1170 |
+
},
|
1171 |
+
{
|
1172 |
+
"epoch": 6.478161668839635,
|
1173 |
+
"grad_norm": 5.549428462982178,
|
1174 |
+
"learning_rate": 0.00010565514993481095,
|
1175 |
+
"loss": 0.4112,
|
1176 |
+
"step": 79500
|
1177 |
+
},
|
1178 |
+
{
|
1179 |
+
"epoch": 6.51890482398957,
|
1180 |
+
"grad_norm": 3.57088041305542,
|
1181 |
+
"learning_rate": 0.0001044328552803129,
|
1182 |
+
"loss": 0.4033,
|
1183 |
+
"step": 80000
|
1184 |
+
},
|
1185 |
+
{
|
1186 |
+
"epoch": 6.559647979139505,
|
1187 |
+
"grad_norm": 4.713502883911133,
|
1188 |
+
"learning_rate": 0.00010321056062581485,
|
1189 |
+
"loss": 0.4129,
|
1190 |
+
"step": 80500
|
1191 |
+
},
|
1192 |
+
{
|
1193 |
+
"epoch": 6.600391134289439,
|
1194 |
+
"grad_norm": 5.755394458770752,
|
1195 |
+
"learning_rate": 0.0001019882659713168,
|
1196 |
+
"loss": 0.4001,
|
1197 |
+
"step": 81000
|
1198 |
+
},
|
1199 |
+
{
|
1200 |
+
"epoch": 6.641134289439374,
|
1201 |
+
"grad_norm": 5.579432010650635,
|
1202 |
+
"learning_rate": 0.00010076597131681877,
|
1203 |
+
"loss": 0.4057,
|
1204 |
+
"step": 81500
|
1205 |
+
},
|
1206 |
+
{
|
1207 |
+
"epoch": 6.681877444589309,
|
1208 |
+
"grad_norm": 5.298727512359619,
|
1209 |
+
"learning_rate": 9.954367666232072e-05,
|
1210 |
+
"loss": 0.4069,
|
1211 |
+
"step": 82000
|
1212 |
+
},
|
1213 |
+
{
|
1214 |
+
"epoch": 6.722620599739244,
|
1215 |
+
"grad_norm": 6.568725109100342,
|
1216 |
+
"learning_rate": 9.832138200782268e-05,
|
1217 |
+
"loss": 0.3974,
|
1218 |
+
"step": 82500
|
1219 |
+
},
|
1220 |
+
{
|
1221 |
+
"epoch": 6.763363754889179,
|
1222 |
+
"grad_norm": 5.902369022369385,
|
1223 |
+
"learning_rate": 9.709908735332463e-05,
|
1224 |
+
"loss": 0.3998,
|
1225 |
+
"step": 83000
|
1226 |
+
},
|
1227 |
+
{
|
1228 |
+
"epoch": 6.804106910039113,
|
1229 |
+
"grad_norm": 5.710369110107422,
|
1230 |
+
"learning_rate": 9.58767926988266e-05,
|
1231 |
+
"loss": 0.4094,
|
1232 |
+
"step": 83500
|
1233 |
+
},
|
1234 |
+
{
|
1235 |
+
"epoch": 6.844850065189048,
|
1236 |
+
"grad_norm": 7.27541446685791,
|
1237 |
+
"learning_rate": 9.465449804432855e-05,
|
1238 |
+
"loss": 0.3889,
|
1239 |
+
"step": 84000
|
1240 |
+
},
|
1241 |
+
{
|
1242 |
+
"epoch": 6.885593220338983,
|
1243 |
+
"grad_norm": 2.905696153640747,
|
1244 |
+
"learning_rate": 9.34322033898305e-05,
|
1245 |
+
"loss": 0.3988,
|
1246 |
+
"step": 84500
|
1247 |
+
},
|
1248 |
+
{
|
1249 |
+
"epoch": 6.926336375488918,
|
1250 |
+
"grad_norm": 4.89232063293457,
|
1251 |
+
"learning_rate": 9.220990873533245e-05,
|
1252 |
+
"loss": 0.3921,
|
1253 |
+
"step": 85000
|
1254 |
+
},
