Delete trainer_state.json
Browse files- trainer_state.json +0 -787
trainer_state.json
DELETED
@@ -1,787 +0,0 @@
|
|
1 |
-
{
|
2 |
-
"best_metric": 0.3883955776691437,
|
3 |
-
"best_model_checkpoint": "exp/vicuna-7b-lora-sft-code_qa_desc_summ_triplet_r_16_alpha_32_8GPUs-0116/checkpoint-1200",
|
4 |
-
"epoch": 4.375569735642662,
|
5 |
-
"eval_steps": 200,
|
6 |
-
"global_step": 1200,
|
7 |
-
"is_hyper_param_search": false,
|
8 |
-
"is_local_process_zero": true,
|
9 |
-
"is_world_process_zero": true,
|
10 |
-
"log_history": [
|
11 |
-
{
|
12 |
-
"epoch": 0.04,
|
13 |
-
"learning_rate": 2.9999999999999997e-05,
|
14 |
-
"loss": 1.4343,
|
15 |
-
"step": 10
|
16 |
-
},
|
17 |
-
{
|
18 |
-
"epoch": 0.07,
|
19 |
-
"learning_rate": 5.9999999999999995e-05,
|
20 |
-
"loss": 1.4848,
|
21 |
-
"step": 20
|
22 |
-
},
|
23 |
-
{
|
24 |
-
"epoch": 0.11,
|
25 |
-
"learning_rate": 8.999999999999999e-05,
|
26 |
-
"loss": 1.1941,
|
27 |
-
"step": 30
|
28 |
-
},
|
29 |
-
{
|
30 |
-
"epoch": 0.15,
|
31 |
-
"learning_rate": 0.00011999999999999999,
|
32 |
-
"loss": 0.8226,
|
33 |
-
"step": 40
|
34 |
-
},
|
35 |
-
{
|
36 |
-
"epoch": 0.18,
|
37 |
-
"learning_rate": 0.00015,
|
38 |
-
"loss": 0.6671,
|
39 |
-
"step": 50
|
40 |
-
},
|
41 |
-
{
|
42 |
-
"epoch": 0.22,
|
43 |
-
"learning_rate": 0.00017999999999999998,
|
44 |
-
"loss": 0.5676,
|
45 |
-
"step": 60
|
46 |
-
},
|
47 |
-
{
|
48 |
-
"epoch": 0.26,
|
49 |
-
"learning_rate": 0.00020999999999999998,
|
50 |
-
"loss": 0.5655,
|
51 |
-
"step": 70
|
52 |
-
},
|
53 |
-
{
|
54 |
-
"epoch": 0.29,
|
55 |
-
"learning_rate": 0.00023999999999999998,
|
56 |
-
"loss": 0.5251,
|
57 |
-
"step": 80
|
58 |
-
},
|
59 |
-
{
|
60 |
-
"epoch": 0.33,
|
61 |
-
"learning_rate": 0.00027,
|
62 |
-
"loss": 0.4845,
|
63 |
-
"step": 90
|
64 |
-
},
|
65 |
-
{
|
66 |
-
"epoch": 0.36,
|
67 |
-
"learning_rate": 0.0003,
|
68 |
-
"loss": 0.481,
|
69 |
-
"step": 100
|
70 |
-
},
|
71 |
-
{
|
72 |
-
"epoch": 0.4,
|
73 |
-
"learning_rate": 0.0002976377952755905,
|
74 |
-
"loss": 0.4565,
|
75 |
-
"step": 110
|
76 |
-
},
|
77 |
-
{
|
78 |
-
"epoch": 0.44,
|
79 |
-
"learning_rate": 0.0002952755905511811,
|
80 |
-
"loss": 0.4625,
|
81 |
-
"step": 120
|
82 |
-
},
|
83 |
-
{
|
84 |
-
"epoch": 0.47,
|
85 |
-
"learning_rate": 0.00029291338582677163,
|
86 |
-
"loss": 0.4584,
|
87 |
-
"step": 130
|
88 |
-
},
|
89 |
-
{
|
90 |
-
"epoch": 0.51,
|
91 |
-
"learning_rate": 0.00029055118110236217,
|
92 |
-
"loss": 0.4425,
|
93 |
-
"step": 140
|
94 |
-
},
|
95 |
-
{
|
96 |
-
