End of training
Browse files- all_results.json +10 -10
- eval_results.json +6 -6
- pytorch_model.bin +1 -1
- train_results.json +5 -5
- trainer_state.json +188 -422
all_results.json
CHANGED
@@ -1,14 +1,14 @@
|
|
1 |
{
|
2 |
-
"epoch":
|
3 |
-
"eval_loss": 0.
|
4 |
-
"eval_runtime":
|
5 |
"eval_samples": 2095,
|
6 |
-
"eval_samples_per_second":
|
7 |
-
"eval_steps_per_second":
|
8 |
-
"eval_wer": 1.
|
9 |
-
"train_loss": 0.
|
10 |
-
"train_runtime":
|
11 |
"train_samples": 4711,
|
12 |
-
"train_samples_per_second":
|
13 |
-
"train_steps_per_second": 0.
|
14 |
}
|
|
|
1 |
{
|
2 |
+
"epoch": 27.12,
|
3 |
+
"eval_loss": 0.7346429228782654,
|
4 |
+
"eval_runtime": 80.4764,
|
5 |
"eval_samples": 2095,
|
6 |
+
"eval_samples_per_second": 26.032,
|
7 |
+
"eval_steps_per_second": 3.256,
|
8 |
+
"eval_wer": 1.0478761110202301,
|
9 |
+
"train_loss": 0.46973063707351687,
|
10 |
+
"train_runtime": 10600.4383,
|
11 |
"train_samples": 4711,
|
12 |
+
"train_samples_per_second": 12.075,
|
13 |
+
"train_steps_per_second": 0.755
|
14 |
}
|
eval_results.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
-
"epoch":
|
3 |
-
"eval_loss": 0.
|
4 |
-
"eval_runtime":
|
5 |
"eval_samples": 2095,
|
6 |
-
"eval_samples_per_second":
|
7 |
-
"eval_steps_per_second":
|
8 |
-
"eval_wer": 1.
|
9 |
}
|
|
|
1 |
{
|
2 |
+
"epoch": 27.12,
|
3 |
+
"eval_loss": 0.7346429228782654,
|
4 |
+
"eval_runtime": 80.4764,
|
5 |
"eval_samples": 2095,
|
6 |
+
"eval_samples_per_second": 26.032,
|
7 |
+
"eval_steps_per_second": 3.256,
|
8 |
+
"eval_wer": 1.0478761110202301
|
9 |
}
|
pytorch_model.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 1262321393
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7586f941c9408ff704686fb109ad61ca786c2510d0f24b51f08ecab6a57debc5
|
3 |
size 1262321393
|
train_results.json
CHANGED
@@ -1,8 +1,8 @@
|
|
1 |
{
|
2 |
-
"epoch":
|
3 |
-
"train_loss": 0.
|
4 |
-
"train_runtime":
|
5 |
"train_samples": 4711,
|
6 |
-
"train_samples_per_second":
|
7 |
-
"train_steps_per_second": 0.
|
8 |
}
|
|
|
1 |
{
|
2 |
+
"epoch": 27.12,
|
3 |
+
"train_loss": 0.46973063707351687,
|
4 |
+
"train_runtime": 10600.4383,
|
5 |
"train_samples": 4711,
|
6 |
+
"train_samples_per_second": 12.075,
|
7 |
+
"train_steps_per_second": 0.755
|
8 |
}
|
trainer_state.json
CHANGED
@@ -1,535 +1,301 @@
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
-
"epoch":
|
5 |
-
"global_step":
|
6 |
"is_hyper_param_search": false,
|
7 |
"is_local_process_zero": true,
|
8 |
"is_world_process_zero": true,
|
9 |
"log_history": [
|
10 |
-
{
|
11 |
-
"epoch": 0.34,
|
12 |
-
"eval_loss": 4.256950378417969,
|
13 |
-
"eval_runtime": 93.7021,
|
14 |
-
"eval_samples_per_second": 22.358,
|
15 |
-
"eval_steps_per_second": 2.796,
|
16 |
-
"eval_wer": 0.9999454713997492,
|
17 |
-
"step": 100
|
18 |
-
},
|
19 |
-
{
|
20 |
-
"epoch": 0.68,
|
21 |
-
"eval_loss": 3.3221993446350098,
|
22 |
-
"eval_runtime": 95.7275,
|
23 |
-
"eval_samples_per_second": 21.885,
|
24 |
-
"eval_steps_per_second": 2.737,
|
25 |
-
"eval_wer": 1.0005452860025084,
|
26 |
-
"step": 200
|
27 |
-
},
|
28 |
-
{
|
29 |
-
"epoch": 1.02,
|
30 |
-
"eval_loss": 2.375530242919922,
|
31 |
-
"eval_runtime": 93.1661,
|
32 |
-
"eval_samples_per_second": 22.487,
|
33 |
-
"eval_steps_per_second": 2.812,
|
34 |
-
"eval_wer": 1.0,
|
35 |
-
"step": 300
|
36 |
-
},
|
37 |
{
|
38 |
"epoch": 1.36,
|
39 |
-
"eval_loss": 1.