|
1255 |
+
{
|
1256 |
+
"epoch": 6.967079530638853,
|
1257 |
+
"grad_norm": 6.2112274169921875,
|
1258 |
+
"learning_rate": 9.098761408083442e-05,
|
1259 |
+
"loss": 0.3907,
|
1260 |
+
"step": 85500
|
1261 |
+
},
|
1262 |
+
{
|
1263 |
+
"epoch": 7.0,
|
1264 |
+
"eval_accuracy": 0.8321285247802734,
|
1265 |
+
"eval_loss": 0.43350040912628174,
|
1266 |
+
"eval_runtime": 3.5095,
|
1267 |
+
"eval_samples_per_second": 709.513,
|
1268 |
+
"eval_steps_per_second": 88.903,
|
1269 |
+
"step": 85904
|
1270 |
+
},
|
1271 |
+
{
|
1272 |
+
"epoch": 7.0078226857887875,
|
1273 |
+
"grad_norm": 6.440418720245361,
|
1274 |
+
"learning_rate": 8.976531942633637e-05,
|
1275 |
+
"loss": 0.3996,
|
1276 |
+
"step": 86000
|
1277 |
+
},
|
1278 |
+
{
|
1279 |
+
"epoch": 7.048565840938722,
|
1280 |
+
"grad_norm": 5.757605075836182,
|
1281 |
+
"learning_rate": 8.854302477183832e-05,
|
1282 |
+
"loss": 0.377,
|
1283 |
+
"step": 86500
|
1284 |
+
},
|
1285 |
+
{
|
1286 |
+
"epoch": 7.089308996088657,
|
1287 |
+
"grad_norm": 5.203352928161621,
|
1288 |
+
"learning_rate": 8.732073011734028e-05,
|
1289 |
+
"loss": 0.3928,
|
1290 |
+
"step": 87000
|
1291 |
+
},
|
1292 |
+
{
|
1293 |
+
"epoch": 7.130052151238592,
|
1294 |
+
"grad_norm": 5.605312347412109,
|
1295 |
+
"learning_rate": 8.609843546284225e-05,
|
1296 |
+
"loss": 0.3863,
|
1297 |
+
"step": 87500
|
1298 |
+
},
|
1299 |
+
{
|
1300 |
+
"epoch": 7.170795306388527,
|
1301 |
+
"grad_norm": 5.28822660446167,
|
1302 |
+
"learning_rate": 8.48761408083442e-05,
|
1303 |
+
"loss": 0.3843,
|
1304 |
+
"step": 88000
|
1305 |
+
},
|
1306 |
+
{
|
1307 |
+
"epoch": 7.211538461538462,
|
1308 |
+
"grad_norm": 4.207997798919678,
|
1309 |
+
"learning_rate": 8.365384615384615e-05,
|
1310 |
+
"loss": 0.3886,
|
1311 |
+
"step": 88500
|
1312 |
+
},
|
1313 |
+
{
|
1314 |
+
"epoch": 7.252281616688396,
|
1315 |
+
"grad_norm": 4.063730239868164,
|
1316 |
+
"learning_rate": 8.24315514993481e-05,
|
1317 |
+
"loss": 0.3914,
|
1318 |
+
"step": 89000
|
1319 |
+
},
|
1320 |
+
{
|
1321 |
+
"epoch": 7.293024771838331,
|
1322 |
+
"grad_norm": 6.433591365814209,
|
1323 |
+
"learning_rate": 8.120925684485006e-05,
|
1324 |
+
"loss": 0.3926,
|
1325 |
+
"step": 89500
|
1326 |
+
},
|
1327 |
+
{
|
1328 |
+
"epoch": 7.333767926988266,
|
1329 |
+
"grad_norm": 3.1585044860839844,
|
1330 |
+
"learning_rate": 7.998696219035201e-05,
|
1331 |
+
"loss": 0.39,
|
1332 |
+
"step": 90000
|
1333 |
+
},
|
1334 |
+
{
|
1335 |
+
"epoch": 7.374511082138201,
|
1336 |
+
"grad_norm": 5.051187992095947,
|
1337 |
+
"learning_rate": 7.876466753585397e-05,