"epoch": 0.55,
|
97 |
-
"learning_rate": 0.0002881889763779527,
|
98 |
-
"loss": 0.4573,
|
99 |
-
"step": 150
|
100 |
-
},
|
101 |
-
{
|
102 |
-
"epoch": 0.58,
|
103 |
-
"learning_rate": 0.0002858267716535433,
|
104 |
-
"loss": 0.4361,
|
105 |
-
"step": 160
|
106 |
-
},
|
107 |
-
{
|
108 |
-
"epoch": 0.62,
|
109 |
-
"learning_rate": 0.00028346456692913383,
|
110 |
-
"loss": 0.4396,
|
111 |
-
"step": 170
|
112 |
-
},
|
113 |
-
{
|
114 |
-
"epoch": 0.66,
|
115 |
-
"learning_rate": 0.00028110236220472436,
|
116 |
-
"loss": 0.4391,
|
117 |
-
"step": 180
|
118 |
-
},
|
119 |
-
{
|
120 |
-
"epoch": 0.69,
|
121 |
-
"learning_rate": 0.00027874015748031495,
|
122 |
-
"loss": 0.418,
|
123 |
-
"step": 190
|
124 |
-
},
|
125 |
-
{
|
126 |
-
"epoch": 0.73,
|
127 |
-
"learning_rate": 0.0002763779527559055,
|
128 |
-
"loss": 0.4469,
|
129 |
-
"step": 200
|
130 |
-
},
|
131 |
-
{
|
132 |
-
"epoch": 0.73,
|
133 |
-
"eval_loss": 0.4269736409187317,
|
134 |
-
"eval_runtime": 19.352,
|
135 |
-
"eval_samples_per_second": 103.348,
|
136 |
-
"eval_steps_per_second": 1.654,
|
137 |
-
"step": 200
|
138 |
-
},
|
139 |
-
{
|
140 |
-
"epoch": 0.77,
|
141 |
-
"learning_rate": 0.0002740157480314961,
|
142 |
-
"loss": 0.4149,
|
143 |
-
"step": 210
|
144 |
-
},
|
145 |
-
{
|
146 |
-
"epoch": 0.8,
|
147 |
-
"learning_rate": 0.00027165354330708656,
|
148 |
-
"loss": 0.428,
|
149 |
-
"step": 220
|
150 |
-
},
|
151 |
-
{
|
152 |
-
"epoch": 0.84,
|
153 |
-
"learning_rate": 0.00026929133858267715,
|
154 |
-
"loss": 0.4248,
|
155 |
-
"step": 230
|
156 |
-
},
|
157 |
-
{
|
158 |
-
"epoch": 0.88,
|
159 |
-
"learning_rate": 0.0002669291338582677,
|
160 |
-
"loss": 0.4249,
|
161 |
-
"step": 240
|
162 |
-
},
|
163 |
-
{
|
164 |
-
"epoch": 0.91,
|
165 |
-
"learning_rate": 0.0002645669291338582,
|
166 |
-
"loss": 0.4331,
|
167 |
-
"step": 250
|
168 |
-
},
|
169 |
-
{
|
170 |
-
"epoch": 0.95,
|
171 |
-
"learning_rate": 0.0002622047244094488,
|
172 |
-
"loss": 0.4192,
|
173 |
-
"step": 260
|
174 |
-
},
|
175 |
-
{
|
176 |
-
"epoch": 0.98,
|
177 |
-
"learning_rate": 0.00025984251968503934,
|
178 |
-
"loss": 0.4204,
|
179 |
-
"step": 270
|
180 |
-
},
|
181 |
-
{
|
182 |
-
"epoch": 1.02,
|
183 |
-
"learning_rate": 0.00025748031496062993,
|
184 |
-
"loss": 0.4318,
|
185 |
-
"step": 280
|
186 |
-
},
|
187 |
-
{
|
188 |
-
"epoch": 1.06,
|
189 |
-
"learning_rate": 0.00025511811023622047,
|
190 |
-
"loss": 0.4229,
|
191 |
-
"step": 290
|
192 |
-
},
|
193 |
-
{
|
194 |
-
"epoch": 1.09,
|
195 |
-
