End of training
Browse files- all_results.json +9 -9
- eval_results.json +5 -5
- runs/Apr07_14-53-27_9b51f9451925/events.out.tfevents.1712502568.9b51f9451925.6697.6 +3 -0
- train_results.json +4 -4
- trainer_state.json +183 -183
all_results.json
CHANGED
@@ -1,13 +1,13 @@
|
|
1 |
{
|
2 |
"epoch": 2.99,
|
3 |
-
"eval_accuracy": 0.
|
4 |
-
"eval_loss":
|
5 |
-
"eval_runtime":
|
6 |
-
"eval_samples_per_second":
|
7 |
-
"eval_steps_per_second": 3.
|
8 |
"total_flos": 1.0585264325663785e+18,
|
9 |
-
"train_loss": 0.
|
10 |
-
"train_runtime":
|
11 |
-
"train_samples_per_second":
|
12 |
-
"train_steps_per_second": 1.
|
13 |
}
|
|
|
1 |
{
|
2 |
"epoch": 2.99,
|
3 |
+
"eval_accuracy": 0.10166666666666667,
|
4 |
+
"eval_loss": NaN,
|
5 |
+
"eval_runtime": 135.6867,
|
6 |
+
"eval_samples_per_second": 110.549,
|
7 |
+
"eval_steps_per_second": 3.456,
|
8 |
"total_flos": 1.0585264325663785e+18,
|
9 |
+
"train_loss": 0.0,
|
10 |
+
"train_runtime": 811.0273,
|
11 |
+
"train_samples_per_second": 129.465,
|
12 |
+
"train_steps_per_second": 1.01
|
13 |
}
|
eval_results.json
CHANGED
@@ -1,8 +1,8 @@
|
|
1 |
{
|
2 |
"epoch": 2.99,
|
3 |
-
"eval_accuracy": 0.
|
4 |
-
"eval_loss":
|
5 |
-
"eval_runtime":
|
6 |
-
"eval_samples_per_second":
|
7 |
-
"eval_steps_per_second": 3.
|
8 |
}
|
|
|
1 |
{
|
2 |
"epoch": 2.99,
|
3 |
+
"eval_accuracy": 0.10166666666666667,
|
4 |
+
"eval_loss": NaN,
|
5 |
+
"eval_runtime": 135.6867,
|
6 |
+
"eval_samples_per_second": 110.549,
|
7 |
+
"eval_steps_per_second": 3.456
|
8 |
}
|
runs/Apr07_14-53-27_9b51f9451925/events.out.tfevents.1712502568.9b51f9451925.6697.6
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:936dea077f72a96abdf489ed31dba84951ee3bac2625c560dbeca6a6348ccae9
|
3 |
+
size 411
|
train_results.json
CHANGED
@@ -1,8 +1,8 @@
|
|
1 |
{
|
2 |
"epoch": 2.99,
|
3 |
"total_flos": 1.0585264325663785e+18,
|
4 |
-
"train_loss": 0.
|
5 |
-
"train_runtime":
|
6 |
-
"train_samples_per_second":
|
7 |
-
"train_steps_per_second": 1.
|
8 |
}
|
|
|
1 |
{
|
2 |
"epoch": 2.99,
|
3 |
"total_flos": 1.0585264325663785e+18,
|
4 |
+
"train_loss": 0.0,
|
5 |
+
"train_runtime": 811.0273,
|
6 |
+
"train_samples_per_second": 129.465,
|
7 |
+
"train_steps_per_second": 1.01
|
8 |
}
|
trainer_state.json
CHANGED
@@ -1,6 +1,6 @@
|
|
1 |
{
|
2 |
-
"best_metric": 0.
|
3 |
-
"best_model_checkpoint": "resnet-18-finetuned-cifar10/checkpoint-
|
4 |
"epoch": 2.9945155393053016,
|
5 |
"eval_steps": 500,
|
6 |
"global_step": 819,
|
@@ -10,606 +10,606 @@
|
|
10 |
"log_history": [
|
11 |
{
|
12 |
"epoch": 0.04,
|
13 |
-
"grad_norm":
|
14 |
"learning_rate": 6.0975609756097564e-06,
|
15 |
-
"loss":
|
16 |
"step": 10
|
17 |
},
|
18 |
{
|
19 |
"epoch": 0.07,
|
20 |
-
"grad_norm":
|
21 |
"learning_rate": 1.2195121951219513e-05,
|
22 |
-
"loss":
|
23 |
"step": 20
|
24 |
},
|
25 |
{
|
26 |
"epoch": 0.11,
|
27 |
-
"grad_norm":
|
28 |
"learning_rate": 1.8292682926829268e-05,
|
29 |
-
"loss":
|
30 |
"step": 30
|
31 |
},
|
32 |
{
|
33 |
"epoch": 0.15,
|
34 |
-
"grad_norm":
|
35 |
"learning_rate": 2.4390243902439026e-05,
|
36 |
-
"loss":
|
37 |
"step": 40
|
38 |
},
|
39 |
{
|
40 |
"epoch": 0.18,
|
41 |
-
"grad_norm":
|
42 |
"learning_rate": 3.048780487804878e-05,
|
43 |
-
"loss":
|
44 |
"step": 50
|
45 |
},
|
46 |
{
|
47 |
"epoch": 0.22,
|
48 |
-
"grad_norm":
|
49 |
"learning_rate": 3.6585365853658535e-05,
|
50 |
-
"loss":
|
51 |
"step": 60
|
52 |
},
|
53 |
{
|
54 |
"epoch": 0.26,
|
55 |
-
"grad_norm":
|
56 |
"learning_rate": 4.26829268292683e-05,
|
57 |
-
"loss":
|
58 |
"step": 70
|
59 |
},
|
60 |
{
|
61 |
"epoch": 0.29,
|
62 |
-
"grad_norm":
|
63 |
"learning_rate": 4.878048780487805e-05,
|
64 |
-
"loss":
|
65 |
"step": 80
|
66 |
},
|
67 |
{
|
68 |
"epoch": 0.33,
|
69 |
-
"grad_norm":
|
70 |
"learning_rate": 4.94572591587517e-05,
|
71 |
-
"loss":
|
72 |
"step": 90
|
73 |
},
|
74 |
{
|
75 |
"epoch": 0.37,
|
76 |
-
"grad_norm":
|
77 |
"learning_rate": 4.877883310719132e-05,
|
78 |
-
"loss":
|
79 |
"step": 100
|
80 |
},
|
81 |
{
|
82 |
"epoch": 0.4,
|
83 |
-
"grad_norm":
|
84 |
"learning_rate": 4.810040705563094e-05,
|
85 |
-
"loss":
|
86 |
"step": 110
|
87 |
},
|
88 |
{
|
89 |
"epoch": 0.44,
|
90 |
-
"grad_norm":
|
91 |
"learning_rate": 4.742198100407056e-05,
|
92 |
-
"loss":
|
93 |
"step": 120
|
94 |
},
|
95 |
{
|
96 |
"epoch": 0.48,
|
97 |
-
"grad_norm":
|
98 |
"learning_rate": 4.674355495251018e-05,
|
99 |
-
"loss":
|
100 |
"step": 130
|
101 |
},
|
102 |
{
|
103 |
"epoch": 0.51,
|
104 |
-
"grad_norm":
|
105 |
"learning_rate": 4.60651289009498e-05,
|
106 |
-
"loss":
|
107 |
"step": 140
|
108 |
},
|
109 |
{
|
110 |
"epoch": 0.55,
|
111 |
-
"grad_norm":
|
112 |
"learning_rate": 4.5386702849389416e-05,
|
113 |
-
"loss":
|
114 |
"step": 150
|
115 |
},
|
116 |
{
|
117 |
"epoch": 0.59,
|
118 |
-
"grad_norm":
|
119 |
"learning_rate": 4.470827679782904e-05,
|
120 |
-
"loss":
|
121 |
"step": 160
|
122 |
},
|
123 |
{
|
124 |
"epoch": 0.62,
|
125 |
-
"grad_norm":
|
126 |
"learning_rate": 4.402985074626866e-05,
|
127 |
-
"loss": 0.
