Training in progress, epoch 0
Browse files- all_results.json +5 -10
- model.safetensors +1 -1
- runs/Dec14_22-06-10_bc148a6f3d4e/events.out.tfevents.1702591592.bc148a6f3d4e.163484.1 +3 -0
- train_results.json +5 -5
- trainer_state.json +143 -143
- training_args.bin +1 -1
all_results.json
CHANGED
@@ -1,13 +1,8 @@
|
|
1 |
{
|
2 |
"epoch": 2.99,
|
3 |
-
"
|
4 |
-
"
|
5 |
-
"
|
6 |
-
"
|
7 |
-
"
|
8 |
-
"total_flos": 3.964011227564194e+18,
|
9 |
-
"train_loss": 1.7593435168462042,
|
10 |
-
"train_runtime": 1295.0132,
|
11 |
-
"train_samples_per_second": 120.603,
|
12 |
-
"train_steps_per_second": 0.941
|
13 |
}
|
|
|
1 |
{
|
2 |
"epoch": 2.99,
|
3 |
+
"total_flos": 3.942238865741697e+18,
|
4 |
+
"train_loss": 2.564211890028028,
|
5 |
+
"train_runtime": 1311.9838,
|
6 |
+
"train_samples_per_second": 119.043,
|
7 |
+
"train_steps_per_second": 0.928
|
|
|
|
|
|
|
|
|
|
|
8 |
}
|
model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 112253100
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:50496fb157afe3ae180172b4762c516ef72aa5ed391148b97994b1f93e441211
|
3 |
size 112253100
|
runs/Dec14_22-06-10_bc148a6f3d4e/events.out.tfevents.1702591592.bc148a6f3d4e.163484.1
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:fc278a44dca6ce61aec84a10f37e8f8988a81df09dd0411ef1c02284fe7ac446
|
3 |
+
size 36078
|
train_results.json
CHANGED
@@ -1,8 +1,8 @@
|
|
1 |
{
|
2 |
"epoch": 2.99,
|
3 |
-
"total_flos": 3.
|
4 |
-
"train_loss":
|
5 |
-
"train_runtime":
|
6 |
-
"train_samples_per_second":
|
7 |
-
"train_steps_per_second": 0.
|
8 |
}
|
|
|
1 |
{
|
2 |
"epoch": 2.99,
|
3 |
+
"total_flos": 3.942238865741697e+18,
|
4 |
+
"train_loss": 2.564211890028028,
|
5 |
+
"train_runtime": 1311.9838,
|
6 |
+
"train_samples_per_second": 119.043,
|
7 |
+
"train_steps_per_second": 0.928
|
8 |
}
|
trainer_state.json
CHANGED
@@ -1,5 +1,5 @@
|
|
1 |
{
|
2 |
-
"best_metric": 0.
|
3 |
"best_model_checkpoint": "swin-tiny-patch4-window7-224-finetuned-eurosat/checkpoint-1218",
|
4 |
"epoch": 2.994468346650277,
|
5 |
"eval_steps": 500,
|
@@ -11,764 +11,764 @@
|
|
11 |
{
|
12 |
"epoch": 0.02,
|
13 |
"learning_rate": 4.098360655737704e-06,
|
14 |
-
"loss":
|
15 |
"step": 10
|
16 |
},
|
17 |
{
|
18 |
"epoch": 0.05,
|
19 |
"learning_rate": 8.196721311475409e-06,
|
20 |
-
"loss":
|
21 |
"step": 20
|
22 |
},
|
23 |
{
|
24 |
"epoch": 0.07,
|
25 |
"learning_rate": 1.2295081967213116e-05,
|
26 |
-
"loss":
|
27 |
"step": 30
|
28 |
},
|
29 |
{
|
30 |
"epoch": 0.1,
|
31 |
"learning_rate": 1.6393442622950818e-05,
|
32 |
-
"loss":
|
33 |
"step": 40
|
34 |
},
|
35 |
{
|
36 |
"epoch": 0.12,
|
37 |
"learning_rate": 2.0491803278688525e-05,
|
38 |
-
"loss":
|
39 |
"step": 50
|
40 |
},
|
41 |
{
|
42 |
"epoch": 0.15,
|
43 |
"learning_rate": 2.459016393442623e-05,
|
44 |
-
"loss":
|
45 |
"step": 60
|
46 |
},
|
47 |
{
|
48 |
"epoch": 0.17,
|
49 |
"learning_rate": 2.8688524590163935e-05,
|
50 |
-
"loss":
|
51 |
"step": 70
|
52 |
},
|
53 |
{
|
54 |
"epoch": 0.2,
|
55 |
"learning_rate": 3.2786885245901635e-05,
|
56 |
-
"loss":
|
57 |
"step": 80
|
58 |
},
|
59 |
{
|
60 |
"epoch": 0.22,
|
61 |
"learning_rate": 3.6885245901639346e-05,
|
62 |
-
"loss":
|
63 |
"step": 90
|
64 |
},
|
65 |
{
|
66 |
"epoch": 0.25,
|
67 |
"learning_rate": 4.098360655737705e-05,
|
68 |
-
"loss":
|
69 |
"step": 100
|
70 |
},
|
71 |
{
|
72 |
"epoch": 0.27,
|
73 |
"learning_rate": 4.508196721311476e-05,
|
74 |
-
"loss":
|
75 |
"step": 110
|
76 |
},
|
77 |
{
|
78 |
"epoch": 0.3,
|
79 |
"learning_rate": 4.918032786885246e-05,
|
80 |
-
"loss":
|
81 |
"step": 120
|
82 |
},
|
83 |
{
|
84 |
"epoch": 0.32,
|
85 |
"learning_rate": 4.963503649635037e-05,
|
86 |
-
"loss":
|
87 |
"step": 130
|
88 |
},
|
89 |
{
|
90 |
"epoch": 0.34,
|
91 |
"learning_rate": 4.9178832116788325e-05,
|
92 |
-
"loss":
|
93 |
"step": 140
|
94 |
},
|
95 |
{
|
96 |
"epoch": 0.37,
|
97 |
"learning_rate": 4.872262773722628e-05,
|
98 |
-
"loss":
|
99 |
"step": 150
|
100 |
},
|
101 |
{
|
102 |
"epoch": 0.39,
|
103 |
"learning_rate": 4.8266423357664235e-05,
|
104 |
-
"loss":
|
105 |
"step": 160
|
106 |
},
|
107 |
{
|
108 |
"epoch": 0.42,
|
109 |
"learning_rate": 4.7810218978102196e-05,
|
110 |
-
"loss":
|
111 |
"step": 170
|
112 |
},
|
113 |
{
|
114 |
"epoch": 0.44,
|
115 |
"learning_rate": 4.7354014598540144e-05,
|
116 |
-
"loss":
|
117 |
"step": 180
|
118 |
},
|
119 |
{
|
120 |
"epoch": 0.47,
|
121 |
"learning_rate": 4.6897810218978106e-05,
|
122 |
-
"loss":
|
123 |
"step": 190
|
124 |
},
|
125 |
{
|
126 |
"epoch": 0.49,
|
127 |
"learning_rate": 4.644160583941606e-05,
|
128 |
-
"loss":
|
129 |
"step": 200
|
130 |
},
|
131 |
{
|
132 |
"epoch": 0.52,
|
133 |
"learning_rate": 4.5985401459854016e-05,
|
134 |
-
"loss":
|
135 |
"step": 210
|
136 |
},
|
137 |
{
|
138 |
"epoch": 0.54,
|
139 |
"learning_rate": 4.552919708029198e-05,
|
140 |
-
"loss":
|
141 |
"step": 220
|
142 |
},
|
143 |
{
|
144 |
"epoch": 0.57,
|
145 |
"learning_rate": 4.5072992700729925e-05,
|
146 |
-
"loss":
|
147 |
"step": 230
|
148 |
},
|
149 |
{
|
150 |
"epoch": 0.59,
|
151 |
"learning_rate": 4.461678832116789e-05,
|
152 |
-
"loss":
|
153 |
"step": 240
|
154 |
},
|
155 |
{
|
156 |
"epoch": 0.61,
|
157 |
"learning_rate": 4.416058394160584e-05,
|
158 |
-
"loss":
|
159 |
"step": 250
|
160 |
},
|
161 |
{
|
162 |
"epoch": 0.64,
|
163 |
"learning_rate": 4.3704379562043796e-05,
|
164 |
-
"loss":
|
165 |
"step": 260
|
166 |
},
|
167 |
{
|
168 |
"epoch": 0.66,
|
169 |
"learning_rate": 4.324817518248175e-05,
|
170 |
-
"loss": 2.