"learning_rate": 0.000252755905511811,
|
196 |
-
"loss": 0.4214,
|
197 |
-
"step": 300
|
198 |
-
},
|
199 |
-
{
|
200 |
-
"epoch": 1.13,
|
201 |
-
"learning_rate": 0.00025039370078740154,
|
202 |
-
"loss": 0.416,
|
203 |
-
"step": 310
|
204 |
-
},
|
205 |
-
{
|
206 |
-
"epoch": 1.17,
|
207 |
-
"learning_rate": 0.00024803149606299207,
|
208 |
-
"loss": 0.4199,
|
209 |
-
"step": 320
|
210 |
-
},
|
211 |
-
{
|
212 |
-
"epoch": 1.2,
|
213 |
-
"learning_rate": 0.00024566929133858266,
|
214 |
-
"loss": 0.4218,
|
215 |
-
"step": 330
|
216 |
-
},
|
217 |
-
{
|
218 |
-
"epoch": 1.24,
|
219 |
-
"learning_rate": 0.0002433070866141732,
|
220 |
-
"loss": 0.4113,
|
221 |
-
"step": 340
|
222 |
-
},
|
223 |
-
{
|
224 |
-
"epoch": 1.28,
|
225 |
-
"learning_rate": 0.00024094488188976376,
|
226 |
-
"loss": 0.4185,
|
227 |
-
"step": 350
|
228 |
-
},
|
229 |
-
{
|
230 |
-
"epoch": 1.31,
|
231 |
-
"learning_rate": 0.00023858267716535432,
|
232 |
-
"loss": 0.4168,
|
233 |
-
"step": 360
|
234 |
-
},
|
235 |
-
{
|
236 |
-
"epoch": 1.35,
|
237 |
-
"learning_rate": 0.00023622047244094488,
|
238 |
-
"loss": 0.4162,
|
239 |
-
"step": 370
|
240 |
-
},
|
241 |
-
{
|
242 |
-
"epoch": 1.39,
|
243 |
-
"learning_rate": 0.0002338582677165354,
|
244 |
-
"loss": 0.4175,
|
245 |
-
"step": 380
|
246 |
-
},
|
247 |
-
{
|
248 |
-
"epoch": 1.42,
|
249 |
-
"learning_rate": 0.00023149606299212595,
|
250 |
-
"loss": 0.4045,
|
251 |
-
"step": 390
|
252 |
-
},
|
253 |
-
{
|
254 |
-
"epoch": 1.46,
|
255 |
-
"learning_rate": 0.00022913385826771652,
|
256 |
-
"loss": 0.4152,
|
257 |
-
"step": 400
|
258 |
-
},
|
259 |
-
{
|
260 |
-
"epoch": 1.46,
|
261 |
-
"eval_loss": 0.4086858630180359,
|
262 |
-
"eval_runtime": 19.2818,
|
263 |
-
"eval_samples_per_second": 103.725,
|
264 |
-
"eval_steps_per_second": 1.66,
|
265 |
-
"step": 400
|
266 |
-
},
|
267 |
-
{
|
268 |
-
"epoch": 1.49,
|
269 |
-
"learning_rate": 0.00022677165354330705,
|
270 |
-
"loss": 0.415,
|
271 |
-
"step": 410
|
272 |
-
},
|
273 |
-
{
|
274 |
-
"epoch": 1.53,
|
275 |
-
"learning_rate": 0.00022440944881889761,
|
276 |
-
"loss": 0.4091,
|
277 |
-
"step": 420
|
278 |
-
},
|
279 |
-
{
|
280 |
-
"epoch": 1.57,
|
281 |
-
"learning_rate": 0.00022204724409448818,
|
282 |
-
"loss": 0.4132,
|
283 |
-
"step": 430
|
284 |
-
},
|
285 |
-
{
|
286 |
-
"epoch": 1.6,
|
287 |
-
"learning_rate": 0.00021968503937007874,
|
288 |
-
"loss": 0.3985,
|
289 |
-
"step": 440
|
290 |
-
},
|
291 |
-
{
|
292 |
-
"epoch": 1.64,
|
293 |
-
"learning_rate": 0.00021732283464566927,
