Training in progress, step 40000, checkpoint
Browse files
last-checkpoint/config.json
CHANGED
@@ -1,5 +1,5 @@
|
|
1 |
{
|
2 |
-
"_name_or_path": "
|
3 |
"architectures": [
|
4 |
"EsmForMaskedLM"
|
5 |
],
|
|
|
1 |
{
|
2 |
+
"_name_or_path": "mgh6/TCS_MLM_All",
|
3 |
"architectures": [
|
4 |
"EsmForMaskedLM"
|
5 |
],
|
last-checkpoint/model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 136000488
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:a68617a5515a903c8c3a3e687c082b3bcc6e64c47b4742c75f65f8881ef02a01
|
3 |
size 136000488
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 268176506
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:161596e82e972e9e4adc90db74dd2d8cb3ab034415c2d88d5245811053e257dd
|
3 |
size 268176506
|
last-checkpoint/rng_state.pth
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:9c4cccf1e358d22fd7f60837b2b16f7000af9c1649ed52f7f376f5fa52b81dfb
|
3 |
+
size 15006
|
last-checkpoint/scheduler.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 1064
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:cc3016eadb91b8fef327a5449310066af9d30d076742d39ccdfdc21dda726ef3
|
3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
@@ -1,886 +1,298 @@
|
|
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": 0.24087679152113695,
|
13 |
-
"grad_norm": 0.
|
14 |
-
"learning_rate":
|
15 |
-
"loss":
|
16 |
"step": 1000
|
17 |
},
|
18 |
{
|
19 |
"epoch": 0.4817535830422739,
|
20 |
-
"grad_norm": 0.
|
21 |
-
"learning_rate":
|
22 |
-
"loss":
|
23 |
"step": 2000
|
24 |
},
|
25 |
{
|
26 |
"epoch": 0.7226303745634108,
|
27 |
-
"grad_norm": 0.
|
28 |
-
"learning_rate":
|
29 |
-
"loss":
|
30 |
"step": 3000
|
31 |
},
|
32 |
{
|
33 |
"epoch": 0.9635071660845478,
|
34 |
-
"grad_norm": 0.
|
35 |
-
"learning_rate":
|
36 |
-
"loss":
|
37 |
"step": 4000
|
38 |
},
|
39 |
{
|
40 |
"epoch": 1.2043839576056847,
|
41 |
-
"grad_norm": 0.
|
42 |
-
"learning_rate":
|
43 |
-
"loss":
|
44 |
"step": 5000
|
45 |
},
|
46 |
{
|
47 |
"epoch": 1.4452607491268217,
|
48 |
-
"grad_norm": 0.
|
49 |
-
"learning_rate":
|
50 |
-
"loss":
|
51 |
"step": 6000
|
52 |
},
|
53 |
{
|
54 |
"epoch": 1.6861375406479586,
|
55 |
-
"grad_norm": 0.
|
56 |
-
"learning_rate":
|
57 |
-
"loss":
|
58 |
"step": 7000
|
59 |
},
|
60 |
{
|
61 |
"epoch": 1.9270143321690956,
|
62 |
-
"grad_norm": 0.
|
63 |
-
"learning_rate":
|
64 |
-
"loss":
|
65 |
"step": 8000
|
66 |
},
|
67 |
{
|
68 |
"epoch": 2.1678911236902323,
|
69 |
-
"grad_norm": 0.
|
70 |
-
"learning_rate":
|
71 |
-
"loss":
|
72 |
"step": 9000
|
73 |
},
|
74 |
{
|
75 |
"epoch": 2.4087679152113695,
|
76 |
-
"grad_norm": 0.
|
77 |
-
"learning_rate":
|
78 |
-
"loss":
|
79 |
"step": 10000
|
80 |
},
|
81 |
{
|
82 |
"epoch": 2.649644706732506,
|
83 |
-
"grad_norm": 0.
|
84 |
-
"learning_rate":
|
85 |
-
"loss":
|
86 |
"step": 11000
|
87 |
},
|
88 |
{
|
89 |
"epoch": 2.8905214982536434,
|
90 |
-
"grad_norm": 0.
|
91 |
-
"learning_rate":
|
92 |
-
"loss":
|
93 |
"step": 12000
|
94 |
},
|
95 |
{
|
96 |
"epoch": 3.13139828977478,
|
97 |
-
"grad_norm": 0.
|
98 |
-
"learning_rate":
|
99 |
-
"loss":
|
100 |
"step": 13000
|
101 |
},
|
102 |
{
|
103 |
"epoch": 3.3722750812959172,
|
104 |
-
"grad_norm": 0.
|
105 |
-
"learning_rate":
|
106 |
-
"loss":
|
107 |
"step": 14000
|
108 |
},
|
109 |
{
|
110 |
"epoch": 3.613151872817054,
|
111 |
-
"grad_norm": 0.
|
112 |
-
"learning_rate":
|
113 |
-
"loss":
|
114 |
"step": 15000
|
115 |
},
|
116 |
{
|
117 |
"epoch": 3.854028664338191,
|
118 |
-
"grad_norm": 0.
|
119 |
-
"learning_rate":
|
120 |
-
"loss":
|
121 |
"step": 16000
|
122 |
},
|
123 |
{
|
124 |
"epoch": 4.094905455859328,
|
125 |
-
"grad_norm": 0.
|
126 |
-
"learning_rate":
|
127 |
-
"loss":
|
128 |
"step": 17000
|
129 |
},
|
130 |
{
|
131 |
"epoch": 4.335782247380465,
|
132 |
-
"grad_norm": 0.
|
133 |
-
"learning_rate":
|
134 |
-
"loss":
|
135 |
"step": 18000
|
136 |
},
|
137 |
{
|
138 |
"epoch": 4.576659038901602,
|
139 |
-
"grad_norm": 0.
|
140 |
-
"learning_rate":
|
141 |
-
"loss":
|
142 |
"step": 19000
|
143 |
},
|
144 |
{
|
145 |
"epoch": 4.817535830422739,
|
146 |
-
"grad_norm": 0.
|
147 |
-
"learning_rate":
|
148 |
-
"loss":
|
149 |
"step": 20000
|
150 |
},
|
151 |
{
|
152 |
"epoch": 5.058412621943876,
|
153 |
-
"grad_norm": 0.
|
154 |
-
"learning_rate":
|
155 |
-
"loss":
|
156 |
"step": 21000
|
157 |
},
|
158 |
{
|
159 |
"epoch": 5.299289413465012,
|
160 |
-
"grad_norm": 0.