|
1338 |
+
"loss": 0.3854,
|
1339 |
+
"step": 90500
|
1340 |
+
},
|
1341 |
+
{
|
1342 |
+
"epoch": 7.415254237288136,
|
1343 |
+
"grad_norm": 5.842960834503174,
|
1344 |
+
"learning_rate": 7.754237288135592e-05,
|
1345 |
+
"loss": 0.3748,
|
1346 |
+
"step": 91000
|
1347 |
+
},
|
1348 |
+
{
|
1349 |
+
"epoch": 7.4559973924380705,
|
1350 |
+
"grad_norm": 3.273320198059082,
|
1351 |
+
"learning_rate": 7.63200782268579e-05,
|
1352 |
+
"loss": 0.3885,
|
1353 |
+
"step": 91500
|
1354 |
+
},
|
1355 |
+
{
|
1356 |
+
"epoch": 7.496740547588005,
|
1357 |
+
"grad_norm": 4.630128860473633,
|
1358 |
+
"learning_rate": 7.509778357235985e-05,
|
1359 |
+
"loss": 0.3914,
|
1360 |
+
"step": 92000
|
1361 |
+
},
|
1362 |
+
{
|
1363 |
+
"epoch": 7.53748370273794,
|
1364 |
+
"grad_norm": 4.444061279296875,
|
1365 |
+
"learning_rate": 7.387548891786178e-05,
|
1366 |
+
"loss": 0.3882,
|
1367 |
+
"step": 92500
|
1368 |
+
},
|
1369 |
+
{
|
1370 |
+
"epoch": 7.578226857887875,
|
1371 |
+
"grad_norm": 4.932531833648682,
|
1372 |
+
"learning_rate": 7.265319426336375e-05,
|
1373 |
+
"loss": 0.3886,
|
1374 |
+
"step": 93000
|
1375 |
+
},
|
1376 |
+
{
|
1377 |
+
"epoch": 7.61897001303781,
|
1378 |
+
"grad_norm": 5.399273872375488,
|
1379 |
+
"learning_rate": 7.14308996088657e-05,
|
1380 |
+
"loss": 0.3937,
|
1381 |
+
"step": 93500
|
1382 |
+
},
|
1383 |
+
{
|
1384 |
+
"epoch": 7.659713168187745,
|
1385 |
+
"grad_norm": 7.287408351898193,
|
1386 |
+
"learning_rate": 7.020860495436766e-05,
|
1387 |
+
"loss": 0.3783,
|
1388 |
+
"step": 94000
|
1389 |
+
},
|
1390 |
+
{
|
1391 |
+
"epoch": 7.700456323337679,
|
1392 |
+
"grad_norm": 6.242406368255615,
|
1393 |
+
"learning_rate": 6.898631029986961e-05,
|
1394 |
+
"loss": 0.395,
|
1395 |
+
"step": 94500
|
1396 |
+
},
|
1397 |
+
{
|
1398 |
+
"epoch": 7.741199478487614,
|
1399 |
+
"grad_norm": 3.794053077697754,
|
1400 |
+
"learning_rate": 6.776401564537158e-05,
|
1401 |
+
"loss": 0.3967,
|
1402 |
+
"step": 95000
|
1403 |
+
},
|
1404 |
+
{
|
1405 |
+
"epoch": 7.781942633637549,
|
1406 |
+
"grad_norm": 7.381150245666504,
|
1407 |
+
"learning_rate": 6.654172099087353e-05,
|
1408 |
+
"loss": 0.3837,
|
1409 |
+
"step": 95500
|
1410 |
+
},
|
1411 |
+
{
|
1412 |
+
"epoch": 7.822685788787483,
|
1413 |
+
"grad_norm": 4.904209613800049,
|
1414 |
+
"learning_rate": 6.531942633637548e-05,
|
1415 |
+
"loss": 0.3874,
|
1416 |
+
"step": 96000
|
1417 |
+
},
|
1418 |
+
{
|
1419 |
+
"epoch": 7.863428943937419,
|
1420 |
+
"grad_norm": 4.649083137512207,
|
1421 |
+
"learning_rate": 6.409713168187743e-05,
|