|
294 |
-
"loss": 0.4056,
|
295 |
-
"step": 450
|
296 |
-
},
|
297 |
-
{
|
298 |
-
"epoch": 1.68,
|
299 |
-
"learning_rate": 0.0002149606299212598,
|
300 |
-
"loss": 0.4005,
|
301 |
-
"step": 460
|
302 |
-
},
|
303 |
-
{
|
304 |
-
"epoch": 1.71,
|
305 |
-
"learning_rate": 0.00021259842519685037,
|
306 |
-
"loss": 0.4059,
|
307 |
-
"step": 470
|
308 |
-
},
|
309 |
-
{
|
310 |
-
"epoch": 1.75,
|
311 |
-
"learning_rate": 0.0002102362204724409,
|
312 |
-
"loss": 0.409,
|
313 |
-
"step": 480
|
314 |
-
},
|
315 |
-
{
|
316 |
-
"epoch": 1.79,
|
317 |
-
"learning_rate": 0.00020787401574803147,
|
318 |
-
"loss": 0.4031,
|
319 |
-
"step": 490
|
320 |
-
},
|
321 |
-
{
|
322 |
-
"epoch": 1.82,
|
323 |
-
"learning_rate": 0.00020551181102362203,
|
324 |
-
"loss": 0.4097,
|
325 |
-
"step": 500
|
326 |
-
},
|
327 |
-
{
|
328 |
-
"epoch": 1.86,
|
329 |
-
"learning_rate": 0.0002031496062992126,
|
330 |
-
"loss": 0.4017,
|
331 |
-
"step": 510
|
332 |
-
},
|
333 |
-
{
|
334 |
-
"epoch": 1.9,
|
335 |
-
"learning_rate": 0.00020078740157480313,
|
336 |
-
"loss": 0.4026,
|
337 |
-
"step": 520
|
338 |
-
},
|
339 |
-
{
|
340 |
-
"epoch": 1.93,
|
341 |
-
"learning_rate": 0.0001984251968503937,
|
342 |
-
"loss": 0.4106,
|
343 |
-
"step": 530
|
344 |
-
},
|
345 |
-
{
|
346 |
-
"epoch": 1.97,
|
347 |
-
"learning_rate": 0.00019606299212598423,
|
348 |
-
"loss": 0.395,
|
349 |
-
"step": 540
|
350 |
-
},
|
351 |
-
{
|
352 |
-
"epoch": 2.01,
|
353 |
-
"learning_rate": 0.0001937007874015748,
|
354 |
-
"loss": 0.3988,
|
355 |
-
"step": 550
|
356 |
-
},
|
357 |
-
{
|
358 |
-
"epoch": 2.04,
|
359 |
-
"learning_rate": 0.00019133858267716532,
|
360 |
-
"loss": 0.409,
|
361 |
-
"step": 560
|
362 |
-
},
|
363 |
-
{
|
364 |
-
"epoch": 2.08,
|
365 |
-
"learning_rate": 0.00018897637795275589,
|
366 |
-
"loss": 0.3997,
|
367 |
-
"step": 570
|
368 |
-
},
|
369 |
-
{
|
370 |
-
"epoch": 2.11,
|
371 |
-
"learning_rate": 0.00018661417322834645,
|
372 |
-
"loss": 0.4007,
|
373 |
-
"step": 580
|
374 |
-
},
|
375 |
-
{
|
376 |
-
"epoch": 2.15,
|
377 |
-
"learning_rate": 0.000184251968503937,
|
378 |
-
"loss": 0.3905,
|
379 |
-
"step": 590
|
380 |
-
},
|
381 |
-
{
|
382 |
-
"epoch": 2.19,
|
383 |
-
"learning_rate": 0.00018188976377952755,
|
384 |
-
"loss": 0.4005,
|
385 |
-
"step": 600
|
386 |
-
},
|
387 |
-
{
|
388 |
-
"epoch": 2.19,
|
389 |
-
"eval_loss": 0.40032637119293213,
|
390 |
-
"eval_runtime": 19.2818,
|
391 |
-
"eval_samples_per_second": 103.725,
|
392 |
-
"eval_steps_per_second": 1.66,