|
161 |
-
"learning_rate":
|
162 |
-
"loss":
|
163 |
"step": 22000
|
164 |
},
|
165 |
{
|
166 |
"epoch": 5.54016620498615,
|
167 |
-
"grad_norm": 0.
|
168 |
-
"learning_rate":
|
169 |
-
"loss":
|
170 |
"step": 23000
|
171 |
},
|
172 |
{
|
173 |
"epoch": 5.781042996507287,
|
174 |
-
"grad_norm": 0.
|
175 |
-
"learning_rate":
|
176 |
-
"loss":
|
177 |
"step": 24000
|
178 |
},
|
179 |
{
|
180 |
"epoch": 6.021919788028423,
|
181 |
-
"grad_norm": 0.
|
182 |
-
"learning_rate":
|
183 |
-
"loss":
|
184 |
"step": 25000
|
185 |
},
|
186 |
{
|
187 |
"epoch": 6.26279657954956,
|
188 |
-
"grad_norm": 0.
|
189 |
-
"learning_rate":
|
190 |
-
"loss":
|
191 |
"step": 26000
|
192 |
},
|
193 |
{
|
194 |
"epoch": 6.503673371070697,
|
195 |
-
"grad_norm": 0.
|
196 |
-
"learning_rate":
|
197 |
-
"loss":
|
198 |
"step": 27000
|
199 |
},
|
200 |
{
|
201 |
"epoch": 6.7445501625918345,
|
202 |
-
"grad_norm": 0.
|
203 |
-
"learning_rate":
|
204 |
-
"loss":
|
205 |
"step": 28000
|
206 |
},
|
207 |
{
|
208 |
"epoch": 6.985426954112971,
|
209 |
-
"grad_norm": 0.
|
210 |
-
"learning_rate":
|
211 |
-
"loss":
|
212 |
"step": 29000
|
213 |
},
|
214 |
{
|
215 |
"epoch": 7.226303745634108,
|
216 |
-
"grad_norm": 0.
|
217 |
-
"learning_rate":
|
218 |
-
"loss":
|
219 |
"step": 30000
|
220 |
},
|
221 |
{
|
222 |
"epoch": 7.4671805371552455,
|
223 |
-
"grad_norm": 0.
|
224 |
-
"learning_rate":
|
225 |
-
"loss":
|
226 |
"step": 31000
|
227 |
},
|
228 |
{
|
229 |
"epoch": 7.708057328676382,
|
230 |
-
"grad_norm": 0.
|
231 |
-
"learning_rate":
|
232 |
-
"loss": 0.
|
233 |
"step": 32000
|
234 |
},
|
235 |
{
|
236 |
"epoch": 7.948934120197519,
|
237 |
-
"grad_norm": 0.
|
238 |
-
"learning_rate":
|
239 |
-
"loss": 0.
|
240 |
"step": 33000
|
241 |
},
|
242 |
{
|
243 |
"epoch": 8.189810911718656,
|
244 |
-
"grad_norm": 0.
|
245 |
-
"learning_rate":
|
246 |
-
"loss": 0.
|
247 |
"step": 34000
|
248 |
},
|
249 |
{
|
250 |
"epoch": 8.430687703239792,
|
251 |
-
"grad_norm": 0.
|
252 |
-
"learning_rate":
|
253 |
-
"loss": 0.
|
254 |
"step": 35000
|
255 |
},
|
256 |
{
|
257 |
"epoch": 8.67156449476093,
|
258 |
-
"grad_norm": 0.
|
259 |
-
"learning_rate":
|
260 |
-
"loss": 0.
|
261 |
"step": 36000
|
262 |
},
|
263 |
{
|
264 |
"epoch": 8.912441286282068,
|
265 |
-
"grad_norm": 0.
|
266 |
-
"learning_rate":
|
267 |
-
"loss": 0.
|
268 |
"step": 37000
|
269 |
},
|
270 |
{
|
271 |
"epoch": 9.153318077803204,
|
272 |
-
"grad_norm": 0.
|
273 |
-
"learning_rate":
|
274 |
-
"loss": 0.
|
275 |
"step": 38000
|
276 |
},
|
277 |
{
|
278 |
"epoch": 9.394194869324341,
|
279 |
-
"grad_norm": 0.
|
280 |
-
"learning_rate":
|
281 |
-
"loss": 0.
|
282 |
"step": 39000
|
283 |
},
|
284 |
{
|
285 |
"epoch": 9.635071660845478,
|
286 |
-
"grad_norm": 0.
|
287 |
-
"learning_rate":
|
288 |
-
"loss": 0.
|
289 |
"step": 40000
|
290 |
-
},
|
291 |
-
{
|
292 |
-
"epoch": 9.875948452366615,
|
293 |
-
"grad_norm": 0.5080583691596985,
|
294 |
-
"learning_rate": 0.0006707620653657753,
|
295 |
-
"loss": 0.9687,
|
296 |
-
"step": 41000
|
297 |
-
},
|
298 |
-
{
|
299 |
-
"epoch": 10.116825243887751,
|
300 |
-
"grad_norm": 0.49223220348358154,
|
301 |
-
"learning_rate": 0.0006627318718381113,
|
302 |
-
"loss": 0.9592,
|
303 |
-
"step": 42000
|
304 |
-
},
|
305 |
-
{
|
306 |
-
"epoch": 10.357702035408888,
|
307 |
-
"grad_norm": 0.5603600740432739,
|
308 |
-
"learning_rate": 0.0006547016783104472,
|
309 |
-
"loss": 0.9559,
|
310 |
-
"step": 43000
|
311 |
-
},
|
312 |
-
{
|
313 |
-
"epoch": 10.598578826930025,
|
314 |
-
"grad_norm": 0.503847599029541,
|
315 |
-
"learning_rate": 0.0006466714847827833,
|
316 |
-
"loss": 0.9533,
|
317 |
-
"step": 44000
|
318 |
-
},
|
319 |
-
{
|
320 |
-
"epoch": 10.839455618451161,
|
321 |
-
"grad_norm": 0.4978269934654236,
|
322 |
-
"learning_rate": 0.0006386412912551193,
|
323 |
-
"loss": 0.9486,
|
324 |
-
"step": 45000
|
325 |
-
},
|
326 |
-
{
|
327 |
-
"epoch": 11.0803324099723,
|
328 |
-
"grad_norm": 0.5506151914596558,
|
329 |
-
"learning_rate": 0.0006306110977274552,
|
330 |
-
"loss": 0.951,
|
331 |
-
"step": 46000
|
332 |
-
},
|
333 |
-
{
|
334 |
-
"epoch": 11.321209201493437,
|
335 |
-
"grad_norm": 0.5171232223510742,
|
336 |
-
"learning_rate": 0.0006225809041997912,