1422 |
+
"loss": 0.389,
|
1423 |
+
"step": 96500
|
1424 |
+
},
|
1425 |
+
{
|
1426 |
+
"epoch": 7.904172099087353,
|
1427 |
+
"grad_norm": 7.420172691345215,
|
1428 |
+
"learning_rate": 6.28748370273794e-05,
|
1429 |
+
"loss": 0.3877,
|
1430 |
+
"step": 97000
|
1431 |
+
},
|
1432 |
+
{
|
1433 |
+
"epoch": 7.944915254237288,
|
1434 |
+
"grad_norm": 4.977739334106445,
|
1435 |
+
"learning_rate": 6.165254237288135e-05,
|
1436 |
+
"loss": 0.3885,
|
1437 |
+
"step": 97500
|
1438 |
+
},
|
1439 |
+
{
|
1440 |
+
"epoch": 7.985658409387223,
|
1441 |
+
"grad_norm": 5.967884063720703,
|
1442 |
+
"learning_rate": 6.0430247718383304e-05,
|
1443 |
+
"loss": 0.3899,
|
1444 |
+
"step": 98000
|
1445 |
+
},
|
1446 |
+
{
|
1447 |
+
"epoch": 8.0,
|
1448 |
+
"eval_accuracy": 0.8401606678962708,
|
1449 |
+
"eval_loss": 0.421245813369751,
|
1450 |
+
"eval_runtime": 3.5351,
|
1451 |
+
"eval_samples_per_second": 704.361,
|
1452 |
+
"eval_steps_per_second": 88.257,
|
1453 |
+
"step": 98176
|
1454 |
+
},
|
1455 |
+
{
|
1456 |
+
"epoch": 8.026401564537158,
|
1457 |
+
"grad_norm": 3.8275697231292725,
|
1458 |
+
"learning_rate": 5.920795306388526e-05,
|
1459 |
+
"loss": 0.3753,
|
1460 |
+
"step": 98500
|
1461 |
+
},
|
1462 |
+
{
|
1463 |
+
"epoch": 8.067144719687093,
|
1464 |
+
"grad_norm": 5.281260967254639,
|
1465 |
+
"learning_rate": 5.798565840938721e-05,
|
1466 |
+
"loss": 0.3657,
|
1467 |
+
"step": 99000
|
1468 |
+
},
|
1469 |
+
{
|
1470 |
+
"epoch": 8.107887874837028,
|
1471 |
+
"grad_norm": 4.99520206451416,
|
1472 |
+
"learning_rate": 5.676336375488918e-05,
|
1473 |
+
"loss": 0.365,
|
1474 |
+
"step": 99500
|
1475 |
+
},
|
1476 |
+
{
|
1477 |
+
"epoch": 8.148631029986962,
|
1478 |
+
"grad_norm": 8.345664978027344,
|
1479 |
+
"learning_rate": 5.554106910039113e-05,
|
1480 |
+
"loss": 0.3766,
|
1481 |
+
"step": 100000
|
1482 |
+
},
|
1483 |
+
{
|
1484 |
+
"epoch": 8.189374185136897,
|
1485 |
+
"grad_norm": 7.439541339874268,
|
1486 |
+
"learning_rate": 5.4318774445893086e-05,
|
1487 |
+
"loss": 0.3695,
|
1488 |
+
"step": 100500
|
1489 |
+
},
|
1490 |
+
{
|
1491 |
+
"epoch": 8.230117340286832,
|
1492 |
+
"grad_norm": 18.46923828125,
|
1493 |
+
"learning_rate": 5.309647979139504e-05,
|
1494 |
+
"loss": 0.3751,
|
1495 |
+
"step": 101000
|
1496 |
+
},
|
1497 |
+
{
|
1498 |
+
"epoch": 8.270860495436766,
|
1499 |
+
"grad_norm": 7.099632740020752,
|
1500 |
+
"learning_rate": 5.1874185136897e-05,
|
1501 |
+
"loss": 0.3747,
|
1502 |
+
"step": 101500
|
1503 |
+
},
|
1504 |
+
{
|
1505 |
+
"epoch": 8.3116036505867,
|
1506 |
+
"grad_norm": 5.036531925201416,