|
393 |
-
"step": 600
|
394 |
-
},
|
395 |
-
{
|
396 |
-
"epoch": 2.22,
|
397 |
-
"learning_rate": 0.0001795275590551181,
|
398 |
-
"loss": 0.3983,
|
399 |
-
"step": 610
|
400 |
-
},
|
401 |
-
{
|
402 |
-
"epoch": 2.26,
|
403 |
-
"learning_rate": 0.00017716535433070864,
|
404 |
-
"loss": 0.3881,
|
405 |
-
"step": 620
|
406 |
-
},
|
407 |
-
{
|
408 |
-
"epoch": 2.3,
|
409 |
-
"learning_rate": 0.00017480314960629918,
|
410 |
-
"loss": 0.4008,
|
411 |
-
"step": 630
|
412 |
-
},
|
413 |
-
{
|
414 |
-
"epoch": 2.33,
|
415 |
-
"learning_rate": 0.00017244094488188974,
|
416 |
-
"loss": 0.3927,
|
417 |
-
"step": 640
|
418 |
-
},
|
419 |
-
{
|
420 |
-
"epoch": 2.37,
|
421 |
-
"learning_rate": 0.0001700787401574803,
|
422 |
-
"loss": 0.4005,
|
423 |
-
"step": 650
|
424 |
-
},
|
425 |
-
{
|
426 |
-
"epoch": 2.41,
|
427 |
-
"learning_rate": 0.00016771653543307086,
|
428 |
-
"loss": 0.3962,
|
429 |
-
"step": 660
|
430 |
-
},
|
431 |
-
{
|
432 |
-
"epoch": 2.44,
|
433 |
-
"learning_rate": 0.0001653543307086614,
|
434 |
-
"loss": 0.3902,
|
435 |
-
"step": 670
|
436 |
-
},
|
437 |
-
{
|
438 |
-
"epoch": 2.48,
|
439 |
-
"learning_rate": 0.00016299212598425196,
|
440 |
-
"loss": 0.3911,
|
441 |
-
"step": 680
|
442 |
-
},
|
443 |
-
{
|
444 |
-
"epoch": 2.52,
|
445 |
-
"learning_rate": 0.00016062992125984252,
|
446 |
-
"loss": 0.3891,
|
447 |
-
"step": 690
|
448 |
-
},
|
449 |
-
{
|
450 |
-
"epoch": 2.55,
|
451 |
-
"learning_rate": 0.00015826771653543303,
|
452 |
-
"loss": 0.3939,
|
453 |
-
"step": 700
|
454 |
-
},
|
455 |
-
{
|
456 |
-
"epoch": 2.59,
|
457 |
-
"learning_rate": 0.0001559055118110236,
|
458 |
-
"loss": 0.4001,
|
459 |
-
"step": 710
|
460 |
-
},
|
461 |
-
{
|
462 |
-
"epoch": 2.63,
|
463 |
-
"learning_rate": 0.00015354330708661416,
|
464 |
-
"loss": 0.3918,
|
465 |
-
"step": 720
|
466 |
-
},
|
467 |
-
{
|
468 |
-
"epoch": 2.66,
|
469 |
-
"learning_rate": 0.00015118110236220472,
|
470 |
-
"loss": 0.3979,
|
471 |
-
"step": 730
|
472 |
-
},
|
473 |
-
{
|
474 |
-
"epoch": 2.7,
|
475 |
-
"learning_rate": 0.00014881889763779525,
|
476 |
-
"loss": 0.3793,
|
477 |
-
"step": 740
|
478 |
-
},
|
479 |
-
{
|
480 |
-
"epoch": 2.73,
|
481 |
-
"learning_rate": 0.00014645669291338582,
|
482 |
-
"loss": 0.3879,
|
483 |
-
"step": 750
|
484 |
-
},
|
485 |
-
{
|
486 |
-
"epoch": 2.77,
|
487 |
-
"learning_rate": 0.00014409448818897635,
|
488 |
-
"loss": 0.3915,
|
489 |
-
"step": 760
|
490 |
-
},
|
491 |
-
{
|
492 |
-
"epoch": 2.81,
|
493 |
-
"learning_rate": 0.00014173228346456691,