|
337 |
-
"loss": 0.9399,
|
338 |
-
"step": 47000
|
339 |
-
},
|
340 |
-
{
|
341 |
-
"epoch": 11.562085993014573,
|
342 |
-
"grad_norm": 0.656745970249176,
|
343 |
-
"learning_rate": 0.0006145507106721272,
|
344 |
-
"loss": 0.937,
|
345 |
-
"step": 48000
|
346 |
-
},
|
347 |
-
{
|
348 |
-
"epoch": 11.80296278453571,
|
349 |
-
"grad_norm": 0.5088077783584595,
|
350 |
-
"learning_rate": 0.0006065205171444633,
|
351 |
-
"loss": 0.9376,
|
352 |
-
"step": 49000
|
353 |
-
},
|
354 |
-
{
|
355 |
-
"epoch": 12.043839576056847,
|
356 |
-
"grad_norm": 0.4850046932697296,
|
357 |
-
"learning_rate": 0.0005984903236167992,
|
358 |
-
"loss": 0.9302,
|
359 |
-
"step": 50000
|
360 |
-
},
|
361 |
-
{
|
362 |
-
"epoch": 12.284716367577984,
|
363 |
-
"grad_norm": 0.5553488731384277,
|
364 |
-
"learning_rate": 0.0005904601300891351,
|
365 |
-
"loss": 0.9302,
|
366 |
-
"step": 51000
|
367 |
-
},
|
368 |
-
{
|
369 |
-
"epoch": 12.52559315909912,
|
370 |
-
"grad_norm": 0.47318387031555176,
|
371 |
-
"learning_rate": 0.0005824299365614711,
|
372 |
-
"loss": 0.9298,
|
373 |
-
"step": 52000
|
374 |
-
},
|
375 |
-
{
|
376 |
-
"epoch": 12.766469950620257,
|
377 |
-
"grad_norm": 0.6019132733345032,
|
378 |
-
"learning_rate": 0.0005743997430338072,
|
379 |
-
"loss": 0.9241,
|
380 |
-
"step": 53000
|
381 |
-
},
|
382 |
-
{
|
383 |
-
"epoch": 13.007346742141396,
|
384 |
-
"grad_norm": 0.49550944566726685,
|
385 |
-
"learning_rate": 0.000566369549506143,
|
386 |
-
"loss": 0.9236,
|
387 |
-
"step": 54000
|
388 |
-
},
|
389 |
-
{
|
390 |
-
"epoch": 13.248223533662532,
|
391 |
-
"grad_norm": 0.5007643103599548,
|
392 |
-
"learning_rate": 0.000558339355978479,
|
393 |
-
"loss": 0.9196,
|
394 |
-
"step": 55000
|
395 |
-
},
|
396 |
-
{
|
397 |
-
"epoch": 13.489100325183669,
|
398 |
-
"grad_norm": 0.5525193810462952,
|
399 |
-
"learning_rate": 0.0005503091624508151,
|
400 |
-
"loss": 0.9113,
|
401 |
-
"step": 56000
|
402 |
-
},
|
403 |
-
{
|
404 |
-
"epoch": 13.729977116704806,
|
405 |
-
"grad_norm": 0.5109050273895264,
|
406 |
-
"learning_rate": 0.0005422789689231511,
|
407 |
-
"loss": 0.9077,
|
408 |
-
"step": 57000
|
409 |
-
},
|
410 |
-
{
|
411 |
-
"epoch": 13.970853908225942,
|
412 |
-
"grad_norm": 0.5197868347167969,
|
413 |
-
"learning_rate": 0.000534248775395487,
|
414 |
-
"loss": 0.9076,
|
415 |
-
"step": 58000
|
416 |
-
},
|
417 |
-
{
|
418 |
-
"epoch": 14.21173069974708,
|
419 |
-
"grad_norm": 0.4690844416618347,
|
420 |
-
"learning_rate": 0.000526218581867823,
|
421 |
-
"loss": 0.9097,
|
422 |
-
"step": 59000
|
423 |
-
},
|
424 |
-
{
|
425 |
-
"epoch": 14.452607491268216,
|
426 |
-
"grad_norm": 0.5010888576507568,
|
427 |
-
"learning_rate": 0.000518188388340159,
|
428 |
-
"loss": 0.9022,
|
429 |
-
"step": 60000
|
430 |
-
},
|
431 |
-
{
|
432 |
-
"epoch": 14.693484282789353,
|
433 |
-
"grad_norm": 0.5394883751869202,
|
434 |
-
"learning_rate": 0.000510158194812495,
|
435 |
-
"loss": 0.8982,
|
436 |
-
"step": 61000
|
437 |
-
},
|
438 |
-
{
|
439 |
-
"epoch": 14.934361074310491,
|
440 |
-
"grad_norm": 0.5398752689361572,
|
441 |
-
"learning_rate": 0.000502128001284831,
|
442 |
-
"loss": 0.8982,
|
443 |
-
"step": 62000
|
444 |
-
},
|
445 |
-
{
|
446 |
-
"epoch": 15.175237865831628,
|
447 |
-
"grad_norm": 0.48452773690223694,
|
448 |
-
"learning_rate": 0.0004940978077571669,
|
449 |
-
"loss": 0.8937,
|
450 |
-
"step": 63000
|
451 |
-
},
|
452 |
-
{
|
453 |
-
"epoch": 15.416114657352765,
|
454 |
-
"grad_norm": 0.5147862434387207,
|
455 |
-
"learning_rate": 0.00048606761422950295,
|
456 |
-
"loss": 0.894,
|
457 |
-
"step": 64000
|
458 |
-
},
|
459 |
-
{
|
460 |
-
"epoch": 15.656991448873901,
|
461 |
-
"grad_norm": 0.5301661491394043,
|
462 |
-
"learning_rate": 0.00047803742070183893,
|
463 |
-
"loss": 0.8865,
|
464 |
-
"step": 65000
|
465 |
-
},
|
466 |
-
{
|
467 |
-
"epoch": 15.897868240395038,
|
468 |
-
"grad_norm": 0.49967125058174133,
|
469 |
-
"learning_rate": 0.0004700072271741749,
|
470 |
-
"loss": 0.8825,
|
471 |
-
"step": 66000
|
472 |
-
},
|
473 |
-
{
|
474 |
-
"epoch": 16.138745031916176,
|
475 |
-
"grad_norm": 0.4977249801158905,
|
476 |
-
"learning_rate": 0.0004619770336465109,
|
477 |
-
"loss": 0.884,
|
478 |
-
"step": 67000
|