|
40 |
-
"eval_runtime":
|
41 |
-
"eval_samples_per_second":
|
42 |
-
"eval_steps_per_second":
|
43 |
-
"eval_wer": 1.
|
44 |
"step": 400
|
45 |
},
|
46 |
{
|
47 |
"epoch": 1.69,
|
48 |
-
"learning_rate": 0.
|
49 |
-
"loss":
|
50 |
"step": 500
|
51 |
},
|
52 |
-
{
|
53 |
-
"epoch": 1.69,
|
54 |
-
"eval_loss": 1.2223190069198608,
|
55 |
-
"eval_runtime": 92.5816,
|
56 |
-
"eval_samples_per_second": 22.629,
|
57 |
-
"eval_steps_per_second": 2.83,
|
58 |
-
"eval_wer": 1.009106276241889,
|
59 |
-
"step": 500
|
60 |
-
},
|
61 |
-
{
|
62 |
-
"epoch": 2.03,
|
63 |
-
"eval_loss": 1.0884724855422974,
|
64 |
-
"eval_runtime": 94.6436,
|
65 |
-
"eval_samples_per_second": 22.136,
|
66 |
-
"eval_steps_per_second": 2.768,
|
67 |
-
"eval_wer": 1.0272643001254158,
|
68 |
-
"step": 600
|
69 |
-
},
|
70 |
-
{
|
71 |
-
"epoch": 2.37,
|
72 |
-
"eval_loss": 0.9818190336227417,
|
73 |
-
"eval_runtime": 92.6738,
|
74 |
-
"eval_samples_per_second": 22.606,
|
75 |
-
"eval_steps_per_second": 2.827,
|
76 |
-
"eval_wer": 1.009760619444899,
|
77 |
-
"step": 700
|
78 |
-
},
|
79 |
{
|
80 |
"epoch": 2.71,
|
81 |
-
"eval_loss": 0.
|
82 |
-
"eval_runtime":
|
83 |
-
"eval_samples_per_second":
|
84 |
-
"eval_steps_per_second":
|
85 |
-
"eval_wer": 1.
|
86 |
"step": 800
|
87 |
},
|
88 |
-
{
|
89 |
-
"epoch": 3.05,
|
90 |
-
"eval_loss": 0.8868876695632935,
|
91 |
-
"eval_runtime": 94.7308,
|
92 |
-
"eval_samples_per_second": 22.115,
|
93 |
-
"eval_steps_per_second": 2.766,
|
94 |
-
"eval_wer": 1.0153770652707346,
|
95 |
-
"step": 900
|
96 |
-
},
|
97 |
-
{
|
98 |
-
"epoch": 3.39,
|
99 |
-
"learning_rate": 0.0002501875,
|
100 |
-
"loss": 0.5888,
|
101 |
-
"step": 1000
|
102 |
-
},
|
103 |
{
|
104 |
"epoch": 3.39,
|
105 |
-
"
|
106 |
-
"
|
107 |
-
"eval_samples_per_second": 22.602,
|
108 |
-
"eval_steps_per_second": 2.827,
|
109 |
-
"eval_wer": 1.0148863078684771,
|
110 |
"step": 1000
|
111 |
},
|
112 |
-
{
|
113 |
-
"epoch": 3.73,
|
114 |
-
"eval_loss": 0.8322562575340271,
|
115 |
-
"eval_runtime": 91.8444,
|
116 |
-
"eval_samples_per_second": 22.81,
|
117 |
-
"eval_steps_per_second": 2.853,
|
118 |
-
"eval_wer": 1.0124325208571896,
|
119 |
-
"step": 1100
|
120 |
-
},
|
121 |
{
|
122 |
"epoch": 4.07,
|
123 |
-
"eval_loss": 0.
|
124 |
-
"eval_runtime":
|
125 |
-
"eval_samples_per_second":
|
126 |
-
"eval_steps_per_second":
|
127 |
-
"eval_wer": 1.
|
128 |
"step": 1200
|
129 |
},
|
130 |
-
{
|
131 |
-
"epoch": 4.41,
|
132 |
-
"eval_loss": 0.8659328818321228,
|
133 |
-
"eval_runtime": 91.959,
|
134 |
-
"eval_samples_per_second": 22.782,
|
135 |
-
"eval_steps_per_second": 2.849,
|
136 |
-
"eval_wer": 1.0069251322318555,
|
137 |
-
"step": 1300
|
138 |
-
},
|
139 |
-
{
|
140 |
-
"epoch": 4.75,
|
141 |
-
"eval_loss": 0.8611888289451599,
|
142 |
-
"eval_runtime": 92.7084,
|
143 |
-
"eval_samples_per_second": 22.598,
|
144 |
-
"eval_steps_per_second": 2.826,
|
145 |
-
"eval_wer": 1.0042532308195649,
|
146 |
-
"step": 1400
|
147 |
-
},
|
148 |
{
|
149 |
"epoch": 5.08,
|
150 |
-
"learning_rate": 0.