|
171 |
"step": 270
|
172 |
},
|
173 |
{
|
174 |
"epoch": 0.69,
|
175 |
"learning_rate": 4.279197080291971e-05,
|
176 |
-
"loss": 2.
|
177 |
"step": 280
|
178 |
},
|
179 |
{
|
180 |
"epoch": 0.71,
|
181 |
"learning_rate": 4.233576642335767e-05,
|
182 |
-
"loss":
|
183 |
"step": 290
|
184 |
},
|
185 |
{
|
186 |
"epoch": 0.74,
|
187 |
"learning_rate": 4.187956204379562e-05,
|
188 |
-
"loss":
|
189 |
"step": 300
|
190 |
},
|
191 |
{
|
192 |
"epoch": 0.76,
|
193 |
"learning_rate": 4.1423357664233584e-05,
|
194 |
-
"loss":
|
195 |
"step": 310
|
196 |
},
|
197 |
{
|
198 |
"epoch": 0.79,
|
199 |
"learning_rate": 4.096715328467153e-05,
|
200 |
-
"loss": 2.
|
201 |
"step": 320
|
202 |
},
|
203 |
{
|
204 |
"epoch": 0.81,
|
205 |
"learning_rate": 4.0510948905109494e-05,
|
206 |
-
"loss":
|
207 |
"step": 330
|
208 |
},
|
209 |
{
|
210 |
"epoch": 0.84,
|
211 |
"learning_rate": 4.005474452554745e-05,
|
212 |
-
"loss":
|
213 |
"step": 340
|
214 |
},
|
215 |
{
|
216 |
"epoch": 0.86,
|
217 |
"learning_rate": 3.95985401459854e-05,
|
218 |
-
"loss":
|
219 |
"step": 350
|
220 |
},
|
221 |
{
|
222 |
"epoch": 0.89,
|
223 |
"learning_rate": 3.914233576642336e-05,
|
224 |
-
"loss":
|
225 |
"step": 360
|
226 |
},
|
227 |
{
|
228 |
"epoch": 0.91,
|
229 |
"learning_rate": 3.868613138686132e-05,
|
230 |
-
"loss":
|
231 |
"step": 370
|
232 |
},
|
233 |
{
|
234 |
"epoch": 0.93,
|
235 |
"learning_rate": 3.822992700729927e-05,
|
236 |
-
"loss":
|
237 |
"step": 380
|
238 |
},
|
239 |
{
|
240 |
"epoch": 0.96,
|
241 |
"learning_rate": 3.777372262773723e-05,
|
242 |
-
"loss":
|
243 |
"step": 390
|
244 |
},
|
245 |
{
|
246 |
"epoch": 0.98,
|
247 |
"learning_rate": 3.7317518248175184e-05,
|
248 |
-
"loss":
|
249 |
"step": 400
|
250 |
},
|
251 |
{
|
252 |
"epoch": 1.0,
|
253 |
-
"eval_accuracy": 0.
|
254 |
-
"eval_loss":
|
255 |
-
"eval_runtime":
|
256 |
-
"eval_samples_per_second":
|
257 |
-
"eval_steps_per_second": 6.
|
258 |
"step": 406
|
259 |
},
|
260 |
{
|
261 |
"epoch": 1.01,
|
262 |
"learning_rate": 3.686131386861314e-05,
|
263 |
-
"loss":
|
264 |
"step": 410
|
265 |
},
|
266 |
{
|
267 |
"epoch": 1.03,
|
268 |
"learning_rate": 3.64051094890511e-05,
|
269 |
-
"loss":
|
270 |
"step": 420
|
271 |
},
|
272 |
{
|
273 |
"epoch": 1.06,
|
274 |
"learning_rate": 3.594890510948905e-05,
|
275 |
-
"loss":
|
276 |
"step": 430
|
277 |
},
|
278 |
{
|
279 |
"epoch": 1.08,
|
280 |
"learning_rate": 3.549270072992701e-05,
|
281 |
-
"loss":
|
282 |
"step": 440
|
283 |
},
|
284 |
{
|
285 |
"epoch": 1.11,
|
286 |
"learning_rate": 3.5036496350364965e-05,
|
287 |
-
"loss":
|
288 |
"step": 450
|
289 |
},
|
290 |
{
|
291 |
"epoch": 1.13,
|
292 |
"learning_rate": 3.458029197080292e-05,
|
293 |
-
"loss":
|
294 |
"step": 460
|
295 |
},
|
296 |
{
|
297 |
"epoch": 1.16,
|
298 |
"learning_rate": 3.4124087591240875e-05,
|
299 |
-
"loss":
|
300 |
"step": 470
|
301 |
},
|
302 |
{
|
303 |
"epoch": 1.18,
|
304 |
"learning_rate": 3.3667883211678836e-05,
|
305 |
-
"loss":
|
306 |
"step": 480
|
307 |
},
|
308 |
{
|
309 |
"epoch": 1.2,
|
310 |
"learning_rate": 3.321167883211679e-05,
|
311 |
-
"loss":
|
312 |
"step": 490
|
313 |
},
|
314 |
{
|
315 |
"epoch": 1.23,
|
316 |
"learning_rate": 3.2755474452554746e-05,
|
317 |
-
"loss":
|
318 |
"step": 500
|
319 |
},
|
320 |
{
|
321 |
"epoch": 1.25,
|
322 |
"learning_rate": 3.229927007299271e-05,
|
323 |
-
"loss":
|
324 |
"step": 510
|
325 |
},
|
326 |
{
|
327 |
"epoch": 1.28,
|
328 |
"learning_rate": 3.1843065693430655e-05,
|
329 |
-
"loss":
|
330 |
"step": 520
|
331 |
},
|
332 |
{
|
333 |
"epoch": 1.3,
|
334 |
"learning_rate": 3.138686131386862e-05,
|
335 |
-
"loss":
|
336 |
"step": 530
|
337 |
},
|
338 |
{
|
339 |
"epoch": 1.33,
|
340 |
"learning_rate": 3.093065693430657e-05,
|
341 |
-
"loss":
|
342 |
"step": 540
|
343 |
},
|
344 |
{
|
345 |
"epoch": 1.35,
|
346 |
"learning_rate": 3.0474452554744527e-05,
|
347 |
-
"loss":
|
348 |
"step": 550
|
349 |
},
|
350 |
{
|
351 |
"epoch": 1.38,
|
352 |
"learning_rate": 3.0018248175182485e-05,
|
353 |
-
"loss":
|
354 |
"step": 560
|
355 |
},
|
356 |
{
|
357 |
"epoch": 1.4,
|
358 |
"learning_rate": 2.9562043795620443e-05,
|
359 |
-
"loss":
|
360 |
"step": 570
|
361 |
},
|
362 |
{
|
363 |
"epoch": 1.43,
|
364 |
"learning_rate": 2.9105839416058394e-05,
|
365 |
-
"loss":
|
366 |
"step": 580
|
367 |
},
|
368 |
{
|
369 |
"epoch": 1.45,
|
370 |
"learning_rate": 2.8649635036496353e-05,
|
371 |
-
"loss":
|
372 |
"step": 590
|
373 |
},
|
374 |
{
|
375 |
"epoch": 1.48,
|
376 |
"learning_rate": 2.8193430656934304e-05,
|
377 |
-
"loss":
|
378 |
"step": 600
|
379 |
},
|
380 |
{
|
381 |
"epoch": 1.5,
|
382 |
"learning_rate": 2.7737226277372262e-05,
|
383 |
-
"loss":
|
384 |
"step": 610
|
385 |
},
|
386 |
{
|
387 |
"epoch": 1.52,
|
388 |
"learning_rate": 2.728102189781022e-05,
|
389 |
-
"loss":
|
390 |
"step": 620
|
391 |
},
|
392 |
{
|
393 |
"epoch": 1.55,
|
394 |
"learning_rate": 2.6824817518248175e-05,
|
395 |
-
"loss": 1.