479 |
-
},
|
480 |
-
{
|
481 |
-
"epoch": 16.37962182343731,
|
482 |
-
"grad_norm": 0.5272189378738403,
|
483 |
-
"learning_rate": 0.0004539468401188468,
|
484 |
-
"loss": 0.8799,
|
485 |
-
"step": 68000
|
486 |
-
},
|
487 |
-
{
|
488 |
-
"epoch": 16.62049861495845,
|
489 |
-
"grad_norm": 0.5125630497932434,
|
490 |
-
"learning_rate": 0.00044591664659118286,
|
491 |
-
"loss": 0.8809,
|
492 |
-
"step": 69000
|
493 |
-
},
|
494 |
-
{
|
495 |
-
"epoch": 16.861375406479585,
|
496 |
-
"grad_norm": 0.4780360460281372,
|
497 |
-
"learning_rate": 0.00043788645306351885,
|
498 |
-
"loss": 0.8744,
|
499 |
-
"step": 70000
|
500 |
-
},
|
501 |
-
{
|
502 |
-
"epoch": 17.102252198000723,
|
503 |
-
"grad_norm": 0.541357159614563,
|
504 |
-
"learning_rate": 0.00042985625953585483,
|
505 |
-
"loss": 0.8729,
|
506 |
-
"step": 71000
|
507 |
-
},
|
508 |
-
{
|
509 |
-
"epoch": 17.34312898952186,
|
510 |
-
"grad_norm": 0.5186867713928223,
|
511 |
-
"learning_rate": 0.0004218260660081908,
|
512 |
-
"loss": 0.8673,
|
513 |
-
"step": 72000
|
514 |
-
},
|
515 |
-
{
|
516 |
-
"epoch": 17.584005781042997,
|
517 |
-
"grad_norm": 0.5490289330482483,
|
518 |
-
"learning_rate": 0.0004137958724805268,
|
519 |
-
"loss": 0.8644,
|
520 |
-
"step": 73000
|
521 |
-
},
|
522 |
-
{
|
523 |
-
"epoch": 17.824882572564132,
|
524 |
-
"grad_norm": 0.5589401125907898,
|
525 |
-
"learning_rate": 0.0004057656789528628,
|
526 |
-
"loss": 0.8676,
|
527 |
-
"step": 74000
|
528 |
-
},
|
529 |
-
{
|
530 |
-
"epoch": 18.06575936408527,
|
531 |
-
"grad_norm": 0.5271314382553101,
|
532 |
-
"learning_rate": 0.00039773548542519876,
|
533 |
-
"loss": 0.861,
|
534 |
-
"step": 75000
|
535 |
-
},
|
536 |
-
{
|
537 |
-
"epoch": 18.30663615560641,
|
538 |
-
"grad_norm": 0.586135983467102,
|
539 |
-
"learning_rate": 0.00038970529189753474,
|
540 |
-
"loss": 0.8556,
|
541 |
-
"step": 76000
|
542 |
-
},
|
543 |
-
{
|
544 |
-
"epoch": 18.547512947127544,
|
545 |
-
"grad_norm": 0.5827994346618652,
|
546 |
-
"learning_rate": 0.00038167509836987073,
|
547 |
-
"loss": 0.8565,
|
548 |
-
"step": 77000
|
549 |
-
},
|
550 |
-
{
|
551 |
-
"epoch": 18.788389738648682,
|
552 |
-
"grad_norm": 0.5451443195343018,
|
553 |
-
"learning_rate": 0.0003736449048422067,
|
554 |
-
"loss": 0.8554,
|
555 |
-
"step": 78000
|
556 |
-
},
|
557 |
-
{
|
558 |
-
"epoch": 19.029266530169817,
|
559 |
-
"grad_norm": 0.6020991206169128,
|
560 |
-
"learning_rate": 0.0003656147113145427,
|
561 |
-
"loss": 0.8506,
|
562 |
-
"step": 79000
|
563 |
-
},
|
564 |
-
{
|
565 |
-
"epoch": 19.270143321690956,
|
566 |
-
"grad_norm": 0.48624420166015625,
|
567 |
-
"learning_rate": 0.0003575845177868787,
|
568 |
-
"loss": 0.8449,
|
569 |
-
"step": 80000
|
570 |
-
},
|
571 |
-
{
|
572 |
-
"epoch": 19.51102011321209,
|
573 |
-
"grad_norm": 0.6073954105377197,
|
574 |
-
"learning_rate": 0.0003495543242592146,
|
575 |
-
"loss": 0.8436,
|
576 |
-
"step": 81000
|
577 |
-
},
|
578 |
-
{
|
579 |
-
"epoch": 19.75189690473323,
|
580 |
-
"grad_norm": 0.5890400409698486,
|
581 |
-
"learning_rate": 0.00034152413073155064,
|
582 |
-
"loss": 0.8445,
|
583 |
-
"step": 82000
|
584 |
-
},
|
585 |
-
{
|
586 |
-
"epoch": 19.992773696254368,
|
587 |
-
"grad_norm": 0.5391818881034851,
|
588 |
-
"learning_rate": 0.0003334939372038866,
|
589 |
-
"loss": 0.8405,
|
590 |
-
"step": 83000
|
591 |
-
},
|
592 |
-
{
|
593 |
-
"epoch": 20.233650487775503,
|
594 |
-
"grad_norm": 0.5576732158660889,
|
595 |
-
"learning_rate": 0.0003254637436762226,
|
596 |
-
"loss": 0.8315,
|
597 |
-
"step": 84000
|
598 |
-
},
|
599 |
-
{
|
600 |
-
"epoch": 20.47452727929664,
|
601 |
-
"grad_norm": 0.5405558347702026,
|
602 |
-
"learning_rate": 0.0003174335501485586,
|
603 |
-
"loss": 0.8348,
|
604 |
-
"step": 85000
|
605 |
-
},
|
606 |
-
{
|
607 |
-
"epoch": 20.715404070817776,
|
608 |
-
"grad_norm": 0.5441027879714966,
|
609 |
-
"learning_rate": 0.0003094033566208946,
|
610 |
-
"loss": 0.8326,
|
611 |
-
"step": 86000
|
612 |
-
},
|
613 |
-
{
|
614 |
-
"epoch": 20.956280862338915,
|
615 |
-
"grad_norm": 0.5216940641403198,
|
616 |
-
"learning_rate": 0.00030137316309323056,
|
617 |
-
"loss": 0.8251,
|
618 |
-
"step": 87000
|
619 |
-
},
|
620 |
-
{
|
621 |
-
"epoch": 21.19715765386005,