|
128 |
"step": 170
|
129 |
},
|
130 |
{
|
131 |
"epoch": 0.66,
|
132 |
-
"grad_norm":
|
133 |
"learning_rate": 4.335142469470828e-05,
|
134 |
-
"loss": 0.
|
135 |
"step": 180
|
136 |
},
|
137 |
{
|
138 |
"epoch": 0.69,
|
139 |
-
"grad_norm":
|
140 |
"learning_rate": 4.26729986431479e-05,
|
141 |
-
"loss": 0.
|
142 |
"step": 190
|
143 |
},
|
144 |
{
|
145 |
"epoch": 0.73,
|
146 |
-
"grad_norm":
|
147 |
"learning_rate": 4.199457259158752e-05,
|
148 |
-
"loss": 0.
|
149 |
"step": 200
|
150 |
},
|
151 |
{
|
152 |
"epoch": 0.77,
|
153 |
-
"grad_norm":
|
154 |
"learning_rate": 4.131614654002714e-05,
|
155 |
-
"loss": 0.
|
156 |
"step": 210
|
157 |
},
|
158 |
{
|
159 |
"epoch": 0.8,
|
160 |
-
"grad_norm":
|
161 |
"learning_rate": 4.063772048846676e-05,
|
162 |
-
"loss": 0.
|
163 |
"step": 220
|
164 |
},
|
165 |
{
|
166 |
"epoch": 0.84,
|
167 |
-
"grad_norm":
|
168 |
"learning_rate": 3.995929443690638e-05,
|
169 |
-
"loss": 0.
|
170 |
"step": 230
|
171 |
},
|
172 |
{
|
173 |
"epoch": 0.88,
|
174 |
-
"grad_norm":
|
175 |
"learning_rate": 3.9280868385345995e-05,
|
176 |
-
"loss": 0.
|
177 |
"step": 240
|
178 |
},
|
179 |
{
|
180 |
"epoch": 0.91,
|
181 |
-
"grad_norm":
|
182 |
"learning_rate": 3.860244233378562e-05,
|
183 |
-
"loss": 0.
|
184 |
"step": 250
|
185 |
},
|
186 |
{
|
187 |
"epoch": 0.95,
|
188 |
-
"grad_norm":
|
189 |
"learning_rate": 3.792401628222524e-05,
|
190 |
-
"loss": 0.
|
191 |
"step": 260
|
192 |
},
|
193 |
{
|
194 |
"epoch": 0.99,
|
195 |
-
"grad_norm":
|
196 |
"learning_rate": 3.724559023066486e-05,
|
197 |
-
"loss": 0.
|
198 |
"step": 270
|
199 |
},
|
200 |
{
|
201 |
"epoch": 1.0,
|
202 |
-
"eval_accuracy": 0.
|
203 |
-
"eval_loss":
|
204 |
-
"eval_runtime":
|
205 |
-
"eval_samples_per_second":
|
206 |
-
"eval_steps_per_second": 3.
|
207 |
"step": 273
|
208 |
},
|
209 |
{
|
210 |
"epoch": 1.02,
|
211 |
-
"grad_norm":
|
212 |
"learning_rate": 3.656716417910448e-05,
|
213 |
-
"loss": 0.
|
214 |
"step": 280
|
215 |
},
|
216 |
{
|
217 |
"epoch": 1.06,
|
218 |
-
"grad_norm":
|
219 |
"learning_rate": 3.58887381275441e-05,
|
220 |
-
"loss": 0.
|
221 |
"step": 290
|
222 |
},
|
223 |
{
|
224 |
"epoch": 1.1,
|
225 |
-
"grad_norm":
|
226 |
"learning_rate": 3.521031207598372e-05,
|
227 |
-
"loss": 0.
|
228 |
"step": 300
|
229 |
},
|
230 |
{
|
231 |
"epoch": 1.13,
|
232 |
-
"grad_norm":
|
233 |
"learning_rate": 3.453188602442334e-05,
|
234 |
-
"loss": 0.
|
235 |
"step": 310
|
236 |
},
|
237 |
{
|
238 |
"epoch": 1.17,
|
239 |
-
"grad_norm":
|
240 |
"learning_rate": 3.385345997286296e-05,
|
241 |
-
"loss": 0.
|
242 |
"step": 320
|
243 |
},
|
244 |
{
|
245 |
"epoch": 1.21,
|
246 |
-
"grad_norm":
|
247 |
"learning_rate": 3.3175033921302575e-05,
|
248 |
-
"loss": 0.