|
396 |
"step": 630
|
397 |
},
|
398 |
{
|
399 |
"epoch": 1.57,
|
400 |
"learning_rate": 2.6368613138686133e-05,
|
401 |
-
"loss":
|
402 |
"step": 640
|
403 |
},
|
404 |
{
|
405 |
"epoch": 1.6,
|
406 |
"learning_rate": 2.591240875912409e-05,
|
407 |
-
"loss":
|
408 |
"step": 650
|
409 |
},
|
410 |
{
|
411 |
"epoch": 1.62,
|
412 |
"learning_rate": 2.5456204379562043e-05,
|
413 |
-
"loss":
|
414 |
"step": 660
|
415 |
},
|
416 |
{
|
417 |
"epoch": 1.65,
|
418 |
"learning_rate": 2.5e-05,
|
419 |
-
"loss":
|
420 |
"step": 670
|
421 |
},
|
422 |
{
|
423 |
"epoch": 1.67,
|
424 |
"learning_rate": 2.4543795620437956e-05,
|
425 |
-
"loss":
|
426 |
"step": 680
|
427 |
},
|
428 |
{
|
429 |
"epoch": 1.7,
|
430 |
"learning_rate": 2.4087591240875914e-05,
|
431 |
-
"loss": 1.
|
432 |
"step": 690
|
433 |
},
|
434 |
{
|
435 |
"epoch": 1.72,
|
436 |
"learning_rate": 2.363138686131387e-05,
|
437 |
-
"loss":
|
438 |
"step": 700
|
439 |
},
|
440 |
{
|
441 |
"epoch": 1.75,
|
442 |
"learning_rate": 2.3175182481751824e-05,
|
443 |
-
"loss": 1.
|
444 |
"step": 710
|
445 |
},
|
446 |
{
|
447 |
"epoch": 1.77,
|
448 |
"learning_rate": 2.2718978102189782e-05,
|
449 |
-
"loss":
|
450 |
"step": 720
|
451 |
},
|
452 |
{
|
453 |
"epoch": 1.79,
|
454 |
"learning_rate": 2.226277372262774e-05,
|
455 |
-
"loss": 1.
|
456 |
"step": 730
|
457 |
},
|
458 |
{
|
459 |
"epoch": 1.82,
|
460 |
"learning_rate": 2.1806569343065695e-05,
|
461 |
-
"loss": 1.
|
462 |
"step": 740
|
463 |
},
|
464 |
{
|
465 |
"epoch": 1.84,
|
466 |
"learning_rate": 2.135036496350365e-05,
|
467 |
-
"loss": 1.
|
468 |
"step": 750
|
469 |
},
|
470 |
{
|
471 |
"epoch": 1.87,
|
472 |
"learning_rate": 2.0894160583941608e-05,
|
473 |
-
"loss": 1.
|
474 |
"step": 760
|
475 |
},
|
476 |
{
|
477 |
"epoch": 1.89,
|
478 |
"learning_rate": 2.0437956204379563e-05,
|
479 |
-
"loss":
|
480 |
"step": 770
|
481 |
},
|
482 |
{
|
483 |
"epoch": 1.92,
|
484 |
"learning_rate": 1.9981751824817518e-05,
|
485 |
-
"loss": 1.
|
486 |
"step": 780
|
487 |
},
|
488 |
{
|
489 |
"epoch": 1.94,
|
490 |
"learning_rate": 1.9525547445255476e-05,
|
491 |
-
"loss": 1.
|
492 |
"step": 790
|
493 |
},
|
494 |
{
|
495 |
"epoch": 1.97,
|
496 |
"learning_rate": 1.906934306569343e-05,
|
497 |
-
"loss": 1.
|
498 |
"step": 800
|
499 |
},
|
500 |
{
|
501 |
"epoch": 1.99,
|
502 |
"learning_rate": 1.861313868613139e-05,
|
503 |
-
"loss": 1.
|
504 |
"step": 810
|
505 |
},
|
506 |
{
|
507 |
"epoch": 2.0,
|
508 |
-
"eval_accuracy": 0.
|
509 |
-
"eval_loss": 1.
|
510 |
-
"eval_runtime":
|
511 |
-
"eval_samples_per_second":
|
512 |
-
"eval_steps_per_second": 6.
|
513 |
"step": 813
|
514 |
},
|
515 |
{
|
516 |
"epoch": 2.02,
|
517 |
"learning_rate": 1.8156934306569344e-05,
|
518 |
-
"loss": 1.
|
519 |
"step": 820
|
520 |
},
|
521 |
{
|
522 |
"epoch": 2.04,
|
523 |
"learning_rate": 1.7700729927007302e-05,
|
524 |
-
"loss": 1.
|
525 |
"step": 830
|
526 |
},
|
527 |
{
|
528 |
"epoch": 2.07,
|
529 |
"learning_rate": 1.7244525547445257e-05,
|
530 |
-
"loss": 1.
|
531 |
"step": 840
|
532 |
},
|
533 |
{
|
534 |
"epoch": 2.09,
|
535 |
"learning_rate": 1.678832116788321e-05,
|
536 |
-
"loss": 1.
|
537 |
"step": 850
|
538 |
},
|
539 |
{
|
540 |
"epoch": 2.11,
|
541 |
"learning_rate": 1.633211678832117e-05,
|
542 |
-
"loss": 1.
|
543 |
"step": 860
|
544 |
},
|
545 |
{
|
546 |
"epoch": 2.14,
|
547 |
"learning_rate": 1.5875912408759125e-05,
|
548 |
-
"loss": 1.
|
549 |
"step": 870
|
550 |
},
|
551 |
{
|
552 |
"epoch": 2.16,
|
553 |
"learning_rate": 1.541970802919708e-05,
|
554 |
-
"loss": 1.
|
555 |
"step": 880
|
556 |
},
|
557 |
{
|
558 |
"epoch": 2.19,
|
559 |
"learning_rate": 1.496350364963504e-05,
|
560 |
-
"loss": 1.
|
561 |
"step": 890
|
562 |
},
|
563 |
{
|
564 |
"epoch": 2.21,
|
565 |
"learning_rate": 1.4507299270072994e-05,
|
566 |
-
"loss": 1.