|
494 |
-
"loss": 0.3831,
|
495 |
-
"step": 770
|
496 |
-
},
|
497 |
-
{
|
498 |
-
"epoch": 2.84,
|
499 |
-
"learning_rate": 0.00013937007874015748,
|
500 |
-
"loss": 0.3838,
|
501 |
-
"step": 780
|
502 |
-
},
|
503 |
-
{
|
504 |
-
"epoch": 2.88,
|
505 |
-
"learning_rate": 0.00013700787401574804,
|
506 |
-
"loss": 0.3734,
|
507 |
-
"step": 790
|
508 |
-
},
|
509 |
-
{
|
510 |
-
"epoch": 2.92,
|
511 |
-
"learning_rate": 0.00013464566929133857,
|
512 |
-
"loss": 0.3872,
|
513 |
-
"step": 800
|
514 |
-
},
|
515 |
-
{
|
516 |
-
"epoch": 2.92,
|
517 |
-
"eval_loss": 0.3944130539894104,
|
518 |
-
"eval_runtime": 19.2596,
|
519 |
-
"eval_samples_per_second": 103.844,
|
520 |
-
"eval_steps_per_second": 1.662,
|
521 |
-
"step": 800
|
522 |
-
},
|
523 |
-
{
|
524 |
-
"epoch": 2.95,
|
525 |
-
"learning_rate": 0.0001322834645669291,
|
526 |
-
"loss": 0.386,
|
527 |
-
"step": 810
|
528 |
-
},
|
529 |
-
{
|
530 |
-
"epoch": 2.99,
|
531 |
-
"learning_rate": 0.00012992125984251967,
|
532 |
-
"loss": 0.3799,
|
533 |
-
"step": 820
|
534 |
-
},
|
535 |
-
{
|
536 |
-
"epoch": 3.03,
|
537 |
-
"learning_rate": 0.00012755905511811023,
|
538 |
-
"loss": 0.3895,
|
539 |
-
"step": 830
|
540 |
-
},
|
541 |
-
{
|
542 |
-
"epoch": 3.06,
|
543 |
-
"learning_rate": 0.00012519685039370077,
|
544 |
-
"loss": 0.3852,
|
545 |
-
"step": 840
|
546 |
-
},
|
547 |
-
{
|
548 |
-
"epoch": 3.1,
|
549 |
-
"learning_rate": 0.00012283464566929133,
|
550 |
-
"loss": 0.3879,
|
551 |
-
"step": 850
|
552 |
-
},
|
553 |
-
{
|
554 |
-
"epoch": 3.14,
|
555 |
-
"learning_rate": 0.00012047244094488188,
|
556 |
-
"loss": 0.3892,
|
557 |
-
"step": 860
|
558 |
-
},
|
559 |
-
{
|
560 |
-
"epoch": 3.17,
|
561 |
-
"learning_rate": 0.00011811023622047244,
|
562 |
-
"loss": 0.3801,
|
563 |
-
"step": 870
|
564 |
-
},
|
565 |
-
{
|
566 |
-
"epoch": 3.21,
|
567 |
-
"learning_rate": 0.00011574803149606298,
|
568 |
-
"loss": 0.3802,
|
569 |
-
"step": 880
|
570 |
-
},
|
571 |
-
{
|
572 |
-
"epoch": 3.25,
|
573 |
-
"learning_rate": 0.00011338582677165353,
|
574 |
-
"loss": 0.3863,
|
575 |
-
"step": 890
|
576 |
-
},
|
577 |
-
{
|
578 |
-
"epoch": 3.28,
|
579 |
-
"learning_rate": 0.00011102362204724409,
|
580 |
-
"loss": 0.3792,
|
581 |
-
"step": 900
|
582 |
-
},
|
583 |
-
{
|
584 |
-
"epoch": 3.32,
|
585 |
-
"learning_rate": 0.00010866141732283464,
|
586 |
-
"loss": 0.3923,
|
587 |
-
"step": 910
|
588 |
-
},
|
589 |
-
{
|
590 |
-
"epoch": 3.35,
|
591 |
-
"learning_rate": 0.00010629921259842519,
|
592 |
-