|
249 |
"step": 330
|
250 |
},
|
251 |
{
|
252 |
"epoch": 1.24,
|
253 |
-
"grad_norm":
|
254 |
"learning_rate": 3.24966078697422e-05,
|
255 |
-
"loss": 0.
|
256 |
"step": 340
|
257 |
},
|
258 |
{
|
259 |
"epoch": 1.28,
|
260 |
-
"grad_norm":
|
261 |
"learning_rate": 3.181818181818182e-05,
|
262 |
-
"loss": 0.
|
263 |
"step": 350
|
264 |
},
|
265 |
{
|
266 |
"epoch": 1.32,
|
267 |
-
"grad_norm":
|
268 |
"learning_rate": 3.113975576662144e-05,
|
269 |
-
"loss": 0.
|
270 |
"step": 360
|
271 |
},
|
272 |
{
|
273 |
"epoch": 1.35,
|
274 |
-
"grad_norm":
|
275 |
"learning_rate": 3.046132971506106e-05,
|
276 |
-
"loss": 0.
|
277 |
"step": 370
|
278 |
},
|
279 |
{
|
280 |
"epoch": 1.39,
|
281 |
-
"grad_norm":
|
282 |
"learning_rate": 2.9782903663500678e-05,
|
283 |
-
"loss": 0.
|
284 |
"step": 380
|
285 |
},
|
286 |
{
|
287 |
"epoch": 1.43,
|
288 |
-
"grad_norm":
|
289 |
"learning_rate": 2.91044776119403e-05,
|
290 |
-
"loss": 0.
|
291 |
"step": 390
|
292 |
},
|
293 |
{
|
294 |
"epoch": 1.46,
|
295 |
-
"grad_norm":
|
296 |
"learning_rate": 2.842605156037992e-05,
|
297 |
-
"loss": 0.
|
298 |
"step": 400
|
299 |
},
|
300 |
{
|
301 |
"epoch": 1.5,
|
302 |
-
"grad_norm":
|
303 |
"learning_rate": 2.7747625508819542e-05,
|
304 |
-
"loss": 0.
|
305 |
"step": 410
|
306 |
},
|
307 |
{
|
308 |
"epoch": 1.54,
|
309 |
-
"grad_norm":
|
310 |
"learning_rate": 2.7069199457259158e-05,
|
311 |
-
"loss": 0.
|
312 |
"step": 420
|
313 |
},
|
314 |
{
|
315 |
"epoch": 1.57,
|
316 |
-
"grad_norm":
|
317 |
"learning_rate": 2.639077340569878e-05,
|
318 |
-
"loss": 0.
|
319 |
"step": 430
|
320 |
},
|
321 |
{
|
322 |
"epoch": 1.61,
|
323 |
-
"grad_norm":
|
324 |
"learning_rate": 2.57123473541384e-05,
|
325 |
-
"loss": 0.
|
326 |
"step": 440
|
327 |
},
|
328 |
{
|
329 |
"epoch": 1.65,
|
330 |
-
"grad_norm":
|
331 |
"learning_rate": 2.5033921302578023e-05,
|
332 |
-
"loss": 0.
|
333 |
"step": 450
|
334 |
},
|
335 |
{
|
336 |
"epoch": 1.68,
|
337 |
-
"grad_norm":
|
338 |
"learning_rate": 2.4355495251017642e-05,
|
339 |
-
"loss": 0.
|
340 |
"step": 460
|
341 |
},
|
342 |
{
|
343 |
"epoch": 1.72,
|
344 |
-
"grad_norm":
|
345 |
"learning_rate": 2.367706919945726e-05,
|
346 |
-
"loss": 0.
|
347 |
"step": 470
|
348 |
},
|
349 |
{
|
350 |
"epoch": 1.76,
|
351 |
-
"grad_norm":
|
352 |
"learning_rate": 2.299864314789688e-05,
|
353 |
-
"loss": 0.
|
354 |
"step": 480
|
355 |
},
|
356 |
{
|
357 |
"epoch": 1.79,
|
358 |
-
"grad_norm":
|
359 |
"learning_rate": 2.2320217096336503e-05,
|
360 |
-
"loss": 0.
|
361 |
"step": 490
|
362 |
},
|
363 |
{
|
364 |
"epoch": 1.83,
|
365 |
-
"grad_norm":
|
366 |
"learning_rate": 2.164179104477612e-05,
|
367 |
-
"loss": 0.
|
368 |
"step": 500
|
369 |
},
|
370 |
{
|
371 |
"epoch": 1.86,
|
372 |
-
"grad_norm":
|
373 |
"learning_rate": 2.0963364993215738e-05,
|
374 |
-
"loss": 0.
|
375 |
"step": 510
|
376 |
},
|
377 |
{
|
378 |
"epoch": 1.9,
|
379 |
-
"grad_norm":
|
380 |
"learning_rate": 2.028493894165536e-05,
|
381 |
-
"loss": 0.
|
382 |
"step": 520
|
383 |
},
|
384 |
{
|
385 |
"epoch": 1.94,
|
386 |
-
"grad_norm":
|
387 |
"learning_rate": 1.960651289009498e-05,
|
388 |
-
"loss": 0.
|
389 |
"step": 530
|
390 |
},
|
391 |
{
|
392 |
"epoch": 1.97,
|
393 |
-
"grad_norm":
|
394 |
"learning_rate": 1.89280868385346e-05,
|
395 |
-
"loss": 0.
|
396 |
"step": 540
|
397 |
},
|
398 |
{
|
399 |
"epoch": 2.0,
|
400 |
-
"eval_accuracy": 0.
|
401 |
-
"eval_loss":
|
402 |
-
"eval_runtime":
|
403 |
-
"eval_samples_per_second":
|
404 |
-
"eval_steps_per_second": 3.
|
405 |
"step": 547
|
406 |
},
|
407 |
{
|
408 |
"epoch": 2.01,
|
409 |
-
"grad_norm":
|
410 |
"learning_rate": 1.824966078697422e-05,
|
411 |
-
"loss": 0.
|
412 |
"step": 550
|
413 |
},
|
414 |
{
|
415 |
"epoch": 2.05,
|
416 |
-
"grad_norm":
|
417 |
"learning_rate": 1.757123473541384e-05,
|
418 |
-
"loss": 0.