|
567 |
"step": 900
|
568 |
},
|
569 |
{
|
570 |
"epoch": 2.24,
|
571 |
"learning_rate": 1.4051094890510949e-05,
|
572 |
-
"loss": 1.
|
573 |
"step": 910
|
574 |
},
|
575 |
{
|
576 |
"epoch": 2.26,
|
577 |
"learning_rate": 1.3594890510948904e-05,
|
578 |
-
"loss": 1.
|
579 |
"step": 920
|
580 |
},
|
581 |
{
|
582 |
"epoch": 2.29,
|
583 |
"learning_rate": 1.3138686131386862e-05,
|
584 |
-
"loss": 1.
|
585 |
"step": 930
|
586 |
},
|
587 |
{
|
588 |
"epoch": 2.31,
|
589 |
"learning_rate": 1.2682481751824818e-05,
|
590 |
-
"loss": 1.
|
591 |
"step": 940
|
592 |
},
|
593 |
{
|
594 |
"epoch": 2.34,
|
595 |
"learning_rate": 1.2226277372262775e-05,
|
596 |
-
"loss": 1.
|
597 |
"step": 950
|
598 |
},
|
599 |
{
|
600 |
"epoch": 2.36,
|
601 |
"learning_rate": 1.177007299270073e-05,
|
602 |
-
"loss": 1.
|
603 |
"step": 960
|
604 |
},
|
605 |
{
|
606 |
"epoch": 2.38,
|
607 |
"learning_rate": 1.1313868613138686e-05,
|
608 |
-
"loss": 1.
|
609 |
"step": 970
|
610 |
},
|
611 |
{
|
612 |
"epoch": 2.41,
|
613 |
"learning_rate": 1.0857664233576643e-05,
|
614 |
-
"loss": 1.
|
615 |
"step": 980
|
616 |
},
|
617 |
{
|
618 |
"epoch": 2.43,
|
619 |
"learning_rate": 1.04014598540146e-05,
|
620 |
-
"loss": 1.
|
621 |
"step": 990
|
622 |
},
|
623 |
{
|
624 |
"epoch": 2.46,
|
625 |
"learning_rate": 9.945255474452556e-06,
|
626 |
-
"loss": 1.
|
627 |
"step": 1000
|
628 |
},
|
629 |
{
|
630 |
"epoch": 2.48,
|
631 |
"learning_rate": 9.48905109489051e-06,
|
632 |
-
"loss": 1.
|
633 |
"step": 1010
|
634 |
},
|
635 |
{
|
636 |
"epoch": 2.51,
|
637 |
"learning_rate": 9.032846715328467e-06,
|
638 |
-
"loss": 1.
|
639 |
"step": 1020
|
640 |
},
|
641 |
{
|
642 |
"epoch": 2.53,
|
643 |
"learning_rate": 8.576642335766425e-06,
|
644 |
-
"loss": 1.
|
645 |
"step": 1030
|
646 |
},
|
647 |
{
|
648 |
"epoch": 2.56,
|
649 |
"learning_rate": 8.12043795620438e-06,
|
650 |
-
"loss": 1.
|
651 |
"step": 1040
|
652 |
},
|
653 |
{
|
654 |
"epoch": 2.58,
|
655 |
"learning_rate": 7.664233576642336e-06,
|
656 |
-
"loss": 1.
|
657 |
"step": 1050
|
658 |
},
|
659 |
{
|
660 |
"epoch": 2.61,
|
661 |
"learning_rate": 7.208029197080292e-06,
|
662 |
-
"loss": 1.
|
663 |
"step": 1060
|
664 |
},
|
665 |
{
|
666 |
"epoch": 2.63,
|
667 |
"learning_rate": 6.751824817518249e-06,
|
668 |
-
"loss": 1.
|
669 |
"step": 1070
|
670 |
},
|
671 |
{
|
672 |
"epoch": 2.66,
|
673 |
"learning_rate": 6.295620437956205e-06,
|
674 |
-
"loss": 1.
|
675 |
"step": 1080
|
676 |
},
|
677 |
{
|
678 |
"epoch": 2.68,
|
679 |
"learning_rate": 5.839416058394161e-06,
|
680 |
-
"loss": 1.
|
681 |
"step": 1090
|
682 |
},
|
683 |
{
|
684 |
"epoch": 2.7,
|
685 |
"learning_rate": 5.3832116788321165e-06,
|
686 |
-
"loss": 1.
|
687 |
"step": 1100
|
688 |
},
|
689 |
{
|
690 |
"epoch": 2.73,
|
691 |
"learning_rate": 4.927007299270074e-06,
|
692 |
-
"loss": 1.
|
693 |
"step": 1110
|
694 |
},
|
695 |
{
|
696 |
"epoch": 2.75,
|
697 |
"learning_rate": 4.4708029197080295e-06,
|
698 |
-
"loss": 1.
|
699 |
"step": 1120
|
700 |
},
|
701 |
{
|
702 |
"epoch": 2.78,
|
703 |
"learning_rate": 4.014598540145985e-06,
|
704 |
-
"loss": 1.
|
705 |
"step": 1130
|
706 |
},
|
707 |
{
|
708 |
"epoch": 2.8,
|
709 |
"learning_rate": 3.5583941605839416e-06,
|
710 |
-
"loss": 1.
|
711 |
"step": 1140
|
712 |
},
|
713 |
{
|
714 |
"epoch": 2.83,
|
715 |
"learning_rate": 3.102189781021898e-06,
|
716 |
-
"loss": 1.
|
717 |
"step": 1150
|
718 |
},
|
719 |
{
|
720 |
"epoch": 2.85,
|
721 |
"learning_rate": 2.6459854014598542e-06,
|
722 |
-
"loss": 1.
|
723 |
"step": 1160
|
724 |
},
|
725 |
{
|
726 |
"epoch": 2.88,
|
727 |
"learning_rate": 2.1897810218978103e-06,
|
728 |
-
"loss": 1.
|
729 |
"step": 1170
|
730 |
},
|
731 |
{
|
732 |
"epoch": 2.9,
|
733 |
"learning_rate": 1.7335766423357664e-06,
|
734 |
-
"loss": 1.
|
735 |
"step": 1180
|
736 |
},
|
737 |
{
|
738 |
"epoch": 2.93,
|
739 |
"learning_rate": 1.2773722627737227e-06,
|
740 |
-
"loss": 1.
|
741 |
"step": 1190
|
742 |
},
|
743 |
{
|
744 |
"epoch": 2.95,
|
745 |
"learning_rate": 8.211678832116789e-07,
|
746 |
-
"loss": 1.
|
747 |
"step": 1200
|
748 |
},
|
749 |
{
|
750 |
"epoch": 2.97,
|
751 |
"learning_rate": 3.6496350364963505e-07,
|
752 |
-
"loss": 1.
|
753 |
"step": 1210
|
754 |
},
|
755 |
{
|
756 |
"epoch": 2.99,
|
757 |
-
"eval_accuracy": 0.
|
758 |
-
"eval_loss":
|
759 |
-
"eval_runtime":
|
760 |
-
"eval_samples_per_second":
|
761 |
-
"eval_steps_per_second": 6.
|
762 |
"step": 1218
|
763 |
},
|
764 |
{
|
765 |
"epoch": 2.99,
|
766 |
"step": 1218,
|
767 |
-
"total_flos": 3.
|
768 |
-
"train_loss":
|
769 |
-
"train_runtime":
|
770 |
-
"train_samples_per_second":
|
771 |
-
"train_steps_per_second": 0.