|
151 |
-
"loss": 0.
|
152 |
-
"step": 1500
|
153 |
-
},
|
154 |
-
{
|
155 |
-
"epoch": 5.08,
|
156 |
-
"eval_loss": 0.8300378918647766,
|
157 |
-
"eval_runtime": 91.9965,
|
158 |
-
"eval_samples_per_second": 22.773,
|
159 |
-
"eval_steps_per_second": 2.848,
|
160 |
-
"eval_wer": 1.0182670810840286,
|
161 |
"step": 1500
|
162 |
},
|
163 |
{
|
164 |
"epoch": 5.42,
|
165 |
-
"eval_loss": 0.
|
166 |
-
"eval_runtime":
|
167 |
-
"eval_samples_per_second":
|
168 |
-
"eval_steps_per_second":
|
169 |
-
"eval_wer": 1.
|
170 |
"step": 1600
|
171 |
},
|
172 |
-
{
|
173 |
-
"epoch": 5.76,
|
174 |
-
"eval_loss": 0.8261289000511169,
|
175 |
-
"eval_runtime": 91.9157,
|
176 |
-
"eval_samples_per_second": 22.793,
|
177 |
-
"eval_steps_per_second": 2.85,
|
178 |
-
"eval_wer": 1.008724576040133,
|
179 |
-
"step": 1700
|
180 |
-
},
|
181 |
-
{
|
182 |
-
"epoch": 6.1,
|
183 |
-
"eval_loss": 0.854759931564331,
|
184 |
-
"eval_runtime": 92.0581,
|
185 |
-
"eval_samples_per_second": 22.757,
|
186 |
-
"eval_steps_per_second": 2.846,
|
187 |
-
"eval_wer": 1.006761546431103,
|
188 |
-
"step": 1800
|
189 |
-
},
|
190 |
-
{
|
191 |
-
"epoch": 6.44,
|
192 |
-
"eval_loss": 0.7983754873275757,
|
193 |
-
"eval_runtime": 92.3841,
|
194 |
-
"eval_samples_per_second": 22.677,
|
195 |
-
"eval_steps_per_second": 2.836,
|
196 |
-
"eval_wer": 1.011014777250668,
|
197 |
-
"step": 1900
|
198 |
-
},
|
199 |
{
|
200 |
"epoch": 6.78,
|
201 |
-
"learning_rate": 0.
|
202 |
-
"loss": 0.
|
203 |
"step": 2000
|
204 |
},
|
205 |
{
|
206 |
"epoch": 6.78,
|
207 |
-
"eval_loss": 0.
|
208 |
-
"eval_runtime":
|
209 |
-
"eval_samples_per_second":
|
210 |
-
"eval_steps_per_second":
|
211 |
-
"eval_wer": 1.
|
212 |
"step": 2000
|
213 |
},
|
214 |
-
{
|
215 |
-
"epoch": 7.12,
|
216 |
-
"eval_loss": 0.8498700857162476,
|
217 |
-
"eval_runtime": 91.7074,
|
218 |
-
"eval_samples_per_second": 22.844,
|
219 |
-
"eval_steps_per_second": 2.857,
|
220 |
-
"eval_wer": 1.0072523038333605,
|
221 |
-
"step": 2100
|
222 |
-
},
|
223 |
-
{
|
224 |
-
"epoch": 7.46,
|
225 |
-
"eval_loss": 0.8480040431022644,
|
226 |
-
"eval_runtime": 92.1173,
|
227 |
-
"eval_samples_per_second": 22.743,
|
228 |
-
"eval_steps_per_second": 2.844,
|
229 |
-
"eval_wer": 1.0112328916516713,
|
230 |
-
"step": 2200
|
231 |
-
},
|
232 |
-
{
|
233 |
-
"epoch": 7.8,
|
234 |
-
"eval_loss": 0.7929303646087646,
|
235 |
-
"eval_runtime": 92.2635,
|
236 |
-
"eval_samples_per_second": 22.707,
|
237 |
-
"eval_steps_per_second": 2.84,
|
238 |
-
"eval_wer": 1.0098696766454005,
|
239 |
-
"step": 2300
|
240 |
-
},
|
241 |
{
|
242 |
"epoch": 8.14,
|
243 |
-
"eval_loss": 0.
|
244 |
-
"eval_runtime":
|
245 |
-
"eval_samples_per_second":
|
246 |
-
"eval_steps_per_second":
|
247 |
-
"eval_wer": 1.
|
248 |
"step": 2400
|
249 |
},
|
250 |
{
|
251 |
"epoch": 8.47,
|
252 |
-
"learning_rate": 0.
|
253 |
-
"loss": 0.