"loss": 0.3753,
|
593 |
-
"step": 920
|
594 |
-
},
|
595 |
-
{
|
596 |
-
"epoch": 3.39,
|
597 |
-
"learning_rate": 0.00010393700787401573,
|
598 |
-
"loss": 0.3777,
|
599 |
-
"step": 930
|
600 |
-
},
|
601 |
-
{
|
602 |
-
"epoch": 3.43,
|
603 |
-
"learning_rate": 0.0001015748031496063,
|
604 |
-
"loss": 0.3849,
|
605 |
-
"step": 940
|
606 |
-
},
|
607 |
-
{
|
608 |
-
"epoch": 3.46,
|
609 |
-
"learning_rate": 9.921259842519685e-05,
|
610 |
-
"loss": 0.3775,
|
611 |
-
"step": 950
|
612 |
-
},
|
613 |
-
{
|
614 |
-
"epoch": 3.5,
|
615 |
-
"learning_rate": 9.68503937007874e-05,
|
616 |
-
"loss": 0.3853,
|
617 |
-
"step": 960
|
618 |
-
},
|
619 |
-
{
|
620 |
-
"epoch": 3.54,
|
621 |
-
"learning_rate": 9.448818897637794e-05,
|
622 |
-
"loss": 0.3719,
|
623 |
-
"step": 970
|
624 |
-
},
|
625 |
-
{
|
626 |
-
"epoch": 3.57,
|
627 |
-
"learning_rate": 9.21259842519685e-05,
|
628 |
-
"loss": 0.3779,
|
629 |
-
"step": 980
|
630 |
-
},
|
631 |
-
{
|
632 |
-
"epoch": 3.61,
|
633 |
-
"learning_rate": 8.976377952755905e-05,
|
634 |
-
"loss": 0.3921,
|
635 |
-
"step": 990
|
636 |
-
},
|
637 |
-
{
|
638 |
-
"epoch": 3.65,
|
639 |
-
"learning_rate": 8.740157480314959e-05,
|
640 |
-
"loss": 0.3776,
|
641 |
-
"step": 1000
|
642 |
-
},
|
643 |
-
{
|
644 |
-
"epoch": 3.65,
|
645 |
-
"eval_loss": 0.3908761739730835,
|
646 |
-
"eval_runtime": 19.2678,
|
647 |
-
"eval_samples_per_second": 103.8,
|
648 |
-
"eval_steps_per_second": 1.661,
|
649 |
-
"step": 1000
|
650 |
-
},
|
651 |
-
{
|
652 |
-
"epoch": 3.68,
|
653 |
-
"learning_rate": 8.503937007874015e-05,
|
654 |
-
"loss": 0.3889,
|
655 |
-
"step": 1010
|
656 |
-
},
|
657 |
-
{
|
658 |
-
"epoch": 3.72,
|
659 |
-
"learning_rate": 8.26771653543307e-05,
|
660 |
-
"loss": 0.3819,
|
661 |
-
"step": 1020
|
662 |
-
},
|
663 |
-
{
|
664 |
-
"epoch": 3.76,
|
665 |
-
"learning_rate": 8.031496062992126e-05,
|
666 |
-
"loss": 0.3758,
|
667 |
-
"step": 1030
|
668 |
-
},
|
669 |
-
{
|
670 |
-
"epoch": 3.79,
|
671 |
-
"learning_rate": 7.79527559055118e-05,
|
672 |
-
"loss": 0.3753,
|
673 |
-
"step": 1040
|
674 |
-
},
|
675 |
-
{
|
676 |
-
"epoch": 3.83,
|
677 |
-
"learning_rate": 7.559055118110236e-05,
|
678 |
-
"loss": 0.3737,
|
679 |
-
"step": 1050
|
680 |
-
},
|
681 |
-
{
|
682 |
-
"epoch": 3.87,
|
683 |
-
"learning_rate": 7.322834645669291e-05,
|
684 |
-
"loss": 0.3833,
|
685 |
-
"step": 1060
|
686 |
-
},
|
687 |
-
{
|
688 |
-
"epoch": 3.9,
|
689 |
-
"learning_rate": 7.086614173228346e-05,
|
690 |
-
"loss": 0.3625,
|
691 |
-
"step": 1070