|
419 |
"step": 560
|
420 |
},
|
421 |
{
|
422 |
"epoch": 2.08,
|
423 |
-
"grad_norm":
|
424 |
"learning_rate": 1.689280868385346e-05,
|
425 |
-
"loss": 0.
|
426 |
"step": 570
|
427 |
},
|
428 |
{
|
429 |
"epoch": 2.12,
|
430 |
-
"grad_norm":
|
431 |
"learning_rate": 1.6214382632293083e-05,
|
432 |
-
"loss": 0.
|
433 |
"step": 580
|
434 |
},
|
435 |
{
|
436 |
"epoch": 2.16,
|
437 |
-
"grad_norm":
|
438 |
"learning_rate": 1.55359565807327e-05,
|
439 |
-
"loss": 0.
|
440 |
"step": 590
|
441 |
},
|
442 |
{
|
443 |
"epoch": 2.19,
|
444 |
-
"grad_norm":
|
445 |
"learning_rate": 1.485753052917232e-05,
|
446 |
-
"loss": 0.
|
447 |
"step": 600
|
448 |
},
|
449 |
{
|
450 |
"epoch": 2.23,
|
451 |
-
"grad_norm":
|
452 |
"learning_rate": 1.417910447761194e-05,
|
453 |
-
"loss": 0.
|
454 |
"step": 610
|
455 |
},
|
456 |
{
|
457 |
"epoch": 2.27,
|
458 |
-
"grad_norm":
|
459 |
"learning_rate": 1.3500678426051561e-05,
|
460 |
-
"loss": 0.
|
461 |
"step": 620
|
462 |
},
|
463 |
{
|
464 |
"epoch": 2.3,
|
465 |
-
"grad_norm":
|
466 |
"learning_rate": 1.282225237449118e-05,
|
467 |
-
"loss": 0.
|
468 |
"step": 630
|
469 |
},
|
470 |
{
|
471 |
"epoch": 2.34,
|
472 |
-
"grad_norm":
|
473 |
"learning_rate": 1.2143826322930801e-05,
|
474 |
-
"loss": 0.
|
475 |
"step": 640
|
476 |
},
|
477 |
{
|
478 |
"epoch": 2.38,
|
479 |
-
"grad_norm":
|
480 |
"learning_rate": 1.1465400271370422e-05,
|
481 |
-
"loss": 0.
|
482 |
"step": 650
|
483 |
},
|
484 |
{
|
485 |
"epoch": 2.41,
|
486 |
-
"grad_norm":
|
487 |
"learning_rate": 1.0786974219810041e-05,
|
488 |
-
"loss": 0.
|
489 |
"step": 660
|
490 |
},
|
491 |
{
|
492 |
"epoch": 2.45,
|
493 |
-
"grad_norm":
|
494 |
"learning_rate": 1.010854816824966e-05,
|
495 |
-
"loss": 0.
|
496 |
"step": 670
|
497 |
},
|
498 |
{
|
499 |
"epoch": 2.49,
|
500 |
-
"grad_norm":
|
501 |
"learning_rate": 9.430122116689281e-06,
|
502 |
-
"loss": 0.
|
503 |
"step": 680
|
504 |
},
|
505 |
{
|
506 |
"epoch": 2.52,
|
507 |
-
"grad_norm":
|
508 |
"learning_rate": 8.751696065128902e-06,
|
509 |
-
"loss": 0.
|
510 |
"step": 690
|
511 |
},
|
512 |
{
|
513 |
"epoch": 2.56,
|
514 |
-
"grad_norm":
|
515 |
"learning_rate": 8.073270013568522e-06,
|
516 |
-
"loss": 0.
|
517 |
"step": 700
|
518 |
},
|
519 |
{
|
520 |
"epoch": 2.6,
|
521 |
-
"grad_norm":
|
522 |
"learning_rate": 7.394843962008141e-06,
|
523 |
-
"loss": 0.
|
524 |
"step": 710
|
525 |
},
|
526 |
{
|
527 |
"epoch": 2.63,
|
528 |
-
"grad_norm":
|
529 |
"learning_rate": 6.716417910447762e-06,
|
530 |
-
"loss": 0.
|
531 |
"step": 720
|
532 |
},
|
533 |
{
|
534 |
"epoch": 2.67,
|
535 |
-
"grad_norm":
|
536 |
"learning_rate": 6.037991858887382e-06,
|
537 |
-
"loss": 0.
|
538 |
"step": 730
|
539 |
},
|
540 |
{
|
541 |
"epoch": 2.71,
|
542 |
-
"grad_norm":
|
543 |
"learning_rate": 5.359565807327002e-06,
|
544 |
-
"loss": 0.
|
545 |
"step": 740
|
546 |
},
|
547 |
{
|
548 |
"epoch": 2.74,
|
549 |
-
"grad_norm":
|
550 |
"learning_rate": 4.681139755766622e-06,
|
551 |
-
"loss": 0.
|
552 |
"step": 750
|
553 |
},
|
554 |
{
|
555 |
"epoch": 2.78,
|
556 |
-
"grad_norm":
|
557 |
"learning_rate": 4.002713704206242e-06,
|
558 |
-
"loss": 0.
|
559 |
"step": 760
|
560 |
},
|
561 |
{
|
562 |
"epoch": 2.82,
|
563 |
-
"grad_norm":
|
564 |
"learning_rate": 3.324287652645862e-06,
|
565 |
-
"loss": 0.
|
566 |
"step": 770
|
567 |
},
|
568 |
{
|
569 |
"epoch": 2.85,
|
570 |
-
"grad_norm":
|
571 |
"learning_rate": 2.645861601085482e-06,
|
572 |
-
"loss": 0.
|
573 |
"step": 780
|
574 |
},
|
575 |
{
|
576 |
"epoch": 2.89,
|
577 |
-
"grad_norm":
|
578 |
"learning_rate": 1.967435549525102e-06,
|
579 |
-
"loss": 0.
|
580 |
"step": 790
|
581 |
},
|
582 |
{
|
583 |
"epoch": 2.93,
|
584 |
-
"grad_norm":
|
585 |
"learning_rate": 1.289009497964722e-06,
|
586 |
-
"loss": 0.