|
1507 |
+
"learning_rate": 5.065189048239895e-05,
|
1508 |
+
"loss": 0.3736,
|
1509 |
+
"step": 102000
|
1510 |
+
},
|
1511 |
+
{
|
1512 |
+
"epoch": 8.352346805736635,
|
1513 |
+
"grad_norm": 5.257627010345459,
|
1514 |
+
"learning_rate": 4.942959582790091e-05,
|
1515 |
+
"loss": 0.3718,
|
1516 |
+
"step": 102500
|
1517 |
+
},
|
1518 |
+
{
|
1519 |
+
"epoch": 8.393089960886572,
|
1520 |
+
"grad_norm": 4.144084453582764,
|
1521 |
+
"learning_rate": 4.820730117340286e-05,
|
1522 |
+
"loss": 0.3831,
|
1523 |
+
"step": 103000
|
1524 |
+
},
|
1525 |
+
{
|
1526 |
+
"epoch": 8.433833116036507,
|
1527 |
+
"grad_norm": 7.010155200958252,
|
1528 |
+
"learning_rate": 4.698500651890482e-05,
|
1529 |
+
"loss": 0.3728,
|
1530 |
+
"step": 103500
|
1531 |
+
},
|
1532 |
+
{
|
1533 |
+
"epoch": 8.474576271186441,
|
1534 |
+
"grad_norm": 4.621219635009766,
|
1535 |
+
"learning_rate": 4.576271186440678e-05,
|
1536 |
+
"loss": 0.3708,
|
1537 |
+
"step": 104000
|
1538 |
+
},
|
1539 |
+
{
|
1540 |
+
"epoch": 8.515319426336376,
|
1541 |
+
"grad_norm": 7.2299394607543945,
|
1542 |
+
"learning_rate": 4.4540417209908735e-05,
|
1543 |
+
"loss": 0.3747,
|
1544 |
+
"step": 104500
|
1545 |
+
},
|
1546 |
+
{
|
1547 |
+
"epoch": 8.55606258148631,
|
1548 |
+
"grad_norm": 7.200563430786133,
|
1549 |
+
"learning_rate": 4.3318122555410686e-05,
|
1550 |
+
"loss": 0.3817,
|
1551 |
+
"step": 105000
|
1552 |
+
},
|
1553 |
+
{
|
1554 |
+
"epoch": 8.596805736636245,
|
1555 |
+
"grad_norm": 4.539661884307861,
|
1556 |
+
"learning_rate": 4.2095827900912643e-05,
|
1557 |
+
"loss": 0.375,
|
1558 |
+
"step": 105500
|
1559 |
+
},
|
1560 |
+
{
|
1561 |
+
"epoch": 8.63754889178618,
|
1562 |
+
"grad_norm": 4.076575756072998,
|
1563 |
+
"learning_rate": 4.08735332464146e-05,
|
1564 |
+
"loss": 0.3761,
|
1565 |
+
"step": 106000
|
1566 |
+
},
|
1567 |
+
{
|
1568 |
+
"epoch": 8.678292046936114,
|
1569 |
+
"grad_norm": 5.755791664123535,
|
1570 |
+
"learning_rate": 3.965123859191656e-05,
|
1571 |
+
"loss": 0.388,
|
1572 |
+
"step": 106500
|
1573 |
+
},
|
1574 |
+
{
|
1575 |
+
"epoch": 8.719035202086049,
|
1576 |
+
"grad_norm": 4.549001216888428,
|
1577 |
+
"learning_rate": 3.842894393741851e-05,
|
1578 |
+
"loss": 0.3794,
|
1579 |
+
"step": 107000
|
1580 |
+
},
|
1581 |
+
{
|
1582 |
+
"epoch": 8.759778357235984,
|
1583 |
+
"grad_norm": 4.821287155151367,
|
1584 |
+
"learning_rate": 3.720664928292047e-05,
|
1585 |
+
"loss": 0.3696,
|
1586 |
+
"step": 107500
|
1587 |
+
},
|
1588 |
+
{
|
1589 |
+
"epoch": 8.800521512385918,
|
1590 |
+