|
772 |
}
|
773 |
],
|
774 |
"logging_steps": 10,
|
@@ -776,7 +776,7 @@
|
|
776 |
"num_input_tokens_seen": 0,
|
777 |
"num_train_epochs": 3,
|
778 |
"save_steps": 500,
|
779 |
-
"total_flos": 3.
|
780 |
"train_batch_size": 32,
|
781 |
"trial_name": null,
|
782 |
"trial_params": null
|
|
|
1 |
{
|
2 |
+
"best_metric": 0.7496974935177182,
|
3 |
"best_model_checkpoint": "swin-tiny-patch4-window7-224-finetuned-eurosat/checkpoint-1218",
|
4 |
"epoch": 2.994468346650277,
|
5 |
"eval_steps": 500,
|
|
|
11 |
{
|
12 |
"epoch": 0.02,
|
13 |
"learning_rate": 4.098360655737704e-06,
|
14 |
+
"loss": 6.433,
|
15 |
"step": 10
|
16 |
},
|
17 |
{
|
18 |
"epoch": 0.05,
|
19 |
"learning_rate": 8.196721311475409e-06,
|
20 |
+
"loss": 6.416,
|
21 |
"step": 20
|
22 |
},
|
23 |
{
|
24 |
"epoch": 0.07,
|
25 |
"learning_rate": 1.2295081967213116e-05,
|
26 |
+
"loss": 6.3382,
|
27 |
"step": 30
|
28 |
},
|
29 |
{
|
30 |
"epoch": 0.1,
|
31 |
"learning_rate": 1.6393442622950818e-05,
|
32 |
+
"loss": 6.1258,
|
33 |
"step": 40
|
34 |
},
|
35 |
{
|
36 |
"epoch": 0.12,
|
37 |
"learning_rate": 2.0491803278688525e-05,
|
38 |
+
"loss": 5.8421,
|
39 |
"step": 50
|
40 |
},
|
41 |
{
|
42 |
"epoch": 0.15,
|
43 |
"learning_rate": 2.459016393442623e-05,
|
44 |
+
"loss": 5.5024,
|
45 |
"step": 60
|
46 |
},
|
47 |
{
|
48 |
"epoch": 0.17,
|
49 |
"learning_rate": 2.8688524590163935e-05,
|
50 |
+
"loss": 5.1467,
|
51 |
"step": 70
|
52 |
},
|
53 |
{
|
54 |
"epoch": 0.2,
|
55 |
"learning_rate": 3.2786885245901635e-05,
|
56 |
+
"loss": 4.7367,
|
57 |
"step": 80
|
58 |
},
|
59 |
{
|
60 |
"epoch": 0.22,
|
61 |
"learning_rate": 3.6885245901639346e-05,
|
62 |
+
"loss": 4.4434,
|
63 |
"step": 90
|
64 |
},
|
65 |
{
|
66 |
"epoch": 0.25,
|
67 |
"learning_rate": 4.098360655737705e-05,
|
68 |
+
"loss": 4.2133,
|
69 |
"step": 100
|
70 |
},
|
71 |
{
|
72 |
"epoch": 0.27,
|
73 |
"learning_rate": 4.508196721311476e-05,
|
74 |
+
"loss": 4.0056,
|
75 |
"step": 110
|
76 |
},
|
77 |
{
|
78 |
"epoch": 0.3,
|
79 |
"learning_rate": 4.918032786885246e-05,
|
80 |
+
"loss": 4.0001,
|
81 |
"step": 120
|
82 |
},
|
83 |
{
|
84 |
"epoch": 0.32,
|
85 |
"learning_rate": 4.963503649635037e-05,
|
86 |
+
"loss": 3.9115,
|
87 |
"step": 130
|
88 |
},
|
89 |
{
|
90 |
"epoch": 0.34,
|
91 |
"learning_rate": 4.9178832116788325e-05,
|
92 |
+
"loss": 3.909,
|
93 |
"step": 140
|
94 |
},
|
95 |
{
|
96 |
"epoch": 0.37,
|
97 |
"learning_rate": 4.872262773722628e-05,
|
98 |
+
"loss": 3.8147,
|
99 |
"step": 150
|
100 |
},
|
101 |
{
|
102 |
"epoch": 0.39,
|
103 |
"learning_rate": 4.8266423357664235e-05,
|
104 |
+
"loss": 3.6984,
|
105 |
"step": 160
|
106 |
},
|
107 |
{
|
108 |
"epoch": 0.42,
|
109 |
"learning_rate": 4.7810218978102196e-05,
|
110 |
+
"loss": 3.5625,
|
111 |
"step": 170
|
112 |
},
|
113 |
{
|
114 |
"epoch": 0.44,
|
115 |
"learning_rate": 4.7354014598540144e-05,
|
116 |
+
"loss": 3.5524,
|
117 |
"step": 180
|
118 |
},
|
119 |
{
|
120 |
"epoch": 0.47,
|
121 |
"learning_rate": 4.6897810218978106e-05,
|
122 |
+
"loss": 3.4258,
|
123 |
"step": 190
|
124 |
},
|
125 |
{
|
126 |
"epoch": 0.49,
|
127 |
"learning_rate": 4.644160583941606e-05,
|
128 |
+
"loss": 3.4335,
|
129 |
"step": 200
|
130 |
},
|
131 |
{
|
132 |
"epoch": 0.52,
|
133 |
"learning_rate": 4.5985401459854016e-05,
|
134 |
+
"loss": 3.4224,
|
135 |
"step": 210
|
136 |
},
|
137 |
{
|
138 |
"epoch": 0.54,
|
139 |
"learning_rate": 4.552919708029198e-05,
|
140 |
+
"loss": 3.2971,
|
141 |
"step": 220
|
142 |
},
|
143 |
{
|
144 |
"epoch": 0.57,
|
145 |
"learning_rate": 4.5072992700729925e-05,
|
146 |
+
"loss": 3.3419,
|
147 |
"step": 230
|
148 |
},
|
149 |
{
|
150 |
"epoch": 0.59,
|
151 |
"learning_rate": 4.461678832116789e-05,
|
152 |
+
"loss": 3.22,
|
153 |
"step": 240
|
154 |
},
|
155 |
{
|
156 |
"epoch": 0.61,
|
157 |
"learning_rate": 4.416058394160584e-05,
|
158 |
+
"loss": 3.2559,
|
159 |
"step": 250
|
160 |
},
|
161 |
{
|
162 |
"epoch": 0.64,
|
163 |
"learning_rate": 4.3704379562043796e-05,
|
164 |
+
"loss": 3.2359,
|
165 |
"step": 260
|
166 |
},
|
167 |
{
|
168 |
"epoch": 0.66,
|
169 |
"learning_rate": 4.324817518248175e-05,
|
170 |
+
"loss": 2.9756,
|
171 |
"step": 270
|
172 |
},
|
173 |
{
|
174 |
"epoch": 0.69,
|
175 |
"learning_rate": 4.279197080291971e-05,
|
176 |
+
"loss": 2.9418,
|
177 |
"step": 280
|
178 |
},
|
179 |
{
|
180 |
"epoch": 0.71,
|
181 |
"learning_rate": 4.233576642335767e-05,
|
182 |
+
"loss": 3.0601,
|
183 |
"step": 290
|
184 |
},
|
185 |
{
|
186 |
"epoch": 0.74,
|
187 |
"learning_rate": 4.187956204379562e-05,
|
188 |
+
"loss": 3.0173,
|
189 |
"step": 300
|
190 |
},
|
191 |
{
|
192 |
"epoch": 0.76,
|
193 |
"learning_rate": 4.1423357664233584e-05,
|
194 |
+
"loss": 2.9081,
|
195 |
"step": 310
|
196 |