|
587 |
"step": 800
|
588 |
},
|
589 |
{
|
590 |
"epoch": 2.96,
|
591 |
-
"grad_norm":
|
592 |
"learning_rate": 6.10583446404342e-07,
|
593 |
-
"loss": 0.
|
594 |
"step": 810
|
595 |
},
|
596 |
{
|
597 |
"epoch": 2.99,
|
598 |
-
"eval_accuracy": 0.
|
599 |
-
"eval_loss":
|
600 |
-
"eval_runtime":
|
601 |
-
"eval_samples_per_second":
|
602 |
-
"eval_steps_per_second": 3.
|
603 |
"step": 819
|
604 |
},
|
605 |
{
|
606 |
"epoch": 2.99,
|
607 |
"step": 819,
|
608 |
"total_flos": 1.0585264325663785e+18,
|
609 |
-
"train_loss": 0.
|
610 |
-
"train_runtime":
|
611 |
-
"train_samples_per_second":
|
612 |
-
"train_steps_per_second": 1.
|
613 |
}
|
614 |
],
|
615 |
"logging_steps": 10,
|
|
|
1 |
{
|
2 |
+
"best_metric": 0.10166666666666667,
|
3 |
+
"best_model_checkpoint": "resnet-18-finetuned-cifar10/checkpoint-273",
|
4 |
"epoch": 2.9945155393053016,
|
5 |
"eval_steps": 500,
|
6 |
"global_step": 819,
|
|
|
10 |
"log_history": [
|
11 |
{
|
12 |
"epoch": 0.04,
|
13 |
+
"grad_norm": NaN,
|
14 |
"learning_rate": 6.0975609756097564e-06,
|
15 |
+
"loss": 0.0,
|
16 |
"step": 10
|
17 |
},
|
18 |
{
|
19 |
"epoch": 0.07,
|
20 |
+
"grad_norm": NaN,
|
21 |
"learning_rate": 1.2195121951219513e-05,
|
22 |
+
"loss": 0.0,
|
23 |
"step": 20
|
24 |
},
|
25 |
{
|
26 |
"epoch": 0.11,
|
27 |
+
"grad_norm": NaN,
|
28 |
"learning_rate": 1.8292682926829268e-05,
|
29 |
+
"loss": 0.0,
|
30 |
"step": 30
|
31 |
},
|
32 |
{
|
33 |
"epoch": 0.15,
|
34 |
+
"grad_norm": NaN,
|
35 |
"learning_rate": 2.4390243902439026e-05,
|
36 |
+
"loss": 0.0,
|
37 |
"step": 40
|
38 |
},
|
39 |
{
|
40 |
"epoch": 0.18,
|
41 |
+
"grad_norm": NaN,
|
42 |
"learning_rate": 3.048780487804878e-05,
|
43 |
+
"loss": 0.0,
|
44 |
"step": 50
|
45 |
},
|
46 |
{
|
47 |
"epoch": 0.22,
|
48 |
+
"grad_norm": NaN,
|
49 |
"learning_rate": 3.6585365853658535e-05,
|
50 |
+
"loss": 0.0,
|
51 |
"step": 60
|
52 |
},
|
53 |
{
|
54 |
"epoch": 0.26,
|
55 |
+
"grad_norm": NaN,
|
56 |
"learning_rate": 4.26829268292683e-05,
|
57 |
+
"loss": 0.0,
|
58 |
"step": 70
|
59 |
},
|
60 |
{
|
61 |
"epoch": 0.29,
|
62 |
+
"grad_norm": NaN,
|
63 |
"learning_rate": 4.878048780487805e-05,
|
64 |
+
"loss": 0.0,
|
65 |
"step": 80
|
66 |
},
|
67 |
{
|
68 |
"epoch": 0.33,
|
69 |
+
"grad_norm": NaN,
|
70 |
"learning_rate": 4.94572591587517e-05,
|
71 |
+
"loss": 0.0,
|
72 |
"step": 90
|
73 |
},
|
74 |
{
|
75 |
"epoch": 0.37,
|
76 |
+
"grad_norm": NaN,
|
77 |
"learning_rate": 4.877883310719132e-05,
|
78 |
+
"loss": 0.0,
|
79 |
"step": 100
|
80 |
},
|
81 |
{
|
82 |
"epoch": 0.4,
|
83 |
+
"grad_norm": NaN,
|
84 |
"learning_rate": 4.810040705563094e-05,
|
85 |
+
"loss": 0.0,
|
86 |
"step": 110
|
87 |
},
|
88 |
{
|
89 |
"epoch": 0.44,
|
90 |
+
"grad_norm": NaN,
|
91 |
"learning_rate": 4.742198100407056e-05,
|
92 |
+
"loss": 0.0,
|
93 |
"step": 120
|
94 |
},
|
95 |
{
|
96 |
"epoch": 0.48,
|
97 |
+
"grad_norm": NaN,
|
98 |
"learning_rate": 4.674355495251018e-05,
|
99 |
+
"loss": 0.0,
|
100 |
"step": 130
|
101 |
},
|
102 |
{
|
103 |
"epoch": 0.51,
|
104 |
+
"grad_norm": NaN,
|
105 |
"learning_rate": 4.60651289009498e-05,
|
106 |
+
"loss": 0.0,
|
107 |
"step": 140
|
108 |
},
|
109 |
{
|
110 |
"epoch": 0.55,
|
111 |
+
"grad_norm": NaN,
|
112 |
"learning_rate": 4.5386702849389416e-05,
|
113 |
+
"loss": 0.0,
|
114 |
"step": 150
|
115 |
},
|
116 |
{
|
117 |
"epoch": 0.59,
|
118 |
+
"grad_norm": NaN,
|
119 |
"learning_rate": 4.470827679782904e-05,
|
120 |
+
"loss": 0.0,
|
121 |
"step": 160
|
122 |
},
|
123 |
{
|
124 |
"epoch": 0.62,
|
125 |
+
"grad_norm": NaN,
|
126 |
"learning_rate": 4.402985074626866e-05,
|
127 |
+
"loss": 0.0,
|
128 |
"step": 170
|
129 |
},
|
130 |
{
|
131 |
"epoch": 0.66,
|
132 |
+
"grad_norm": NaN,
|
133 |
"learning_rate": 4.335142469470828e-05,
|
134 |
+
"loss": 0.0,
|
135 |
"step": 180
|
136 |
},
|
137 |
{
|