"grad_norm": 5.3173089027404785,
|
1591 |
+
"learning_rate": 3.5984354628422425e-05,
|
1592 |
+
"loss": 0.3777,
|
1593 |
+
"step": 108000
|
1594 |
+
},
|
1595 |
+
{
|
1596 |
+
"epoch": 8.841264667535853,
|
1597 |
+
"grad_norm": 5.278952121734619,
|
1598 |
+
"learning_rate": 3.4762059973924376e-05,
|
1599 |
+
"loss": 0.3743,
|
1600 |
+
"step": 108500
|
1601 |
+
},
|
1602 |
+
{
|
1603 |
+
"epoch": 8.88200782268579,
|
1604 |
+
"grad_norm": 6.104799270629883,
|
1605 |
+
"learning_rate": 3.3539765319426334e-05,
|
1606 |
+
"loss": 0.3791,
|
1607 |
+
"step": 109000
|
1608 |
+
},
|
1609 |
+
{
|
1610 |
+
"epoch": 8.922750977835724,
|
1611 |
+
"grad_norm": 4.673792839050293,
|
1612 |
+
"learning_rate": 3.2317470664928285e-05,
|
1613 |
+
"loss": 0.3774,
|
1614 |
+
"step": 109500
|
1615 |
+
},
|
1616 |
+
{
|
1617 |
+
"epoch": 8.963494132985659,
|
1618 |
+
"grad_norm": 6.051920413970947,
|
1619 |
+
"learning_rate": 3.109517601043025e-05,
|
1620 |
+
"loss": 0.3752,
|
1621 |
+
"step": 110000
|
1622 |
+
},
|
1623 |
+
{
|
1624 |
+
"epoch": 9.0,
|
1625 |
+
"eval_accuracy": 0.8373494148254395,
|
1626 |
+
"eval_loss": 0.43148961663246155,
|
1627 |
+
"eval_runtime": 3.528,
|
1628 |
+
"eval_samples_per_second": 705.779,
|
1629 |
+
"eval_steps_per_second": 88.435,
|
1630 |
+
"step": 110448
|
1631 |
+
},
|
1632 |
+
{
|
1633 |
+
"epoch": 9.004237288135593,
|
1634 |
+
"grad_norm": 3.2586278915405273,
|
1635 |
+
"learning_rate": 2.98728813559322e-05,
|
1636 |
+
"loss": 0.3684,
|
1637 |
+
"step": 110500
|
1638 |
+
},
|
1639 |
+
{
|
1640 |
+
"epoch": 9.044980443285528,
|
1641 |
+
"grad_norm": 3.5761301517486572,
|
1642 |
+
"learning_rate": 2.8650586701434158e-05,
|
1643 |
+
"loss": 0.3634,
|
1644 |
+
"step": 111000
|
1645 |
+
},
|
1646 |
+
{
|
1647 |
+
"epoch": 9.085723598435463,
|
1648 |
+
"grad_norm": 8.008382797241211,
|
1649 |
+
"learning_rate": 2.7428292046936113e-05,
|
1650 |
+
"loss": 0.3574,
|
1651 |
+
"step": 111500
|
1652 |
+
},
|
1653 |
+
{
|
1654 |
+
"epoch": 9.126466753585397,
|
1655 |
+
"grad_norm": 3.5123190879821777,
|
1656 |
+
"learning_rate": 2.6205997392438067e-05,
|
1657 |
+
"loss": 0.3794,
|
1658 |
+
"step": 112000
|
1659 |
+
},
|
1660 |
+
{
|
1661 |
+
"epoch": 9.167209908735332,
|
1662 |
+
"grad_norm": 6.476950168609619,
|
1663 |
+
"learning_rate": 2.4983702737940025e-05,
|
1664 |
+
"loss": 0.3677,
|
1665 |
+
"step": 112500
|
1666 |
+
},
|
1667 |
+
{
|
1668 |
+
"epoch": 9.207953063885267,
|
1669 |
+
"grad_norm": 3.6542065143585205,
|
1670 |
+
"learning_rate": 2.376140808344198e-05,
|
1671 |
+
"loss": 0.3666,
|
1672 |
+
"step": 113000