|
622 |
-
"grad_norm": 0.6281733512878418,
|
623 |
-
"learning_rate": 0.00029334296956556654,
|
624 |
-
"loss": 0.8219,
|
625 |
-
"step": 88000
|
626 |
-
},
|
627 |
-
{
|
628 |
-
"epoch": 21.438034445381188,
|
629 |
-
"grad_norm": 0.6415626406669617,
|
630 |
-
"learning_rate": 0.0002853127760379025,
|
631 |
-
"loss": 0.8225,
|
632 |
-
"step": 89000
|
633 |
-
},
|
634 |
-
{
|
635 |
-
"epoch": 21.678911236902323,
|
636 |
-
"grad_norm": 0.5944454073905945,
|
637 |
-
"learning_rate": 0.0002772825825102385,
|
638 |
-
"loss": 0.8223,
|
639 |
-
"step": 90000
|
640 |
-
},
|
641 |
-
{
|
642 |
-
"epoch": 21.91978802842346,
|
643 |
-
"grad_norm": 0.5434209704399109,
|
644 |
-
"learning_rate": 0.0002692523889825745,
|
645 |
-
"loss": 0.8161,
|
646 |
-
"step": 91000
|
647 |
-
},
|
648 |
-
{
|
649 |
-
"epoch": 22.1606648199446,
|
650 |
-
"grad_norm": 0.5393619537353516,
|
651 |
-
"learning_rate": 0.0002612221954549105,
|
652 |
-
"loss": 0.8126,
|
653 |
-
"step": 92000
|
654 |
-
},
|
655 |
-
{
|
656 |
-
"epoch": 22.401541611465735,
|
657 |
-
"grad_norm": 0.6141464710235596,
|
658 |
-
"learning_rate": 0.00025319200192724646,
|
659 |
-
"loss": 0.8119,
|
660 |
-
"step": 93000
|
661 |
-
},
|
662 |
-
{
|
663 |
-
"epoch": 22.642418402986873,
|
664 |
-
"grad_norm": 0.6085337400436401,
|
665 |
-
"learning_rate": 0.00024516180839958244,
|
666 |
-
"loss": 0.8082,
|
667 |
-
"step": 94000
|
668 |
-
},
|
669 |
-
{
|
670 |
-
"epoch": 22.88329519450801,
|
671 |
-
"grad_norm": 0.6050975322723389,
|
672 |
-
"learning_rate": 0.00023713161487191843,
|
673 |
-
"loss": 0.8065,
|
674 |
-
"step": 95000
|
675 |
-
},
|
676 |
-
{
|
677 |
-
"epoch": 23.124171986029147,
|
678 |
-
"grad_norm": 0.5710690021514893,
|
679 |
-
"learning_rate": 0.0002291014213442544,
|
680 |
-
"loss": 0.7971,
|
681 |
-
"step": 96000
|
682 |
-
},
|
683 |
-
{
|
684 |
-
"epoch": 23.36504877755028,
|
685 |
-
"grad_norm": 0.5681021213531494,
|
686 |
-
"learning_rate": 0.0002210712278165904,
|
687 |
-
"loss": 0.7996,
|
688 |
-
"step": 97000
|
689 |
-
},
|
690 |
-
{
|
691 |
-
"epoch": 23.60592556907142,
|
692 |
-
"grad_norm": 0.5795422792434692,
|
693 |
-
"learning_rate": 0.00021304103428892638,
|
694 |
-
"loss": 0.8006,
|
695 |
-
"step": 98000
|
696 |
-
},
|
697 |
-
{
|
698 |
-
"epoch": 23.846802360592555,
|
699 |
-
"grad_norm": 0.6810296773910522,
|
700 |
-
"learning_rate": 0.00020501084076126236,
|
701 |
-
"loss": 0.7931,
|
702 |
-
"step": 99000
|
703 |
-
},
|
704 |
-
{
|
705 |
-
"epoch": 24.087679152113694,
|
706 |
-
"grad_norm": 0.5023326277732849,
|
707 |
-
"learning_rate": 0.00019698064723359831,
|
708 |
-
"loss": 0.7895,
|
709 |
-
"step": 100000
|
710 |
-
},
|
711 |
-
{
|
712 |
-
"epoch": 24.328555943634832,
|
713 |
-
"grad_norm": 0.6325027346611023,
|
714 |
-
"learning_rate": 0.0001889504537059343,
|
715 |
-
"loss": 0.7902,
|
716 |
-
"step": 101000
|
717 |
-
},
|
718 |
-
{
|
719 |
-
"epoch": 24.569432735155967,
|
720 |
-
"grad_norm": 0.5804798007011414,
|
721 |
-
"learning_rate": 0.00018092026017827028,
|
722 |
-
"loss": 0.7912,
|
723 |
-
"step": 102000
|
724 |
-
},
|
725 |
-
{
|
726 |
-
"epoch": 24.810309526677106,
|
727 |
-
"grad_norm": 0.5688096284866333,
|
728 |
-
"learning_rate": 0.00017289006665060626,
|
729 |
-
"loss": 0.7886,
|
730 |
-
"step": 103000
|
731 |
-
},
|
732 |
-
{
|
733 |
-
"epoch": 25.05118631819824,
|
734 |
-
"grad_norm": 0.5234955549240112,
|
735 |
-
"learning_rate": 0.00016485987312294225,
|
736 |
-
"loss": 0.7828,
|
737 |
-
"step": 104000
|
738 |
-
},
|
739 |
-
{
|
740 |
-
"epoch": 25.29206310971938,
|
741 |
-
"grad_norm": 0.5501936078071594,
|
742 |
-
"learning_rate": 0.00015682967959527826,
|
743 |
-
"loss": 0.7769,
|
744 |
-
"step": 105000
|
745 |
-
},
|
746 |
-
{
|
747 |
-
"epoch": 25.532939901240514,
|
748 |
-
"grad_norm": 0.6095595359802246,
|
749 |
-
"learning_rate": 0.00014879948606761424,
|
750 |
-
"loss": 0.7788,
|
751 |
-
"step": 106000
|
752 |
-
},
|
753 |
-
{
|
754 |
-
"epoch": 25.773816692761653,
|
755 |
-
"grad_norm": 0.6600815057754517,
|
756 |
-
"learning_rate": 0.00014076929253995022,
|
757 |
-
"loss": 0.778,
|
758 |
-
"step": 107000
|
759 |
-
},
|
760 |
-
{
|
761 |
-
"epoch": 26.01469348428279,
|
762 |
-
"grad_norm": 0.6249046921730042,