|
254 |
-
"step": 2500
|
255 |
-
},
|
256 |
-
{
|
257 |
-
"epoch": 8.47,
|
258 |
-
"eval_loss": 0.8583062291145325,
|
259 |
-
"eval_runtime": 96.9455,
|
260 |
-
"eval_samples_per_second": 21.61,
|
261 |
-
"eval_steps_per_second": 2.703,
|
262 |
-
"eval_wer": 1.0062707890288456,
|
263 |
"step": 2500
|
264 |
},
|
265 |
-
{
|
266 |
-
"epoch": 8.81,
|
267 |
-
"eval_loss": 0.8326291441917419,
|
268 |
-
"eval_runtime": 93.0622,
|
269 |
-
"eval_samples_per_second": 22.512,
|
270 |
-
"eval_steps_per_second": 2.815,
|
271 |
-
"eval_wer": 1.0109602486504172,
|
272 |
-
"step": 2600
|
273 |
-
},
|
274 |
-
{
|
275 |
-
"epoch": 9.15,
|
276 |
-
"eval_loss": 0.8759247064590454,
|
277 |
-
"eval_runtime": 92.8699,
|
278 |
-
"eval_samples_per_second": 22.558,
|
279 |
-
"eval_steps_per_second": 2.821,
|
280 |
-
"eval_wer": 1.0036534162168058,
|
281 |
-
"step": 2700
|
282 |
-
},
|
283 |
{
|
284 |
"epoch": 9.49,
|
285 |
-
"eval_loss": 0.
|
286 |
-
"eval_runtime":
|
287 |
-
"eval_samples_per_second":
|
288 |
-
"eval_steps_per_second":
|
289 |
-
"eval_wer": 1.
|
290 |
"step": 2800
|
291 |
},
|
292 |
-
{
|
293 |
-
"epoch": 9.83,
|
294 |
-
"eval_loss": 0.8777465224266052,
|
295 |
-
"eval_runtime": 92.008,
|
296 |
-
"eval_samples_per_second": 22.77,
|
297 |
-
"eval_steps_per_second": 2.848,
|
298 |
-
"eval_wer": 1.0224112547030917,
|
299 |
-
"step": 2900
|
300 |
-
},
|
301 |
-
{
|
302 |
-
"epoch": 10.17,
|
303 |
-
"learning_rate": 0.0001251875,
|
304 |
-
"loss": 0.1682,
|
305 |
-
"step": 3000
|
306 |
-
},
|
307 |
{
|
308 |
"epoch": 10.17,
|
309 |
-
"
|
310 |
-
"
|
311 |
-
"eval_samples_per_second": 22.713,
|
312 |
-
"eval_steps_per_second": 2.84,
|
313 |
-
"eval_wer": 1.0280277005289273,
|
314 |
"step": 3000
|
315 |
},
|
316 |
-
{
|
317 |
-
"epoch": 10.51,
|
318 |
-
"eval_loss": 0.9213446378707886,
|
319 |
-
"eval_runtime": 93.1739,
|
320 |
-
"eval_samples_per_second": 22.485,
|
321 |
-
"eval_steps_per_second": 2.812,
|
322 |
-
"eval_wer": 1.006761546431103,
|
323 |
-
"step": 3100
|
324 |
-
},
|
325 |
{
|
326 |
"epoch": 10.85,
|
327 |
-
"eval_loss": 0.
|
328 |
-
"eval_runtime":
|
329 |
-
"eval_samples_per_second":
|
330 |
-
"eval_steps_per_second":
|
331 |
-
"eval_wer": 1.
|
332 |
"step": 3200
|
333 |
},
|
334 |
{
|
335 |
-
"epoch": 11.
|
336 |
-
"
|
337 |
-
"
|
338 |
-
"
|
339 |
-
"eval_steps_per_second": 2.794,
|
340 |
-
"eval_wer": 1.010033262446153,
|
341 |
-
"step": 3300
|
342 |
},
|
343 |
{
|
344 |
-
"epoch":
|
345 |
-
"eval_loss": 0.
|
346 |
-
"eval_runtime":
|
347 |
-
"eval_samples_per_second":
|
348 |
-
"eval_steps_per_second":
|
349 |
-
"eval_wer": 1.
|
350 |
-
"step":
|
351 |
},
|
352 |
{
|
353 |
-
"epoch":
|
354 |
-
"learning_rate":
|
355 |
-
"loss": 0.
|
356 |
-
"step":
|
357 |
},
|
358 |
{
|
359 |
-
"epoch":
|
360 |
-
"eval_loss": 0.
|
361 |
-
"eval_runtime":
|
362 |
-
"eval_samples_per_second":
|
363 |
-
"eval_steps_per_second":
|
364 |
-
"eval_wer": 1.
|
365 |
-
"step":
|
366 |
},
|
367 |
{
|
368 |
-
"epoch":
|
369 |
-
"eval_loss": 0.