|
1673 |
+
},
|
1674 |
+
{
|
1675 |
+
"epoch": 9.248696219035201,
|
1676 |
+
"grad_norm": 5.919882297515869,
|
1677 |
+
"learning_rate": 2.2539113428943937e-05,
|
1678 |
+
"loss": 0.3605,
|
1679 |
+
"step": 113500
|
1680 |
+
},
|
1681 |
+
{
|
1682 |
+
"epoch": 9.289439374185136,
|
1683 |
+
"grad_norm": 10.532210350036621,
|
1684 |
+
"learning_rate": 2.131681877444589e-05,
|
1685 |
+
"loss": 0.3703,
|
1686 |
+
"step": 114000
|
1687 |
+
},
|
1688 |
+
{
|
1689 |
+
"epoch": 9.330182529335072,
|
1690 |
+
"grad_norm": 4.157111167907715,
|
1691 |
+
"learning_rate": 2.009452411994785e-05,
|
1692 |
+
"loss": 0.3712,
|
1693 |
+
"step": 114500
|
1694 |
+
},
|
1695 |
+
{
|
1696 |
+
"epoch": 9.370925684485007,
|
1697 |
+
"grad_norm": 4.405088424682617,
|
1698 |
+
"learning_rate": 1.8872229465449803e-05,
|
1699 |
+
"loss": 0.3595,
|
1700 |
+
"step": 115000
|
1701 |
+
},
|
1702 |
+
{
|
1703 |
+
"epoch": 9.411668839634942,
|
1704 |
+
"grad_norm": 7.088386535644531,
|
1705 |
+
"learning_rate": 1.7649934810951758e-05,
|
1706 |
+
"loss": 0.3637,
|
1707 |
+
"step": 115500
|
1708 |
+
},
|
1709 |
+
{
|
1710 |
+
"epoch": 9.452411994784876,
|
1711 |
+
"grad_norm": 3.80808162689209,
|
1712 |
+
"learning_rate": 1.6427640156453715e-05,
|
1713 |
+
"loss": 0.3662,
|
1714 |
+
"step": 116000
|
1715 |
+
},
|
1716 |
+
{
|
1717 |
+
"epoch": 9.493155149934811,
|
1718 |
+
"grad_norm": 5.665896415710449,
|
1719 |
+
"learning_rate": 1.520534550195567e-05,
|
1720 |
+
"loss": 0.369,
|
1721 |
+
"step": 116500
|
1722 |
+
},
|
1723 |
+
{
|
1724 |
+
"epoch": 9.533898305084746,
|
1725 |
+
"grad_norm": 4.733582496643066,
|
1726 |
+
"learning_rate": 1.3983050847457626e-05,
|
1727 |
+
"loss": 0.3665,
|
1728 |
+
"step": 117000
|
1729 |
+
},
|
1730 |
+
{
|
1731 |
+
"epoch": 9.57464146023468,
|
1732 |
+
"grad_norm": 6.934778690338135,
|
1733 |
+
"learning_rate": 1.2760756192959582e-05,
|
1734 |
+
"loss": 0.3567,
|
1735 |
+
"step": 117500
|
1736 |
+
},
|
1737 |
+
{
|
1738 |
+
"epoch": 9.615384615384615,
|
1739 |
+
"grad_norm": 6.5998311042785645,
|
1740 |
+
"learning_rate": 1.1538461538461538e-05,
|
1741 |
+
"loss": 0.3612,
|
1742 |
+
"step": 118000
|
1743 |
+
},
|
1744 |
+
{
|
1745 |
+
"epoch": 9.65612777053455,
|
1746 |
+
"grad_norm": 7.71038293838501,
|
1747 |
+
"learning_rate": 1.0316166883963494e-05,
|
1748 |
+
"loss": 0.3562,
|
1749 |
+
"step": 118500
|
1750 |
+
},
|
1751 |
+
{
|
1752 |
+
"epoch": 9.696870925684484,
|
1753 |
+
"grad_norm": 4.586871147155762,
|
1754 |
+
"learning_rate": 9.093872229465448e-06,
|
1755 |
+
"loss": 0.3661,
|
1756 |
+