},
|
197 |
{
|
198 |
"epoch": 0.79,
|
199 |
"learning_rate": 4.096715328467153e-05,
|
200 |
+
"loss": 2.7525,
|
201 |
"step": 320
|
202 |
},
|
203 |
{
|
204 |
"epoch": 0.81,
|
205 |
"learning_rate": 4.0510948905109494e-05,
|
206 |
+
"loss": 2.774,
|
207 |
"step": 330
|
208 |
},
|
209 |
{
|
210 |
"epoch": 0.84,
|
211 |
"learning_rate": 4.005474452554745e-05,
|
212 |
+
"loss": 2.7805,
|
213 |
"step": 340
|
214 |
},
|
215 |
{
|
216 |
"epoch": 0.86,
|
217 |
"learning_rate": 3.95985401459854e-05,
|
218 |
+
"loss": 2.6359,
|
219 |
"step": 350
|
220 |
},
|
221 |
{
|
222 |
"epoch": 0.89,
|
223 |
"learning_rate": 3.914233576642336e-05,
|
224 |
+
"loss": 2.7126,
|
225 |
"step": 360
|
226 |
},
|
227 |
{
|
228 |
"epoch": 0.91,
|
229 |
"learning_rate": 3.868613138686132e-05,
|
230 |
+
"loss": 2.5696,
|
231 |
"step": 370
|
232 |
},
|
233 |
{
|
234 |
"epoch": 0.93,
|
235 |
"learning_rate": 3.822992700729927e-05,
|
236 |
+
"loss": 2.5396,
|
237 |
"step": 380
|
238 |
},
|
239 |
{
|
240 |
"epoch": 0.96,
|
241 |
"learning_rate": 3.777372262773723e-05,
|
242 |
+
"loss": 2.5119,
|
243 |
"step": 390
|
244 |
},
|
245 |
{
|
246 |
"epoch": 0.98,
|
247 |
"learning_rate": 3.7317518248175184e-05,
|
248 |
+
"loss": 2.5294,
|
249 |
"step": 400
|
250 |
},
|
251 |
{
|
252 |
"epoch": 1.0,
|
253 |
+
"eval_accuracy": 0.5865168539325842,
|
254 |
+
"eval_loss": 2.107146739959717,
|
255 |
+
"eval_runtime": 28.2223,
|
256 |
+
"eval_samples_per_second": 204.98,
|
257 |
+
"eval_steps_per_second": 6.413,
|
258 |
"step": 406
|
259 |
},
|
260 |
{
|
261 |
"epoch": 1.01,
|
262 |
"learning_rate": 3.686131386861314e-05,
|
263 |
+
"loss": 2.5622,
|
264 |
"step": 410
|
265 |
},
|
266 |
{
|
267 |
"epoch": 1.03,
|
268 |
"learning_rate": 3.64051094890511e-05,
|
269 |
+
"loss": 2.4528,
|
270 |
"step": 420
|
271 |
},
|
272 |
{
|
273 |
"epoch": 1.06,
|
274 |
"learning_rate": 3.594890510948905e-05,
|
275 |
+
"loss": 2.4302,
|
276 |
"step": 430
|
277 |
},
|
278 |
{
|
279 |
"epoch": 1.08,
|
280 |
"learning_rate": 3.549270072992701e-05,
|
281 |
+
"loss": 2.3664,
|
282 |
"step": 440
|
283 |
},
|
284 |
{
|
285 |
"epoch": 1.11,
|
286 |
"learning_rate": 3.5036496350364965e-05,
|
287 |
+
"loss": 2.3382,
|
288 |
"step": 450
|
289 |
},
|
290 |
{
|
291 |
"epoch": 1.13,
|
292 |
"learning_rate": 3.458029197080292e-05,
|
293 |
+
"loss": 2.386,
|
294 |
"step": 460
|
295 |
},
|
296 |
{
|
297 |
"epoch": 1.16,
|
298 |
"learning_rate": 3.4124087591240875e-05,
|
299 |
+
"loss": 2.3621,
|
300 |
"step": 470
|
301 |
},
|
302 |
{
|
303 |
"epoch": 1.18,
|
304 |
"learning_rate": 3.3667883211678836e-05,
|
305 |
+
"loss": 2.3706,
|
306 |
"step": 480
|
307 |
},
|
308 |
{
|
309 |
"epoch": 1.2,
|
310 |
"learning_rate": 3.321167883211679e-05,
|
311 |
+
"loss": 2.3005,
|
312 |
"step": 490
|
313 |
},
|
314 |
{
|
315 |
"epoch": 1.23,
|
316 |
"learning_rate": 3.2755474452554746e-05,
|
317 |
+
"loss": 2.3632,
|
318 |
"step": 500
|
319 |
},
|
320 |
{
|
321 |
"epoch": 1.25,
|
322 |
"learning_rate": 3.229927007299271e-05,
|
323 |
+
"loss": 2.2058,
|
324 |
"step": 510
|
325 |
},
|
326 |
{
|
327 |
"epoch": 1.28,
|
328 |
"learning_rate": 3.1843065693430655e-05,
|
329 |
+
"loss": 2.231,
|
330 |
"step": 520
|
331 |
},
|
332 |
{
|
333 |
"epoch": 1.3,
|
334 |
"learning_rate": 3.138686131386862e-05,
|
335 |
+
"loss": 2.3454,
|
336 |
"step": 530
|
337 |
},
|
338 |
{
|
339 |
"epoch": 1.33,
|
340 |
"learning_rate": 3.093065693430657e-05,
|
341 |
+
"loss": 2.1551,
|
342 |
"step": 540
|
343 |
},
|
344 |
{
|
345 |
"epoch": 1.35,
|
346 |
"learning_rate": 3.0474452554744527e-05,
|
347 |
+
"loss": 2.1461,
|
348 |
"step": 550
|
349 |
},
|
350 |
{
|
351 |
"epoch": 1.38,
|
352 |
"learning_rate": 3.0018248175182485e-05,
|
353 |
+
"loss": 2.2473,
|
354 |
"step": 560
|
355 |
},
|
356 |
{
|
357 |
"epoch": 1.4,
|
358 |
"learning_rate": 2.9562043795620443e-05,
|
359 |
+
"loss": 2.2178,
|
360 |
"step": 570
|
361 |
},
|
362 |
{
|
363 |
"epoch": 1.43,
|
364 |
"learning_rate": 2.9105839416058394e-05,
|
365 |
+
"loss": 2.0885,
|
366 |
"step": 580
|
367 |
},
|
368 |
{
|
369 |
"epoch": 1.45,
|
370 |
"learning_rate": 2.8649635036496353e-05,
|
371 |
+
"loss": 2.1198,
|
372 |
"step": 590
|
373 |
},
|
374 |
{
|
375 |
"epoch": 1.48,
|
376 |
"learning_rate": 2.8193430656934304e-05,
|
377 |
+
"loss": 2.1566,
|
378 |
"step": 600
|
379 |
},
|
380 |
{
|
381 |
"epoch": 1.5,
|
382 |
"learning_rate": 2.7737226277372262e-05,
|
383 |
+
"loss": 2.2305,
|
384 |
"step": 610
|
385 |
},
|
386 |
{
|
387 |
"epoch": 1.52,
|
388 |
"learning_rate": 2.728102189781022e-05,
|
389 |
+
"loss": 2.0076,
|
390 |
"step": 620
|
391 |
},
|
392 |
{
|
393 |
"epoch": 1.55,
|
394 |
"learning_rate": 2.6824817518248175e-05,
|
395 |
+
"loss": 1.9865,
|
396 |
"step": 630