|
370 |
-
"eval_runtime":
|
371 |
-
"eval_samples_per_second":
|
372 |
-
"eval_steps_per_second":
|
373 |
-
"eval_wer": 1.
|
374 |
-
"step":
|
375 |
},
|
376 |
{
|
377 |
-
"epoch":
|
378 |
-
"
|
379 |
-
"
|
380 |
-
"
|
381 |
-
"eval_steps_per_second": 2.809,
|
382 |
-
"eval_wer": 1.0083428758383772,
|
383 |
-
"step": 3700
|
384 |
},
|
385 |
{
|
386 |
-
"epoch":
|
387 |
-
"eval_loss": 0.
|
388 |
-
"eval_runtime":
|
389 |
-
"eval_samples_per_second":
|
390 |
-
"eval_steps_per_second":
|
391 |
-
"eval_wer": 1.
|
392 |
-
"step":
|
393 |
},
|
394 |
{
|
395 |
-
"epoch":
|
396 |
-
"
|
397 |
-
"
|
398 |
-
"
|
399 |
-
"eval_steps_per_second": 2.854,
|
400 |
-
"eval_wer": 1.010687605649163,
|
401 |
-
"step": 3900
|
402 |
},
|
403 |
{
|
404 |
-
"epoch":
|
405 |
-
"
|
406 |
-
"
|
407 |
-
"
|
|
|
|
|
|
|
408 |
},
|
409 |
{
|
410 |
-
"epoch":
|
411 |
-
"
|
412 |
-
"
|
413 |
-
"
|
414 |
-
"eval_steps_per_second": 2.825,
|
415 |
-
"eval_wer": 1.0098696766454005,
|
416 |
-
"step": 4000
|
417 |
},
|
418 |
{
|
419 |
-
"epoch":
|
420 |
-
"eval_loss": 0.
|
421 |
-
"eval_runtime":
|
422 |
-
"eval_samples_per_second":
|
423 |
-
"eval_steps_per_second":
|
424 |
-
"eval_wer": 1.
|
425 |
-
"step":
|
426 |
},
|
427 |
{
|
428 |
-
"epoch":
|
429 |
-
"
|
430 |
-
"
|
431 |
-
"
|
432 |
-
"eval_steps_per_second": 2.83,
|
433 |
-
"eval_wer": 1.0081247614373738,
|
434 |
-
"step": 4200
|
435 |
},
|
436 |
{
|
437 |
-
"epoch":
|
438 |
-
"eval_loss": 0.
|
439 |
-
"eval_runtime":
|
440 |
-
"eval_samples_per_second":
|
441 |
-
"eval_steps_per_second":
|
442 |
-
"eval_wer": 1.
|
443 |
-
"step":
|
444 |
},
|
445 |
{
|
446 |
-
"epoch":
|
447 |
-
"eval_loss": 0.
|
448 |
-
"eval_runtime":
|
449 |
-
"eval_samples_per_second":
|
450 |
-
"eval_steps_per_second":
|
451 |
-
"eval_wer": 1.
|
452 |
-
"step":
|
453 |
},
|
454 |
{
|
455 |
-
"epoch":
|
456 |
-
"learning_rate":
|
457 |
-
"loss": 0.
|
458 |
-
"step":
|
459 |
},
|
460 |
{
|
461 |
-
"epoch":
|
462 |
-
"eval_loss": 0.
|
463 |
-
"eval_runtime":
|
464 |
-
"eval_samples_per_second":
|
465 |
-
"eval_steps_per_second":
|
466 |
-
"eval_wer": 1.
|
467 |
-
"step":
|
468 |
},
|
469 |
{
|
470 |
-
"epoch":
|
471 |
-
"
|
472 |
-
"
|
473 |
-
"
|
474 |
-
"eval_steps_per_second": 2.85,
|
475 |
-
"eval_wer": 1.009051747641638,
|
476 |
-
"step": 4600
|
477 |
},
|
478 |
{
|
479 |
-
"epoch":
|
480 |
-
"eval_loss": 0.
|
481 |
-
"eval_runtime":
|
482 |
-
"eval_samples_per_second":
|
483 |
-
"eval_steps_per_second":
|
484 |
-
"eval_wer": 1.
|
485 |
-
"step":
|
486 |
},
|
487 |
{
|
488 |
-
"epoch":
|
489 |
-
"
|
490 |
-
"
|
491 |
-
"
|
492 |
-
"eval_steps_per_second": 2.841,
|
493 |
-
"eval_wer": 1.0091608048421397,
|
494 |
-
"step": 4800
|
495 |
},
|
496 |
{
|
497 |
-
"epoch":
|
498 |
-
"eval_loss": 0.
|
499 |
-
"eval_runtime":
|
500 |
-
"eval_samples_per_second":
|
501 |
-
"eval_steps_per_second":
|
502 |
-
"eval_wer": 1.