138 |
"epoch": 0.69,
|
139 |
+
"grad_norm": NaN,
|
140 |
"learning_rate": 4.26729986431479e-05,
|
141 |
+
"loss": 0.0,
|
142 |
"step": 190
|
143 |
},
|
144 |
{
|
145 |
"epoch": 0.73,
|
146 |
+
"grad_norm": NaN,
|
147 |
"learning_rate": 4.199457259158752e-05,
|
148 |
+
"loss": 0.0,
|
149 |
"step": 200
|
150 |
},
|
151 |
{
|
152 |
"epoch": 0.77,
|
153 |
+
"grad_norm": NaN,
|
154 |
"learning_rate": 4.131614654002714e-05,
|
155 |
+
"loss": 0.0,
|
156 |
"step": 210
|
157 |
},
|
158 |
{
|
159 |
"epoch": 0.8,
|
160 |
+
"grad_norm": NaN,
|
161 |
"learning_rate": 4.063772048846676e-05,
|
162 |
+
"loss": 0.0,
|
163 |
"step": 220
|
164 |
},
|
165 |
{
|
166 |
"epoch": 0.84,
|
167 |
+
"grad_norm": NaN,
|
168 |
"learning_rate": 3.995929443690638e-05,
|
169 |
+
"loss": 0.0,
|
170 |
"step": 230
|
171 |
},
|
172 |
{
|
173 |
"epoch": 0.88,
|
174 |
+
"grad_norm": NaN,
|
175 |
"learning_rate": 3.9280868385345995e-05,
|
176 |
+
"loss": 0.0,
|
177 |
"step": 240
|
178 |
},
|
179 |
{
|
180 |
"epoch": 0.91,
|
181 |
+
"grad_norm": NaN,
|
182 |
"learning_rate": 3.860244233378562e-05,
|
183 |
+
"loss": 0.0,
|
184 |
"step": 250
|
185 |
},
|
186 |
{
|
187 |
"epoch": 0.95,
|
188 |
+
"grad_norm": NaN,
|
189 |
"learning_rate": 3.792401628222524e-05,
|
190 |
+
"loss": 0.0,
|
191 |
"step": 260
|
192 |
},
|
193 |
{
|
194 |
"epoch": 0.99,
|
195 |
+
"grad_norm": NaN,
|
196 |
"learning_rate": 3.724559023066486e-05,
|
197 |
+
"loss": 0.0,
|
198 |
"step": 270
|
199 |
},
|
200 |
{
|
201 |
"epoch": 1.0,
|
202 |
+
"eval_accuracy": 0.10166666666666667,
|
203 |
+
"eval_loss": NaN,
|
204 |
+
"eval_runtime": 135.8532,
|
205 |
+
"eval_samples_per_second": 110.413,
|
206 |
+
"eval_steps_per_second": 3.452,
|
207 |
"step": 273
|
208 |
},
|
209 |
{
|
210 |
"epoch": 1.02,
|
211 |
+
"grad_norm": NaN,
|
212 |
"learning_rate": 3.656716417910448e-05,
|
213 |
+
"loss": 0.0,
|
214 |
"step": 280
|
215 |
},
|
216 |
{
|
217 |
"epoch": 1.06,
|
218 |
+
"grad_norm": NaN,
|
219 |
"learning_rate": 3.58887381275441e-05,
|
220 |
+
"loss": 0.0,
|
221 |
"step": 290
|
222 |
},
|
223 |
{
|
224 |
"epoch": 1.1,
|
225 |
+
"grad_norm": NaN,
|
226 |
"learning_rate": 3.521031207598372e-05,
|
227 |
+
"loss": 0.0,
|
228 |
"step": 300
|
229 |
},
|
230 |
{
|
231 |
"epoch": 1.13,
|
232 |
+
"grad_norm": NaN,
|
233 |
"learning_rate": 3.453188602442334e-05,
|
234 |
+
"loss": 0.0,
|
235 |
"step": 310
|
236 |
},
|
237 |
{
|
238 |
"epoch": 1.17,
|
239 |
+
"grad_norm": NaN,
|
240 |
"learning_rate": 3.385345997286296e-05,
|
241 |
+
"loss": 0.0,
|
242 |
"step": 320
|
243 |
},
|
244 |
{
|
245 |
"epoch": 1.21,
|
246 |
+
"grad_norm": NaN,
|
247 |
"learning_rate": 3.3175033921302575e-05,
|
248 |
+
"loss": 0.0,
|
249 |
"step": 330
|
250 |
},
|
251 |
{
|
252 |
"epoch": 1.24,
|
253 |
+
"grad_norm": NaN,
|
254 |
"learning_rate": 3.24966078697422e-05,
|
255 |
+
"loss": 0.0,
|
256 |
"step": 340
|
257 |
},
|
258 |
{
|
259 |
"epoch": 1.28,
|
260 |
+
"grad_norm": NaN,
|
261 |
"learning_rate": 3.181818181818182e-05,
|
262 |
+
"loss": 0.0,
|
263 |
"step": 350
|
264 |
},
|
265 |
{
|
266 |
"epoch": 1.32,
|
267 |
+
"grad_norm": NaN,
|
268 |
"learning_rate": 3.113975576662144e-05,
|
269 |
+
"loss": 0.0,
|
270 |
"step": 360
|
271 |
},
|
272 |
{
|
273 |
"epoch": 1.35,
|
274 |
+
"grad_norm": NaN,
|
275 |
"learning_rate": 3.046132971506106e-05,
|
276 |
+
"loss": 0.0,
|
277 |
"step": 370
|
278 |
},
|
279 |
{
|
280 |
"epoch": 1.39,
|
281 |
+
"grad_norm": NaN,
|
282 |
"learning_rate": 2.9782903663500678e-05,
|
283 |
+
"loss": 0.0,
|
284 |
"step": 380
|
285 |
},
|
286 |
{
|
287 |
"epoch": 1.43,
|
288 |
+
"grad_norm": NaN,
|
289 |
"learning_rate": 2.91044776119403e-05,
|
290 |
+
"loss": 0.0,
|
291 |
"step": 390
|
292 |
},
|
293 |
{
|
294 |
"epoch": 1.46,
|
295 |
+
"grad_norm": NaN,
|
296 |
"learning_rate": 2.842605156037992e-05,
|