|
397 |
},
|
398 |
{
|
399 |
"epoch": 1.57,
|
400 |
"learning_rate": 2.6368613138686133e-05,
|
401 |
+
"loss": 2.151,
|
402 |
"step": 640
|
403 |
},
|
404 |
{
|
405 |
"epoch": 1.6,
|
406 |
"learning_rate": 2.591240875912409e-05,
|
407 |
+
"loss": 2.0612,
|
408 |
"step": 650
|
409 |
},
|
410 |
{
|
411 |
"epoch": 1.62,
|
412 |
"learning_rate": 2.5456204379562043e-05,
|
413 |
+
"loss": 2.0431,
|
414 |
"step": 660
|
415 |
},
|
416 |
{
|
417 |
"epoch": 1.65,
|
418 |
"learning_rate": 2.5e-05,
|
419 |
+
"loss": 2.0422,
|
420 |
"step": 670
|
421 |
},
|
422 |
{
|
423 |
"epoch": 1.67,
|
424 |
"learning_rate": 2.4543795620437956e-05,
|
425 |
+
"loss": 2.0565,
|
426 |
"step": 680
|
427 |
},
|
428 |
{
|
429 |
"epoch": 1.7,
|
430 |
"learning_rate": 2.4087591240875914e-05,
|
431 |
+
"loss": 1.9515,
|
432 |
"step": 690
|
433 |
},
|
434 |
{
|
435 |
"epoch": 1.72,
|
436 |
"learning_rate": 2.363138686131387e-05,
|
437 |
+
"loss": 2.1044,
|
438 |
"step": 700
|
439 |
},
|
440 |
{
|
441 |
"epoch": 1.75,
|
442 |
"learning_rate": 2.3175182481751824e-05,
|
443 |
+
"loss": 1.9952,
|
444 |
"step": 710
|
445 |
},
|
446 |
{
|
447 |
"epoch": 1.77,
|
448 |
"learning_rate": 2.2718978102189782e-05,
|
449 |
+
"loss": 2.0582,
|
450 |
"step": 720
|
451 |
},
|
452 |
{
|
453 |
"epoch": 1.79,
|
454 |
"learning_rate": 2.226277372262774e-05,
|
455 |
+
"loss": 1.9372,
|
456 |
"step": 730
|
457 |
},
|
458 |
{
|
459 |
"epoch": 1.82,
|
460 |
"learning_rate": 2.1806569343065695e-05,
|
461 |
+
"loss": 1.9758,
|
462 |
"step": 740
|
463 |
},
|
464 |
{
|
465 |
"epoch": 1.84,
|
466 |
"learning_rate": 2.135036496350365e-05,
|
467 |
+
"loss": 1.9974,
|
468 |
"step": 750
|
469 |
},
|
470 |
{
|
471 |
"epoch": 1.87,
|
472 |
"learning_rate": 2.0894160583941608e-05,
|
473 |
+
"loss": 1.9725,
|
474 |
"step": 760
|
475 |
},
|
476 |
{
|
477 |
"epoch": 1.89,
|
478 |
"learning_rate": 2.0437956204379563e-05,
|
479 |
+
"loss": 2.0613,
|
480 |
"step": 770
|
481 |
},
|
482 |
{
|
483 |
"epoch": 1.92,
|
484 |
"learning_rate": 1.9981751824817518e-05,
|
485 |
+
"loss": 1.8908,
|
486 |
"step": 780
|
487 |
},
|
488 |
{
|
489 |
"epoch": 1.94,
|
490 |
"learning_rate": 1.9525547445255476e-05,
|
491 |
+
"loss": 1.8355,
|
492 |
"step": 790
|
493 |
},
|
494 |
{
|
495 |
"epoch": 1.97,
|
496 |
"learning_rate": 1.906934306569343e-05,
|
497 |
+
"loss": 1.8955,
|
498 |
"step": 800
|
499 |
},
|
500 |
{
|
501 |
"epoch": 1.99,
|
502 |
"learning_rate": 1.861313868613139e-05,
|
503 |
+
"loss": 1.9362,
|
504 |
"step": 810
|
505 |
},
|
506 |
{
|
507 |
"epoch": 2.0,
|
508 |
+
"eval_accuracy": 0.7218668971477961,
|
509 |
+
"eval_loss": 1.4541077613830566,
|
510 |
+
"eval_runtime": 27.9276,
|
511 |
+
"eval_samples_per_second": 207.143,
|
512 |
+
"eval_steps_per_second": 6.481,
|
513 |
"step": 813
|
514 |
},
|
515 |
{
|
516 |
"epoch": 2.02,
|
517 |
"learning_rate": 1.8156934306569344e-05,
|
518 |
+
"loss": 1.8781,
|
519 |
"step": 820
|
520 |
},
|
521 |
{
|
522 |
"epoch": 2.04,
|
523 |
"learning_rate": 1.7700729927007302e-05,
|
524 |
+
"loss": 1.9182,
|
525 |
"step": 830
|
526 |
},
|
527 |
{
|
528 |
"epoch": 2.07,
|
529 |
"learning_rate": 1.7244525547445257e-05,
|
530 |
+
"loss": 1.8944,
|
531 |
"step": 840
|
532 |
},
|
533 |
{
|
534 |
"epoch": 2.09,
|
535 |
"learning_rate": 1.678832116788321e-05,
|
536 |
+
"loss": 1.8667,
|
537 |
"step": 850
|
538 |
},
|
539 |
{
|
540 |
"epoch": 2.11,
|
541 |
"learning_rate": 1.633211678832117e-05,
|
542 |
+
"loss": 1.879,
|
543 |
"step": 860
|
544 |
},
|
545 |
{
|
546 |
"epoch": 2.14,
|
547 |
"learning_rate": 1.5875912408759125e-05,
|
548 |
+
"loss": 1.8376,
|
549 |
"step": 870
|
550 |
},
|
551 |
{
|
552 |
"epoch": 2.16,
|
553 |
"learning_rate": 1.541970802919708e-05,
|
554 |
+
"loss": 1.8446,
|
555 |
"step": 880
|
556 |
},
|
557 |
{
|
558 |
"epoch": 2.19,
|
559 |
"learning_rate": 1.496350364963504e-05,
|
560 |
+
"loss": 1.8271,
|
561 |
"step": 890
|
562 |
},
|
563 |
{
|
564 |
"epoch": 2.21,
|
565 |
"learning_rate": 1.4507299270072994e-05,
|
566 |
+
"loss": 1.8052,
|
567 |
"step": 900
|
568 |
},
|
569 |
{
|
570 |
"epoch": 2.24,
|
571 |
"learning_rate": 1.4051094890510949e-05,
|
572 |
+
"loss": 1.7987,
|
573 |
"step": 910
|
574 |
},
|
575 |
{
|
576 |
"epoch": 2.26,
|
577 |
"learning_rate": 1.3594890510948904e-05,
|
578 |
+
"loss": 1.9407,
|
579 |
"step": 920
|
580 |
},
|
581 |
{
|
582 |
"epoch": 2.29,
|
583 |
"learning_rate": 1.3138686131386862e-05,
|
584 |
+
"loss": 1.9835,
|
585 |
"step": 930
|
586 |
},
|
587 |
{
|
588 |
"epoch": 2.31,
|
589 |
"learning_rate": 1.2682481751824818e-05,
|
590 |
+
"loss": 1.7249,
|
591 |
"step": 940
|
592 |
},
|
593 |
{
|
594 |
"epoch": 2.34,
|
595 |
"learning_rate": 1.2226277372262775e-05,
|
596 |
+
"loss": 1.7717,
|