|
503 |
-
"step":
|
504 |
},
|
505 |
{
|
506 |
-
"epoch":
|
507 |
-
"learning_rate": 1.
|
508 |
-
"loss": 0.
|
509 |
-
"step":
|
510 |
},
|
511 |
{
|
512 |
-
"epoch":
|
513 |
-
"eval_loss": 0.
|
514 |
-
"eval_runtime":
|
515 |
-
"eval_samples_per_second":
|
516 |
-
"eval_steps_per_second":
|
517 |
-
"eval_wer": 1.
|
518 |
-
"step":
|
519 |
},
|
520 |
{
|
521 |
-
"epoch":
|
522 |
-
"step":
|
523 |
-
"total_flos": 1.
|
524 |
-
"train_loss": 0.
|
525 |
-
"train_runtime":
|
526 |
-
"train_samples_per_second":
|
527 |
-
"train_steps_per_second": 0.
|
528 |
}
|
529 |
],
|
530 |
-
"max_steps":
|
531 |
-
"num_train_epochs":
|
532 |
-
"total_flos": 1.
|
533 |
"trial_name": null,
|
534 |
"trial_params": null
|
535 |
}
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 27.11864406779661,
|
5 |
+
"global_step": 8000,
|
6 |
"is_hyper_param_search": false,
|
7 |
"is_local_process_zero": true,
|
8 |
"is_world_process_zero": true,
|
9 |
"log_history": [
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
10 |
{
|
11 |
"epoch": 1.36,
|
12 |
+
"eval_loss": 1.4595407247543335,
|
13 |
+
"eval_runtime": 81.6295,
|
14 |
+
"eval_samples_per_second": 25.665,
|
15 |
+
"eval_steps_per_second": 3.21,
|
16 |
+
"eval_wer": 1.0039260592180599,
|
17 |
"step": 400
|
18 |
},
|
19 |
{
|
20 |
"epoch": 1.69,
|
21 |
+
"learning_rate": 0.0002982,
|
22 |
+
"loss": 4.7778,
|
23 |
"step": 500
|
24 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
25 |
{
|
26 |
"epoch": 2.71,
|
27 |
+
"eval_loss": 0.8082281947135925,
|
28 |
+
"eval_runtime": 80.8384,
|
29 |
+
"eval_samples_per_second": 25.916,
|
30 |
+
"eval_steps_per_second": 3.241,
|
31 |
+
"eval_wer": 1.0115055346529254,
|
32 |
"step": 800
|
33 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
34 |
{
|
35 |
"epoch": 3.39,
|
36 |
+
"learning_rate": 0.00028011999999999997,
|
37 |
+
"loss": 0.6408,
|
|
|
|
|
|
|
38 |
"step": 1000
|
39 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
40 |
{
|
41 |
"epoch": 4.07,
|
42 |
+
"eval_loss": 0.7031667828559875,
|
43 |
+
"eval_runtime": 84.117,
|
44 |
+
"eval_samples_per_second": 24.906,
|
45 |
+
"eval_steps_per_second": 3.115,
|
46 |
+
"eval_wer": 1.0078521184361198,
|
47 |
"step": 1200
|
48 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
49 |
{
|
50 |
"epoch": 5.08,
|
51 |
+
"learning_rate": 0.00026011999999999997,
|
52 |
+
"loss": 0.3937,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
53 |
"step": 1500
|
54 |
},
|
55 |
{
|
56 |
"epoch": 5.42,
|
57 |
+
"eval_loss": 0.6889204382896423,
|
58 |
+
"eval_runtime": 80.1903,
|
59 |
+
"eval_samples_per_second": 26.125,
|
60 |
+
"eval_steps_per_second": 3.267,
|
61 |
+
"eval_wer": 1.0432957085991603,
|
62 |
"step": 1600
|
63 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
64 |
{
|
65 |
"epoch": 6.78,
|
66 |
+
"learning_rate": 0.00024011999999999997,
|
67 |
+
"loss": 0.3,
|
68 |
"step": 2000
|
69 |
},
|
70 |
{
|
71 |
"epoch": 6.78,
|
72 |
+
"eval_loss": 0.6820310354232788,
|
73 |
+
"eval_runtime": 80.2232,
|
74 |
+
"eval_samples_per_second": 26.115,
|
75 |
+
"eval_steps_per_second": 3.266,
|
76 |
+
"eval_wer": 1.0068706036316049,
|
77 |
"step": 2000
|
78 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
79 |
{
|
80 |
"epoch": 8.14,
|
81 |
+