|
763 |
-
"learning_rate": 0.0001327390990122862,
|
764 |
-
"loss": 0.7701,
|
765 |
-
"step": 108000
|
766 |
-
},
|
767 |
-
{
|
768 |
-
"epoch": 26.255570275803926,
|
769 |
-
"grad_norm": 0.7678042054176331,
|
770 |
-
"learning_rate": 0.0001247089054846222,
|
771 |
-
"loss": 0.7661,
|
772 |
-
"step": 109000
|
773 |
-
},
|
774 |
-
{
|
775 |
-
"epoch": 26.496447067325064,
|
776 |
-
"grad_norm": 0.5331607460975647,
|
777 |
-
"learning_rate": 0.00011667871195695816,
|
778 |
-
"loss": 0.7689,
|
779 |
-
"step": 110000
|
780 |
-
},
|
781 |
-
{
|
782 |
-
"epoch": 26.7373238588462,
|
783 |
-
"grad_norm": 0.6330080628395081,
|
784 |
-
"learning_rate": 0.00010864851842929414,
|
785 |
-
"loss": 0.7655,
|
786 |
-
"step": 111000
|
787 |
-
},
|
788 |
-
{
|
789 |
-
"epoch": 26.978200650367338,
|
790 |
-
"grad_norm": 0.6966120004653931,
|
791 |
-
"learning_rate": 0.00010061832490163013,
|
792 |
-
"loss": 0.7593,
|
793 |
-
"step": 112000
|
794 |
-
},
|
795 |
-
{
|
796 |
-
"epoch": 27.219077441888473,
|
797 |
-
"grad_norm": 0.6162911653518677,
|
798 |
-
"learning_rate": 9.258813137396611e-05,
|
799 |
-
"loss": 0.7619,
|
800 |
-
"step": 113000
|
801 |
-
},
|
802 |
-
{
|
803 |
-
"epoch": 27.45995423340961,
|
804 |
-
"grad_norm": 0.6430843472480774,
|
805 |
-
"learning_rate": 8.45579378463021e-05,
|
806 |
-
"loss": 0.7598,
|
807 |
-
"step": 114000
|
808 |
-
},
|
809 |
-
{
|
810 |
-
"epoch": 27.700831024930746,
|
811 |
-
"grad_norm": 0.6248263120651245,
|
812 |
-
"learning_rate": 7.652774431863809e-05,
|
813 |
-
"loss": 0.7584,
|
814 |
-
"step": 115000
|
815 |
-
},
|
816 |
-
{
|
817 |
-
"epoch": 27.941707816451885,
|
818 |
-
"grad_norm": 0.6525952219963074,
|
819 |
-
"learning_rate": 6.849755079097407e-05,
|
820 |
-
"loss": 0.7536,
|
821 |
-
"step": 116000
|
822 |
-
},
|
823 |
-
{
|
824 |
-
"epoch": 28.182584607973023,
|
825 |
-
"grad_norm": 0.64836186170578,
|
826 |
-
"learning_rate": 6.046735726331005e-05,
|
827 |
-
"loss": 0.7535,
|
828 |
-
"step": 117000
|
829 |
-
},
|
830 |
-
{
|
831 |
-
"epoch": 28.42346139949416,
|
832 |
-
"grad_norm": 0.5937727689743042,
|
833 |
-
"learning_rate": 5.243716373564603e-05,
|
834 |
-
"loss": 0.7453,
|
835 |
-
"step": 118000
|
836 |
-
},
|
837 |
-
{
|
838 |
-
"epoch": 28.664338191015297,
|
839 |
-
"grad_norm": 0.5951708555221558,
|
840 |
-
"learning_rate": 4.440697020798201e-05,
|
841 |
-
"loss": 0.7462,
|
842 |
-
"step": 119000
|
843 |
-
},
|
844 |
-
{
|
845 |
-
"epoch": 28.90521498253643,
|
846 |
-
"grad_norm": 0.5529988408088684,
|
847 |
-
"learning_rate": 3.6376776680318e-05,
|
848 |
-
"loss": 0.7478,
|
849 |
-
"step": 120000
|
850 |
-
},
|
851 |
-
{
|
852 |
-
"epoch": 29.14609177405757,
|
853 |
-
"grad_norm": 0.5422804355621338,
|
854 |
-
"learning_rate": 2.834658315265398e-05,
|
855 |
-
"loss": 0.7454,
|
856 |
-
"step": 121000
|
857 |
-
},
|
858 |
-
{
|
859 |
-
"epoch": 29.386968565578705,
|
860 |
-
"grad_norm": 0.5853854417800903,
|
861 |
-
"learning_rate": 2.0316389624989963e-05,
|
862 |
-
"loss": 0.7424,
|
863 |
-
"step": 122000
|
864 |
-
},
|
865 |
-
{
|
866 |
-
"epoch": 29.627845357099844,
|
867 |
-
"grad_norm": 0.6154918074607849,
|
868 |
-
"learning_rate": 1.2286196097325946e-05,
|
869 |
-
"loss": 0.7431,
|
870 |
-
"step": 123000
|
871 |
-
},
|
872 |
-
{
|
873 |
-
"epoch": 29.868722148620982,
|
874 |
-
"grad_norm": 0.596747875213623,
|
875 |
-
"learning_rate": 4.256002569661929e-06,
|
876 |
-
"loss": 0.7426,
|
877 |
-
"step": 124000
|
878 |
}
|
879 |
],
|
880 |
"logging_steps": 1000,
|
881 |
-
"max_steps":
|
882 |
"num_input_tokens_seen": 0,
|
883 |
-
"num_train_epochs":
|
884 |
"save_steps": 1000,
|
885 |
"stateful_callbacks": {
|
886 |
"TrainerControl": {
|
@@ -894,7 +306,7 @@
|
|
894 |
"attributes": {}
|
895 |
}
|
896 |
},
|
897 |
-
"total_flos":
|
898 |
"train_batch_size": 64,
|
899 |
"trial_name": null,
|
900 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 9.635071660845478,
|
5 |
"eval_steps": 500,
|
6 |
+
"global_step": 40000,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
10 |
"log_history": [
|
11 |
{
|
12 |
"epoch": 0.24087679152113695,
|
13 |
+
"grad_norm": 0.6258379220962524,
|
14 |
+
"learning_rate": 4.153472996796067e-06,
|
15 |
+
"loss": 0.7631,