|
692 |
-
},
|
693 |
-
{
|
694 |
-
"epoch": 3.94,
|
695 |
-
"learning_rate": 6.850393700787402e-05,
|
696 |
-
"loss": 0.3809,
|
697 |
-
"step": 1080
|
698 |
-
},
|
699 |
-
{
|
700 |
-
"epoch": 3.97,
|
701 |
-
"learning_rate": 6.614173228346455e-05,
|
702 |
-
"loss": 0.3751,
|
703 |
-
"step": 1090
|
704 |
-
},
|
705 |
-
{
|
706 |
-
"epoch": 4.01,
|
707 |
-
"learning_rate": 6.377952755905512e-05,
|
708 |
-
"loss": 0.3776,
|
709 |
-
"step": 1100
|
710 |
-
},
|
711 |
-
{
|
712 |
-
"epoch": 4.05,
|
713 |
-
"learning_rate": 6.141732283464567e-05,
|
714 |
-
"loss": 0.3748,
|
715 |
-
"step": 1110
|
716 |
-
},
|
717 |
-
{
|
718 |
-
"epoch": 4.08,
|
719 |
-
"learning_rate": 5.905511811023622e-05,
|
720 |
-
"loss": 0.3636,
|
721 |
-
"step": 1120
|
722 |
-
},
|
723 |
-
{
|
724 |
-
"epoch": 4.12,
|
725 |
-
"learning_rate": 5.669291338582676e-05,
|
726 |
-
"loss": 0.372,
|
727 |
-
"step": 1130
|
728 |
-
},
|
729 |
-
{
|
730 |
-
"epoch": 4.16,
|
731 |
-
"learning_rate": 5.433070866141732e-05,
|
732 |
-
"loss": 0.3795,
|
733 |
-
"step": 1140
|
734 |
-
},
|
735 |
-
{
|
736 |
-
"epoch": 4.19,
|
737 |
-
"learning_rate": 5.196850393700787e-05,
|
738 |
-
"loss": 0.3632,
|
739 |
-
"step": 1150
|
740 |
-
},
|
741 |
-
{
|
742 |
-
"epoch": 4.23,
|
743 |
-
"learning_rate": 4.960629921259842e-05,
|
744 |
-
"loss": 0.3806,
|
745 |
-
"step": 1160
|
746 |
-
},
|
747 |
-
{
|
748 |
-
"epoch": 4.27,
|
749 |
-
"learning_rate": 4.724409448818897e-05,
|
750 |
-
"loss": 0.3732,
|
751 |
-
"step": 1170
|
752 |
-
},
|
753 |
-
{
|
754 |
-
"epoch": 4.3,
|
755 |
-
"learning_rate": 4.488188976377953e-05,
|
756 |
-
"loss": 0.3818,
|
757 |
-
"step": 1180
|
758 |
-
},
|
759 |
-
{
|
760 |
-
"epoch": 4.34,
|
761 |
-
"learning_rate": 4.2519685039370076e-05,
|
762 |
-
"loss": 0.3766,
|
763 |
-
"step": 1190
|
764 |
-
},
|
765 |
-
{
|
766 |
-
"epoch": 4.38,
|
767 |
-
"learning_rate": 4.015748031496063e-05,
|
768 |
-
"loss": 0.3587,
|
769 |
-
"step": 1200
|
770 |
-
},
|
771 |
-
{
|
772 |
-
"epoch": 4.38,
|
773 |
-
"eval_loss": 0.3883955776691437,
|
774 |
-
"eval_runtime": 19.3219,
|
775 |
-
"eval_samples_per_second": 103.51,
|
776 |
-
"eval_steps_per_second": 1.656,
|
777 |
-
"step": 1200
|
778 |
-
}
|
779 |
-
],
|
780 |
-
"logging_steps": 10,
|
781 |
-
"max_steps": 1370,
|
782 |
-
"num_train_epochs": 5,
|
783 |
-
"save_steps": 200,
|
784 |
-
"total_flos": 2.2400975031249142e+18,
|
785 |
-
"trial_name": null,
|
786 |
-
"trial_params": null
|
787 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|