"eval_loss": 0.6669920086860657,
|
82 |
+
"eval_runtime": 79.89,
|
83 |
+
"eval_samples_per_second": 26.224,
|
84 |
+
"eval_steps_per_second": 3.28,
|
85 |
+
"eval_wer": 1.0196302960902994,
|
86 |
"step": 2400
|
87 |
},
|
88 |
{
|
89 |
"epoch": 8.47,
|
90 |
+
"learning_rate": 0.00022011999999999997,
|
91 |
+
"loss": 0.226,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
92 |
"step": 2500
|
93 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
94 |
{
|
95 |
"epoch": 9.49,
|
96 |
+
"eval_loss": 0.7215595841407776,
|
97 |
+
"eval_runtime": 80.0002,
|
98 |
+
"eval_samples_per_second": 26.187,
|
99 |
+
"eval_steps_per_second": 3.275,
|
100 |
+
"eval_wer": 1.0422051365941436,
|
101 |
"step": 2800
|
102 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
103 |
{
|
104 |
"epoch": 10.17,
|
105 |
+
"learning_rate": 0.00020012,
|
106 |
+
"loss": 0.197,
|
|
|
|
|
|
|
107 |
"step": 3000
|
108 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
109 |
{
|
110 |
"epoch": 10.85,
|
111 |
+
"eval_loss": 0.7669464945793152,
|
112 |
+
"eval_runtime": 80.1105,
|
113 |
+
"eval_samples_per_second": 26.151,
|
114 |
+
"eval_steps_per_second": 3.27,
|
115 |
+
"eval_wer": 1.053383499645564,
|
116 |
"step": 3200
|
117 |
},
|
118 |
{
|
119 |
+
"epoch": 11.86,
|
120 |
+
"learning_rate": 0.00018012,
|
121 |
+
"loss": 0.165,
|
122 |
+
"step": 3500
|
|
|
|
|
|
|
123 |
},
|
124 |
{
|
125 |
+
"epoch": 12.2,
|
126 |
+
"eval_loss": 0.7517344951629639,
|
127 |
+
"eval_runtime": 79.716,
|
128 |
+
"eval_samples_per_second": 26.281,
|
129 |
+
"eval_steps_per_second": 3.287,
|
130 |
+
"eval_wer": 1.0199574676918044,
|
131 |
+
"step": 3600
|
132 |
},
|
133 |
{
|
134 |
+
"epoch": 13.56,
|
135 |
+
"learning_rate": 0.00016011999999999998,
|
136 |
+
"loss": 0.1486,
|
137 |
+
"step": 4000
|
138 |
},
|
139 |
{
|
140 |
+
"epoch": 13.56,
|
141 |
+
"eval_loss": 0.7124771475791931,
|
142 |
+
"eval_runtime": 79.8981,
|
143 |
+
"eval_samples_per_second": 26.221,
|
144 |
+
"eval_steps_per_second": 3.279,
|
145 |
+
"eval_wer": 1.0357162331642946,
|
146 |
+
"step": 4000
|
147 |
},
|
148 |
{
|
149 |
+
"epoch": 14.92,
|
150 |
+
"eval_loss": 0.7447456121444702,
|
151 |
+
"eval_runtime": 82.2103,
|
152 |
+
"eval_samples_per_second": 25.483,
|
153 |
+
"eval_steps_per_second": 3.187,
|
154 |
+
"eval_wer": 1.0347347183597797,
|
155 |
+
"step": 4400
|
156 |
},
|
157 |
{
|
158 |
+
"epoch": 15.25,
|
159 |
+
"learning_rate": 0.00014012,
|
160 |
+
"loss": 0.122,
|
161 |
+
"step": 4500
|
|
|
|
|
|
|
162 |
},
|
163 |
{
|
164 |
+
"epoch": 16.27,
|
165 |
+
"eval_loss": 0.6899322271347046,
|
166 |
+
"eval_runtime": 79.6624,
|
167 |
+
"eval_samples_per_second": 26.298,
|
168 |
+
"eval_steps_per_second": 3.289,
|
169 |
+
"eval_wer": 1.0440045804024212,
|
170 |
+
"step": 4800
|
171 |
},
|
172 |
{
|
173 |
+
"epoch": 16.95,
|
174 |
+
"learning_rate": 0.00012011999999999998,
|
175 |
+
"loss": 0.1069,
|
176 |
+
"step": 5000
|
|
|
|
|
|
|
177 |
},
|
178 |
{
|
179 |
+
"epoch": 17.63,
|
180 |
+
"eval_loss": 0.7212241291999817,
|
181 |
+
"eval_runtime": 80.3141,
|
182 |
+
"eval_samples_per_second": 26.085,
|
183 |
+
"eval_steps_per_second": 3.262,
|
184 |
+
"eval_wer": 1.0350073613610338,
|
185 |
+
"step": 5200
|
186 |
},
|
187 |
{
|
188 |
+
"epoch": 18.64,
|
189 |
+