297 |
+
"loss": 0.0,
|
298 |
"step": 400
|
299 |
},
|
300 |
{
|
301 |
"epoch": 1.5,
|
302 |
+
"grad_norm": NaN,
|
303 |
"learning_rate": 2.7747625508819542e-05,
|
304 |
+
"loss": 0.0,
|
305 |
"step": 410
|
306 |
},
|
307 |
{
|
308 |
"epoch": 1.54,
|
309 |
+
"grad_norm": NaN,
|
310 |
"learning_rate": 2.7069199457259158e-05,
|
311 |
+
"loss": 0.0,
|
312 |
"step": 420
|
313 |
},
|
314 |
{
|
315 |
"epoch": 1.57,
|
316 |
+
"grad_norm": NaN,
|
317 |
"learning_rate": 2.639077340569878e-05,
|
318 |
+
"loss": 0.0,
|
319 |
"step": 430
|
320 |
},
|
321 |
{
|
322 |
"epoch": 1.61,
|
323 |
+
"grad_norm": NaN,
|
324 |
"learning_rate": 2.57123473541384e-05,
|
325 |
+
"loss": 0.0,
|
326 |
"step": 440
|
327 |
},
|
328 |
{
|
329 |
"epoch": 1.65,
|
330 |
+
"grad_norm": NaN,
|
331 |
"learning_rate": 2.5033921302578023e-05,
|
332 |
+
"loss": 0.0,
|
333 |
"step": 450
|
334 |
},
|
335 |
{
|
336 |
"epoch": 1.68,
|
337 |
+
"grad_norm": NaN,
|
338 |
"learning_rate": 2.4355495251017642e-05,
|
339 |
+
"loss": 0.0,
|
340 |
"step": 460
|
341 |
},
|
342 |
{
|
343 |
"epoch": 1.72,
|
344 |
+
"grad_norm": NaN,
|
345 |
"learning_rate": 2.367706919945726e-05,
|
346 |
+
"loss": 0.0,
|
347 |
"step": 470
|
348 |
},
|
349 |
{
|
350 |
"epoch": 1.76,
|
351 |
+
"grad_norm": NaN,
|
352 |
"learning_rate": 2.299864314789688e-05,
|
353 |
+
"loss": 0.0,
|
354 |
"step": 480
|
355 |
},
|
356 |
{
|
357 |
"epoch": 1.79,
|
358 |
+
"grad_norm": NaN,
|
359 |
"learning_rate": 2.2320217096336503e-05,
|
360 |
+
"loss": 0.0,
|
361 |
"step": 490
|
362 |
},
|
363 |
{
|
364 |
"epoch": 1.83,
|
365 |
+
"grad_norm": NaN,
|
366 |
"learning_rate": 2.164179104477612e-05,
|
367 |
+
"loss": 0.0,
|
368 |
"step": 500
|
369 |
},
|
370 |
{
|
371 |
"epoch": 1.86,
|
372 |
+
"grad_norm": NaN,
|
373 |
"learning_rate": 2.0963364993215738e-05,
|
374 |
+
"loss": 0.0,
|
375 |
"step": 510
|
376 |
},
|
377 |
{
|
378 |
"epoch": 1.9,
|
379 |
+
"grad_norm": NaN,
|
380 |
"learning_rate": 2.028493894165536e-05,
|
381 |
+
"loss": 0.0,
|
382 |
"step": 520
|
383 |
},
|
384 |
{
|
385 |
"epoch": 1.94,
|
386 |
+
"grad_norm": NaN,
|
387 |
"learning_rate": 1.960651289009498e-05,
|
388 |
+
"loss": 0.0,
|
389 |
"step": 530
|
390 |
},
|
391 |
{
|
392 |
"epoch": 1.97,
|
393 |
+
"grad_norm": NaN,
|
394 |
"learning_rate": 1.89280868385346e-05,
|
395 |
+
"loss": 0.0,
|
396 |
"step": 540
|
397 |
},
|
398 |
{
|
399 |
"epoch": 2.0,
|
400 |
+
"eval_accuracy": 0.10166666666666667,
|
401 |
+
"eval_loss": NaN,
|
402 |
+
"eval_runtime": 132.0195,
|
403 |
+
"eval_samples_per_second": 113.62,
|
404 |
+
"eval_steps_per_second": 3.553,
|
405 |
"step": 547
|
406 |
},
|
407 |
{
|
408 |
"epoch": 2.01,
|
409 |
+
"grad_norm": NaN,
|
410 |
"learning_rate": 1.824966078697422e-05,
|
411 |
+
"loss": 0.0,
|
412 |
"step": 550
|
413 |
},
|
414 |
{
|
415 |
"epoch": 2.05,
|
416 |
+
"grad_norm": NaN,
|
417 |
"learning_rate": 1.757123473541384e-05,
|
418 |
+
"loss": 0.0,
|
419 |
"step": 560
|
420 |
},
|
421 |
{
|
422 |
"epoch": 2.08,
|
423 |
+
"grad_norm": NaN,
|
424 |
"learning_rate": 1.689280868385346e-05,
|
425 |
+
"loss": 0.0,
|
426 |
"step": 570
|
427 |
},
|
428 |
{
|
429 |
"epoch": 2.12,
|
430 |
+
"grad_norm": NaN,
|
431 |
"learning_rate": 1.6214382632293083e-05,
|
432 |
+
"loss": 0.0,
|
433 |
"step": 580
|
434 |
},
|
435 |
{
|
436 |
"epoch": 2.16,
|
437 |
+
"grad_norm": NaN,
|
438 |
"learning_rate": 1.55359565807327e-05,
|
439 |
+
"loss": 0.0,
|
440 |
"step": 590
|
441 |
},
|
442 |
{
|
443 |
"epoch": 2.19,
|
444 |
+
"grad_norm": NaN,
|
445 |
"learning_rate": 1.485753052917232e-05,
|
446 |
+
"loss": 0.0,
|
447 |
"step": 600
|
448 |
},
|
449 |
{
|
450 |
"epoch": 2.23,
|
451 |
+
"grad_norm": NaN,
|
452 |
"learning_rate": 1.417910447761194e-05,
|
453 |
+
"loss": 0.0,
|
454 |
"step": 610
|
455 |
},
|
456 |
{
|
457 |
"epoch": 2.27,
|
458 |
+