597 |
"step": 950
|
598 |
},
|
599 |
{
|
600 |
"epoch": 2.36,
|
601 |
"learning_rate": 1.177007299270073e-05,
|
602 |
+
"loss": 1.7976,
|
603 |
"step": 960
|
604 |
},
|
605 |
{
|
606 |
"epoch": 2.38,
|
607 |
"learning_rate": 1.1313868613138686e-05,
|
608 |
+
"loss": 1.7857,
|
609 |
"step": 970
|
610 |
},
|
611 |
{
|
612 |
"epoch": 2.41,
|
613 |
"learning_rate": 1.0857664233576643e-05,
|
614 |
+
"loss": 1.8899,
|
615 |
"step": 980
|
616 |
},
|
617 |
{
|
618 |
"epoch": 2.43,
|
619 |
"learning_rate": 1.04014598540146e-05,
|
620 |
+
"loss": 1.7449,
|
621 |
"step": 990
|
622 |
},
|
623 |
{
|
624 |
"epoch": 2.46,
|
625 |
"learning_rate": 9.945255474452556e-06,
|
626 |
+
"loss": 1.7657,
|
627 |
"step": 1000
|
628 |
},
|
629 |
{
|
630 |
"epoch": 2.48,
|
631 |
"learning_rate": 9.48905109489051e-06,
|
632 |
+
"loss": 1.8645,
|
633 |
"step": 1010
|
634 |
},
|
635 |
{
|
636 |
"epoch": 2.51,
|
637 |
"learning_rate": 9.032846715328467e-06,
|
638 |
+
"loss": 1.8106,
|
639 |
"step": 1020
|
640 |
},
|
641 |
{
|
642 |
"epoch": 2.53,
|
643 |
"learning_rate": 8.576642335766425e-06,
|
644 |
+
"loss": 1.8082,
|
645 |
"step": 1030
|
646 |
},
|
647 |
{
|
648 |
"epoch": 2.56,
|
649 |
"learning_rate": 8.12043795620438e-06,
|
650 |
+
"loss": 1.5872,
|
651 |
"step": 1040
|
652 |
},
|
653 |
{
|
654 |
"epoch": 2.58,
|
655 |
"learning_rate": 7.664233576642336e-06,
|
656 |
+
"loss": 1.8594,
|
657 |
"step": 1050
|
658 |
},
|
659 |
{
|
660 |
"epoch": 2.61,
|
661 |
"learning_rate": 7.208029197080292e-06,
|
662 |
+
"loss": 1.7495,
|
663 |
"step": 1060
|
664 |
},
|
665 |
{
|
666 |
"epoch": 2.63,
|
667 |
"learning_rate": 6.751824817518249e-06,
|
668 |
+
"loss": 1.6744,
|
669 |
"step": 1070
|
670 |
},
|
671 |
{
|
672 |
"epoch": 2.66,
|
673 |
"learning_rate": 6.295620437956205e-06,
|
674 |
+
"loss": 1.6667,
|
675 |
"step": 1080
|
676 |
},
|
677 |
{
|
678 |
"epoch": 2.68,
|
679 |
"learning_rate": 5.839416058394161e-06,
|
680 |
+
"loss": 1.7831,
|
681 |
"step": 1090
|
682 |
},
|
683 |
{
|
684 |
"epoch": 2.7,
|
685 |
"learning_rate": 5.3832116788321165e-06,
|
686 |
+
"loss": 1.6088,
|
687 |
"step": 1100
|
688 |
},
|
689 |
{
|
690 |
"epoch": 2.73,
|
691 |
"learning_rate": 4.927007299270074e-06,
|
692 |
+
"loss": 1.6938,
|
693 |
"step": 1110
|
694 |
},
|
695 |
{
|
696 |
"epoch": 2.75,
|
697 |
"learning_rate": 4.4708029197080295e-06,
|
698 |
+
"loss": 1.794,
|
699 |
"step": 1120
|
700 |
},
|
701 |
{
|
702 |
"epoch": 2.78,
|
703 |
"learning_rate": 4.014598540145985e-06,
|
704 |
+
"loss": 1.8384,
|
705 |
"step": 1130
|
706 |
},
|
707 |
{
|
708 |
"epoch": 2.8,
|
709 |
"learning_rate": 3.5583941605839416e-06,
|
710 |
+
"loss": 1.6364,
|
711 |
"step": 1140
|
712 |
},
|
713 |
{
|
714 |
"epoch": 2.83,
|
715 |
"learning_rate": 3.102189781021898e-06,
|
716 |
+
"loss": 1.7934,
|
717 |
"step": 1150
|
718 |
},
|
719 |
{
|
720 |
"epoch": 2.85,
|
721 |
"learning_rate": 2.6459854014598542e-06,
|
722 |
+
"loss": 1.7715,
|
723 |
"step": 1160
|
724 |
},
|
725 |
{
|
726 |
"epoch": 2.88,
|
727 |
"learning_rate": 2.1897810218978103e-06,
|
728 |
+
"loss": 1.7242,
|
729 |
"step": 1170
|
730 |
},
|
731 |
{
|
732 |
"epoch": 2.9,
|
733 |
"learning_rate": 1.7335766423357664e-06,
|
734 |
+
"loss": 1.8898,
|
735 |
"step": 1180
|
736 |
},
|
737 |
{
|
738 |
"epoch": 2.93,
|
739 |
"learning_rate": 1.2773722627737227e-06,
|
740 |
+
"loss": 1.6993,
|
741 |
"step": 1190
|
742 |
},
|
743 |
{
|
744 |
"epoch": 2.95,
|
745 |
"learning_rate": 8.211678832116789e-07,
|
746 |
+
"loss": 1.8267,
|
747 |
"step": 1200
|
748 |
},
|
749 |
{
|
750 |
"epoch": 2.97,
|
751 |
"learning_rate": 3.6496350364963505e-07,
|
752 |
+
"loss": 1.8206,
|
753 |
"step": 1210
|
754 |
},
|
755 |
{
|
756 |
"epoch": 2.99,
|
757 |
+
"eval_accuracy": 0.7496974935177182,
|
758 |
+
"eval_loss": 1.28416907787323,
|
759 |
+
"eval_runtime": 27.5596,
|
760 |
+
"eval_samples_per_second": 209.909,
|
761 |
+
"eval_steps_per_second": 6.568,
|
762 |
"step": 1218
|
763 |
},
|
764 |
{
|
765 |
"epoch": 2.99,
|
766 |
"step": 1218,
|
767 |
+
"total_flos": 3.942238865741697e+18,
|
768 |
+
"train_loss": 2.564211890028028,
|
769 |
+
"train_runtime": 1311.9838,
|
770 |
+
"train_samples_per_second": 119.043,
|
771 |
+
"train_steps_per_second": 0.928
|
772 |
}
|
773 |
],
|
774 |
"logging_steps": 10,
|
|
|
776 |
"num_input_tokens_seen": 0,
|
777 |
"num_train_epochs": 3,
|
778 |
"save_steps": 500,
|
779 |
+
"total_flos": 3.942238865741697e+18,
|
780 |
"train_batch_size": 32,
|
781 |
"trial_name": null,
|
782 |
"trial_params": null
|
training_args.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 4792
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:921ef97d4a35a1853a68ba7b3ab1cd7c82448b978014cfdaaa14d2ee5888e1bd
|
3 |
size 4792
|