"learning_rate": 0.00010011999999999998,
|
190 |
+
"loss": 0.0961,
|
191 |
+
"step": 5500
|
|
|
|
|
|
|
192 |
},
|
193 |
{
|
194 |
+
"epoch": 18.98,
|
195 |
+
"eval_loss": 0.7417359352111816,
|
196 |
+
"eval_runtime": 80.0211,
|
197 |
+
"eval_samples_per_second": 26.181,
|
198 |
+
"eval_steps_per_second": 3.274,
|
199 |
+
"eval_wer": 1.0408419215878728,
|
200 |
+
"step": 5600
|
201 |
},
|
202 |
{
|
203 |
+
"epoch": 20.34,
|
204 |
+
"learning_rate": 8.012e-05,
|
205 |
+
"loss": 0.086,
|
206 |
+
"step": 6000
|
|
|
|
|
|
|
207 |
},
|
208 |
{
|
209 |
+
"epoch": 20.34,
|
210 |
+
"eval_loss": 0.7402010560035706,
|
211 |
+
"eval_runtime": 80.0522,
|
212 |
+
"eval_samples_per_second": 26.17,
|
213 |
+
"eval_steps_per_second": 3.273,
|
214 |
+
"eval_wer": 1.0355526473635421,
|
215 |
+
"step": 6000
|
216 |
},
|
217 |
{
|
218 |
+
"epoch": 21.69,
|
219 |
+
"eval_loss": 0.7760854959487915,
|
220 |
+
"eval_runtime": 80.138,
|
221 |
+
"eval_samples_per_second": 26.142,
|
222 |
+
"eval_steps_per_second": 3.269,
|
223 |
+
"eval_wer": 1.0419870221931402,
|
224 |
+
"step": 6400
|
225 |
},
|
226 |
{
|
227 |
+
"epoch": 22.03,
|
228 |
+
"learning_rate": 6.0119999999999994e-05,
|
229 |
+
"loss": 0.0756,
|
230 |
+
"step": 6500
|
231 |
},
|
232 |
{
|
233 |
+
"epoch": 23.05,
|
234 |
+
"eval_loss": 0.7345677614212036,
|
235 |
+
"eval_runtime": 80.4841,
|
236 |
+
"eval_samples_per_second": 26.03,
|
237 |
+
"eval_steps_per_second": 3.255,
|
238 |
+
"eval_wer": 1.036915862369813,
|
239 |
+
"step": 6800
|
240 |
},
|
241 |
{
|
242 |
+
"epoch": 23.73,
|
243 |
+
"learning_rate": 4.012e-05,
|
244 |
+
"loss": 0.0666,
|
245 |
+
"step": 7000
|
|
|
|
|
|
|
246 |
},
|
247 |
{
|
248 |
+
"epoch": 24.41,
|
249 |
+
"eval_loss": 0.7506045699119568,
|
250 |
+
"eval_runtime": 82.6434,
|
251 |
+
"eval_samples_per_second": 25.35,
|
252 |
+
"eval_steps_per_second": 3.17,
|
253 |
+
"eval_wer": 1.0449315666066852,
|
254 |
+
"step": 7200
|
255 |
},
|
256 |
{
|
257 |
+
"epoch": 25.42,
|
258 |
+
"learning_rate": 2.0119999999999997e-05,
|
259 |
+
"loss": 0.0595,
|
260 |
+
"step": 7500
|
|
|
|
|
|
|
261 |
},
|
262 |
{
|
263 |
+
"epoch": 25.76,
|
264 |
+
"eval_loss": 0.7319227457046509,
|
265 |
+
"eval_runtime": 79.8082,
|
266 |
+
"eval_samples_per_second": 26.25,
|
267 |
+
"eval_steps_per_second": 3.283,
|
268 |
+
"eval_wer": 1.0476034680189759,
|
269 |
+
"step": 7600
|
270 |
},
|
271 |
{
|
272 |
+
"epoch": 27.12,
|
273 |
+
"learning_rate": 1.6e-07,
|
274 |
+
"loss": 0.054,
|
275 |
+
"step": 8000
|
276 |
},
|
277 |
{
|
278 |
+
"epoch": 27.12,
|
279 |
+
"eval_loss": 0.7346429228782654,
|
280 |
+
"eval_runtime": 79.9162,
|
281 |
+
"eval_samples_per_second": 26.215,
|
282 |
+
"eval_steps_per_second": 3.278,
|
283 |
+
"eval_wer": 1.0478761110202301,
|
284 |
+
"step": 8000
|
285 |
},
|
286 |
{
|
287 |
+
"epoch": 27.12,
|
288 |
+
"step": 8000,
|
289 |
+
"total_flos": 1.592873144248711e+19,
|
290 |
+
"train_loss": 0.46973063707351687,
|
291 |
+
"train_runtime": 10600.4383,
|
292 |
+
"train_samples_per_second": 12.075,
|
293 |
+
"train_steps_per_second": 0.755
|
294 |
}
|
295 |
],
|
296 |
+
"max_steps": 8000,
|
297 |
+
"num_train_epochs": 28,
|
298 |
+
"total_flos": 1.592873144248711e+19,
|
299 |
"trial_name": null,
|
300 |
"trial_params": null
|
301 |
}
|