"step": 119000
|
1757 |
+
},
|
1758 |
+
{
|
1759 |
+
"epoch": 9.737614080834419,
|
1760 |
+
"grad_norm": 5.962682247161865,
|
1761 |
+
"learning_rate": 7.871577574967404e-06,
|
1762 |
+
"loss": 0.3675,
|
1763 |
+
"step": 119500
|
1764 |
+
},
|
1765 |
+
{
|
1766 |
+
"epoch": 9.778357235984355,
|
1767 |
+
"grad_norm": 3.3349926471710205,
|
1768 |
+
"learning_rate": 6.649282920469361e-06,
|
1769 |
+
"loss": 0.3571,
|
1770 |
+
"step": 120000
|
1771 |
+
},
|
1772 |
+
{
|
1773 |
+
"epoch": 9.81910039113429,
|
1774 |
+
"grad_norm": 5.066919326782227,
|
1775 |
+
"learning_rate": 5.426988265971316e-06,
|
1776 |
+
"loss": 0.3558,
|
1777 |
+
"step": 120500
|
1778 |
+
},
|
1779 |
+
{
|
1780 |
+
"epoch": 9.859843546284225,
|
1781 |
+
"grad_norm": 5.792361259460449,
|
1782 |
+
"learning_rate": 4.2046936114732716e-06,
|
1783 |
+
"loss": 0.368,
|
1784 |
+
"step": 121000
|
1785 |
+
},
|
1786 |
+
{
|
1787 |
+
"epoch": 9.90058670143416,
|
1788 |
+
"grad_norm": 6.927473545074463,
|
1789 |
+
"learning_rate": 2.982398956975228e-06,
|
1790 |
+
"loss": 0.3631,
|
1791 |
+
"step": 121500
|
1792 |
+
},
|
1793 |
+
{
|
1794 |
+
"epoch": 9.941329856584094,
|
1795 |
+
"grad_norm": 6.935751914978027,
|
1796 |
+
"learning_rate": 1.7601043024771837e-06,
|
1797 |
+
"loss": 0.3623,
|
1798 |
+
"step": 122000
|
1799 |
+
},
|
1800 |
+
{
|
1801 |
+
"epoch": 9.982073011734029,
|
1802 |
+
"grad_norm": 1.9150322675704956,
|
1803 |
+
"learning_rate": 5.378096479791394e-07,
|
1804 |
+
"loss": 0.3639,
|
1805 |
+
"step": 122500
|
1806 |
+
},
|
1807 |
+
{
|
1808 |
+
"epoch": 10.0,
|
1809 |
+
"eval_accuracy": 0.8357429504394531,
|
1810 |
+
"eval_loss": 0.4303068220615387,
|
1811 |
+
"eval_runtime": 3.5313,
|
1812 |
+
"eval_samples_per_second": 705.119,
|
1813 |
+
"eval_steps_per_second": 88.352,
|
1814 |
+
"step": 122720
|
1815 |
+
},
|
1816 |
+
{
|
1817 |
+
"epoch": 10.0,
|
1818 |
+
"step": 122720,
|
1819 |
+
"total_flos": 2.6187994846190592e+17,
|
1820 |
+
"train_loss": 0.4432659422714054,
|
1821 |
+
"train_runtime": 8098.9158,
|
1822 |
+
"train_samples_per_second": 484.882,
|
1823 |
+
"train_steps_per_second": 15.153
|
1824 |
}
|
1825 |
],
|
1826 |
"logging_steps": 500,
|
1827 |
+
"max_steps": 122720,
|
1828 |
"num_input_tokens_seen": 0,
|
1829 |
+
"num_train_epochs": 10,
|
1830 |
"save_steps": 500,
|
1831 |
"stateful_callbacks": {
|
1832 |
"TrainerControl": {
|
|
|
1840 |
"attributes": {}
|
1841 |
}
|
1842 |
},
|
1843 |
+
"total_flos": 2.6187994846190592e+17,
|
1844 |
"train_batch_size": 32,
|
1845 |
"trial_name": null,
|
1846 |
"trial_params": null
|