"grad_norm": NaN,
|
459 |
"learning_rate": 1.3500678426051561e-05,
|
460 |
+
"loss": 0.0,
|
461 |
"step": 620
|
462 |
},
|
463 |
{
|
464 |
"epoch": 2.3,
|
465 |
+
"grad_norm": NaN,
|
466 |
"learning_rate": 1.282225237449118e-05,
|
467 |
+
"loss": 0.0,
|
468 |
"step": 630
|
469 |
},
|
470 |
{
|
471 |
"epoch": 2.34,
|
472 |
+
"grad_norm": NaN,
|
473 |
"learning_rate": 1.2143826322930801e-05,
|
474 |
+
"loss": 0.0,
|
475 |
"step": 640
|
476 |
},
|
477 |
{
|
478 |
"epoch": 2.38,
|
479 |
+
"grad_norm": NaN,
|
480 |
"learning_rate": 1.1465400271370422e-05,
|
481 |
+
"loss": 0.0,
|
482 |
"step": 650
|
483 |
},
|
484 |
{
|
485 |
"epoch": 2.41,
|
486 |
+
"grad_norm": NaN,
|
487 |
"learning_rate": 1.0786974219810041e-05,
|
488 |
+
"loss": 0.0,
|
489 |
"step": 660
|
490 |
},
|
491 |
{
|
492 |
"epoch": 2.45,
|
493 |
+
"grad_norm": NaN,
|
494 |
"learning_rate": 1.010854816824966e-05,
|
495 |
+
"loss": 0.0,
|
496 |
"step": 670
|
497 |
},
|
498 |
{
|
499 |
"epoch": 2.49,
|
500 |
+
"grad_norm": NaN,
|
501 |
"learning_rate": 9.430122116689281e-06,
|
502 |
+
"loss": 0.0,
|
503 |
"step": 680
|
504 |
},
|
505 |
{
|
506 |
"epoch": 2.52,
|
507 |
+
"grad_norm": NaN,
|
508 |
"learning_rate": 8.751696065128902e-06,
|
509 |
+
"loss": 0.0,
|
510 |
"step": 690
|
511 |
},
|
512 |
{
|
513 |
"epoch": 2.56,
|
514 |
+
"grad_norm": NaN,
|
515 |
"learning_rate": 8.073270013568522e-06,
|
516 |
+
"loss": 0.0,
|
517 |
"step": 700
|
518 |
},
|
519 |
{
|
520 |
"epoch": 2.6,
|
521 |
+
"grad_norm": NaN,
|
522 |
"learning_rate": 7.394843962008141e-06,
|
523 |
+
"loss": 0.0,
|
524 |
"step": 710
|
525 |
},
|
526 |
{
|
527 |
"epoch": 2.63,
|
528 |
+
"grad_norm": NaN,
|
529 |
"learning_rate": 6.716417910447762e-06,
|
530 |
+
"loss": 0.0,
|
531 |
"step": 720
|
532 |
},
|
533 |
{
|
534 |
"epoch": 2.67,
|
535 |
+
"grad_norm": NaN,
|
536 |
"learning_rate": 6.037991858887382e-06,
|
537 |
+
"loss": 0.0,
|
538 |
"step": 730
|
539 |
},
|
540 |
{
|
541 |
"epoch": 2.71,
|
542 |
+
"grad_norm": NaN,
|
543 |
"learning_rate": 5.359565807327002e-06,
|
544 |
+
"loss": 0.0,
|
545 |
"step": 740
|
546 |
},
|
547 |
{
|
548 |
"epoch": 2.74,
|
549 |
+
"grad_norm": NaN,
|
550 |
"learning_rate": 4.681139755766622e-06,
|
551 |
+
"loss": 0.0,
|
552 |
"step": 750
|
553 |
},
|
554 |
{
|
555 |
"epoch": 2.78,
|
556 |
+
"grad_norm": NaN,
|
557 |
"learning_rate": 4.002713704206242e-06,
|
558 |
+
"loss": 0.0,
|
559 |
"step": 760
|
560 |
},
|
561 |
{
|
562 |
"epoch": 2.82,
|
563 |
+
"grad_norm": NaN,
|
564 |
"learning_rate": 3.324287652645862e-06,
|
565 |
+
"loss": 0.0,
|
566 |
"step": 770
|
567 |
},
|
568 |
{
|
569 |
"epoch": 2.85,
|
570 |
+
"grad_norm": NaN,
|
571 |
"learning_rate": 2.645861601085482e-06,
|
572 |
+
"loss": 0.0,
|
573 |
"step": 780
|
574 |
},
|
575 |
{
|
576 |
"epoch": 2.89,
|
577 |
+
"grad_norm": NaN,
|
578 |
"learning_rate": 1.967435549525102e-06,
|
579 |
+
"loss": 0.0,
|
580 |
"step": 790
|
581 |
},
|
582 |
{
|
583 |
"epoch": 2.93,
|
584 |
+
"grad_norm": NaN,
|
585 |
"learning_rate": 1.289009497964722e-06,
|
586 |
+
"loss": 0.0,
|
587 |
"step": 800
|
588 |
},
|
589 |
{
|
590 |
"epoch": 2.96,
|
591 |
+
"grad_norm": NaN,
|
592 |
"learning_rate": 6.10583446404342e-07,
|
593 |
+
"loss": 0.0,
|
594 |
"step": 810
|
595 |
},
|
596 |
{
|
597 |
"epoch": 2.99,
|
598 |
+
"eval_accuracy": 0.10166666666666667,
|
599 |
+
"eval_loss": NaN,
|
600 |
+
"eval_runtime": 130.8726,
|
601 |
+
"eval_samples_per_second": 114.615,
|
602 |
+
"eval_steps_per_second": 3.584,
|
603 |
"step": 819
|
604 |
},
|
605 |
{
|
606 |
"epoch": 2.99,
|
607 |
"step": 819,
|
608 |
"total_flos": 1.0585264325663785e+18,
|
609 |
+
"train_loss": 0.0,
|
610 |
+
"train_runtime": 811.0273,
|
611 |
+
"train_samples_per_second": 129.465,
|
612 |
+
"train_steps_per_second": 1.01
|
613 |
}
|
614 |
],
|
615 |
"logging_steps": 10,
|