|
16 |
"step": 1000
|
17 |
},
|
18 |
{
|
19 |
"epoch": 0.4817535830422739,
|
20 |
+
"grad_norm": 0.6458311676979065,
|
21 |
+
"learning_rate": 4.050943423930205e-06,
|
22 |
+
"loss": 0.7552,
|
23 |
"step": 2000
|
24 |
},
|
25 |
{
|
26 |
"epoch": 0.7226303745634108,
|
27 |
+
"grad_norm": 0.645728349685669,
|
28 |
+
"learning_rate": 3.948413851064344e-06,
|
29 |
+
"loss": 0.7559,
|
30 |
"step": 3000
|
31 |
},
|
32 |
{
|
33 |
"epoch": 0.9635071660845478,
|
34 |
+
"grad_norm": 0.6008012294769287,
|
35 |
+
"learning_rate": 3.845884278198482e-06,
|
36 |
+
"loss": 0.7608,
|
37 |
"step": 4000
|
38 |
},
|
39 |
{
|
40 |
"epoch": 1.2043839576056847,
|
41 |
+
"grad_norm": 0.7096830606460571,
|
42 |
+
"learning_rate": 3.743354705332619e-06,
|
43 |
+
"loss": 0.7574,
|
44 |
"step": 5000
|
45 |
},
|
46 |
{
|
47 |
"epoch": 1.4452607491268217,
|
48 |
+
"grad_norm": 0.5911938548088074,
|
49 |
+
"learning_rate": 3.640825132466757e-06,
|
50 |
+
"loss": 0.7549,
|
51 |
"step": 6000
|
52 |
},
|
53 |
{
|
54 |
"epoch": 1.6861375406479586,
|
55 |
+
"grad_norm": 0.7647172212600708,
|
56 |
+
"learning_rate": 3.5382955596008955e-06,
|
57 |
+
"loss": 0.7557,
|
58 |
"step": 7000
|
59 |
},
|
60 |
{
|
61 |
"epoch": 1.9270143321690956,
|
62 |
+
"grad_norm": 0.5935486555099487,
|
63 |
+
"learning_rate": 3.435765986735034e-06,
|
64 |
+
"loss": 0.7594,
|
65 |
"step": 8000
|
66 |
},
|
67 |
{
|
68 |
"epoch": 2.1678911236902323,
|
69 |
+
"grad_norm": 0.6902479529380798,
|
70 |
+
"learning_rate": 3.333236413869172e-06,
|
71 |
+
"loss": 0.7593,
|
72 |
"step": 9000
|
73 |
},
|
74 |
{
|
75 |
"epoch": 2.4087679152113695,
|
76 |
+
"grad_norm": 0.650104284286499,
|
77 |
+
"learning_rate": 3.2307068410033098e-06,
|
78 |
+
"loss": 0.7583,
|
79 |
"step": 10000
|
80 |
},
|
81 |
{
|
82 |
"epoch": 2.649644706732506,
|
83 |
+
"grad_norm": 0.5951535701751709,
|
84 |
+
"learning_rate": 3.1281772681374477e-06,
|
85 |
+
"loss": 0.7506,
|
86 |
"step": 11000
|
87 |
},
|
88 |
{
|
89 |
"epoch": 2.8905214982536434,
|
90 |
+
"grad_norm": 0.643113911151886,
|
91 |
+
"learning_rate": 3.0256476952715857e-06,
|
92 |
+
"loss": 0.7577,
|
93 |
"step": 12000
|
94 |
},
|
95 |
{
|
96 |
"epoch": 3.13139828977478,
|
97 |
+
"grad_norm": 0.5796828866004944,
|
98 |
+
"learning_rate": 2.923118122405724e-06,
|
99 |
+
"loss": 0.7594,
|
100 |
"step": 13000
|
101 |
},
|
102 |
{
|
103 |
"epoch": 3.3722750812959172,
|
104 |
+
"grad_norm": 0.6443358063697815,
|
105 |
+
"learning_rate": 2.820588549539862e-06,
|
106 |
+
"loss": 0.7539,
|
107 |
"step": 14000
|
108 |
},
|
109 |
{
|
110 |
"epoch": 3.613151872817054,
|
111 |
+
"grad_norm": 0.5989753603935242,
|
112 |
+
"learning_rate": 2.718058976674e-06,
|
113 |
+
"loss": 0.7578,
|
114 |
"step": 15000
|
115 |
},
|
116 |
{
|
117 |
"epoch": 3.854028664338191,
|
118 |
+
"grad_norm": 0.5480288863182068,
|
119 |
+
"learning_rate": 2.615529403808138e-06,
|
120 |
+
"loss": 0.7531,
|
121 |
"step": 16000
|
122 |
},
|
123 |
{
|
124 |
"epoch": 4.094905455859328,
|
125 |
+
"grad_norm": 0.6654678583145142,
|
126 |
+
"learning_rate": 2.512999830942276e-06,
|
127 |
+
"loss": 0.7579,
|
128 |
"step": 17000
|
129 |
},
|
130 |
{
|
131 |
"epoch": 4.335782247380465,
|
132 |
+
"grad_norm": 0.6215384006500244,
|
133 |
+
"learning_rate": 2.4104702580764137e-06,
|
134 |
+
"loss": 0.7545,
|
135 |
"step": 18000
|
136 |
},
|
137 |
{
|
138 |
"epoch": 4.576659038901602,
|
139 |
+
"grad_norm": 0.6459180116653442,
|
140 |
+
"learning_rate": 2.3079406852105525e-06,
|
141 |
+
"loss": 0.7566,
|
142 |
"step": 19000
|
143 |
},
|
144 |
{
|
145 |
"epoch": 4.817535830422739,
|
146 |
+
"grad_norm": 0.6795477867126465,
|
147 |
+
"learning_rate": 2.2054111123446905e-06,
|
148 |
+
"loss": 0.7555,
|
149 |
"step": 20000
|
150 |
},
|
151 |
{
|
152 |
"epoch": 5.058412621943876,
|
153 |
+
"grad_norm": 0.6410796046257019,
|
154 |
+
"learning_rate": 2.1028815394788284e-06,
|
155 |
+
"loss": 0.7503,
|
156 |
"step": 21000
|
157 |
},
|
158 |
{
|
159 |
"epoch": 5.299289413465012,
|
160 |
+
"grad_norm": 0.5807074308395386,
|
161 |
+
"learning_rate": 2.0003519666129664e-06,
|
162 |
+
"loss": 0.7529,
|
163 |
"step": 22000
|
164 |
},
|
165 |
{
|
166 |
"epoch": 5.54016620498615,
|
167 |
+
"grad_norm": 0.5771316885948181,
|
168 |
+
"learning_rate": 1.8978223937471045e-06,
|
169 |
+
"loss": 0.751,
|
170 |
"step": 23000
|
171 |
},
|
172 |
{
|
173 |
"epoch": 5.781042996507287,
|
174 |
+
"grad_norm": 0.6301162838935852,
|
175 |
+
"learning_rate": 1.7952928208812425e-06,
|
176 |
+
"loss": 0.7534,
|
177 |
"step": 24000
|
178 |
},
|
179 |
{
|
180 |
"epoch": 6.021919788028423,
|
181 |
+
"grad_norm": 0.5753132104873657,
|
182 |
+
"learning_rate": 1.6927632480153806e-06,
|
183 |
+
"loss": 0.7511,
|
184 |
"step": 25000
|
185 |
},
|
186 |
{
|
187 |
"epoch": 6.26279657954956,
|
188 |
+
"grad_norm": 0.6592470407485962,
|
189 |
+
"learning_rate": 1.5902336751495186e-06,
|
190 |
+
"loss": 0.7554,
|
191 |
"step": 26000
|
192 |
},
|
193 |
{
|
194 |
"epoch": 6.503673371070697,
|
195 |
+
"grad_norm": 0.6203874349594116,
|
196 |
+
"learning_rate": 1.4877041022836565e-06,
|
197 |
+
"loss": 0.7568,
|
198 |
"step": 27000
|
199 |
},
|
200 |
{
|
201 |
"epoch": 6.7445501625918345,
|
202 |
+
"grad_norm": 0.6534298062324524,
|
203 |
+
"learning_rate": 1.385174529417795e-06,
|
204 |
+
"loss": 0.7553,
|
205 |
"step": 28000
|
206 |
},
|
207 |
{
|
208 |
"epoch": 6.985426954112971,
|
209 |
+
"grad_norm": 0.6775331497192383,
|
210 |
+
"learning_rate": 1.2826449565519328e-06,
|
211 |
+
"loss": 0.7525,
|
212 |
"step": 29000
|
213 |
},
|
214 |
{
|
215 |
"epoch": 7.226303745634108,
|
216 |
+
"grad_norm": 0.640843391418457,
|
217 |
+
"learning_rate": 1.180115383686071e-06,
|
218 |
+
"loss": 0.7559,
|
219 |
"step": 30000
|
220 |
},
|
221 |
{
|
222 |
"epoch": 7.4671805371552455,
|
223 |
+
"grad_norm": 0.6803774833679199,
|
224 |
+
"learning_rate": 1.077585810820209e-06,
|
225 |
+
"loss": 0.7508,
|
226 |
"step": 31000
|
227 |
},
|
228 |
{
|
229 |
"epoch": 7.708057328676382,
|
230 |
+
"grad_norm": 0.6412643790245056,
|
231 |
+
"learning_rate": 9.75056237954347e-07,
|
232 |
+
"loss": 0.7534,
|
233 |
"step": 32000
|
234 |
},
|
235 |
{
|
236 |
"epoch": 7.948934120197519,
|
237 |
+
"grad_norm": 0.6142516136169434,
|
238 |
+
"learning_rate": 8.725266650884852e-07,
|
239 |
+
"loss": 0.7531,
|
240 |
"step": 33000
|
241 |
},
|
242 |
{
|
243 |
"epoch": 8.189810911718656,
|
244 |
+
"grad_norm": 0.670280933380127,
|
245 |
+
"learning_rate": 7.699970922226231e-07,
|
246 |
+
"loss": 0.7506,
|
247 |
"step": 34000
|
248 |
},
|
249 |
{
|
250 |
"epoch": 8.430687703239792,
|
251 |
+
"grad_norm": 0.6336104869842529,
|
252 |
+
"learning_rate": 6.674675193567612e-07,
|
253 |
+
"loss": 0.7538,
|
254 |
"step": 35000
|
255 |
},
|
256 |
{
|
257 |
"epoch": 8.67156449476093,
|
258 |
+
"grad_norm": 0.6225976347923279,
|
259 |
+
"learning_rate": 5.649379464908993e-07,
|
260 |
+
"loss": 0.7521,
|
261 |
"step": 36000
|
262 |
},
|
263 |
{
|
264 |
"epoch": 8.912441286282068,
|
265 |
+
"grad_norm": 0.7998696565628052,
|
266 |
+
"learning_rate": 4.624083736250373e-07,
|
267 |
+
"loss": 0.7559,
|
268 |
"step": 37000
|
269 |
},
|
270 |
{
|
271 |
"epoch": 9.153318077803204,
|
272 |
+
"grad_norm": 0.6381024122238159,
|
273 |
+
"learning_rate": 3.598788007591754e-07,
|
274 |
+
"loss": 0.7547,
|
275 |
"step": 38000
|
276 |
},
|
277 |
{
|
278 |
"epoch": 9.394194869324341,
|
279 |
+
"grad_norm": 0.558660089969635,
|
280 |
+
"learning_rate": 2.5734922789331346e-07,
|
281 |
+
"loss": 0.7539,
|
282 |
"step": 39000
|
283 |
},
|
284 |
{
|
285 |
"epoch": 9.635071660845478,
|
286 |
+
"grad_norm": 0.7711436748504639,
|
287 |
+
"learning_rate": 1.5481965502745152e-07,
|
288 |
+
"loss": 0.7484,
|
289 |
"step": 40000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
290 |
}
|
291 |
],
|
292 |
"logging_steps": 1000,
|
293 |
+
"max_steps": 41510,
|
294 |
"num_input_tokens_seen": 0,
|
295 |
+
"num_train_epochs": 10,
|
296 |
"save_steps": 1000,
|
297 |
"stateful_callbacks": {
|
298 |
"TrainerControl": {
|
|
|
306 |
"attributes": {}
|
307 |
}
|
308 |
},
|
309 |
+
"total_flos": 4.00145917384065e+17,
|
310 |
"train_batch_size": 64,
|
311 |
"trial_name": null,
|
312 |
"trial_params": null
|
last-checkpoint/training_args.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 5112
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:1b19c50ebceb96bf9ab86b4ad80b203101b877c903c2c529277d18a83643e787
|
3 |
size 5112
|