DewiBrynJones
commited on
Commit
•
7606543
1
Parent(s):
782bf7c
End of training
Browse files- README.md +15 -2
- all_results.json +9 -9
- eval_results.json +5 -5
- runs/Aug22_06-11-06_202a94c24ee9/events.out.tfevents.1724374762.202a94c24ee9.32.1 +3 -0
- train_results.json +4 -4
- trainer_state.json +472 -472
README.md
CHANGED
@@ -3,11 +3,24 @@ license: apache-2.0
|
|
3 |
base_model: openai/whisper-large-v3
|
4 |
tags:
|
5 |
- generated_from_trainer
|
|
|
|
|
6 |
metrics:
|
7 |
- wer
|
8 |
model-index:
|
9 |
- name: whisper-large-v3-ft-btb-cy
|
10 |
-
results:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
11 |
---
|
12 |
|
13 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
@@ -15,7 +28,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
15 |
|
16 |
# whisper-large-v3-ft-btb-cy
|
17 |
|
18 |
-
This model is a fine-tuned version of [openai/whisper-large-v3](https://huggingface.co/openai/whisper-large-v3) on
|
19 |
It achieves the following results on the evaluation set:
|
20 |
- Loss: 0.5776
|
21 |
- Wer: 0.3064
|
|
|
3 |
base_model: openai/whisper-large-v3
|
4 |
tags:
|
5 |
- generated_from_trainer
|
6 |
+
datasets:
|
7 |
+
- DewiBrynJones/banc-trawsgrifiadau-bangor-clean
|
8 |
metrics:
|
9 |
- wer
|
10 |
model-index:
|
11 |
- name: whisper-large-v3-ft-btb-cy
|
12 |
+
results:
|
13 |
+
- task:
|
14 |
+
name: Automatic Speech Recognition
|
15 |
+
type: automatic-speech-recognition
|
16 |
+
dataset:
|
17 |
+
name: DewiBrynJones/banc-trawsgrifiadau-bangor-clean default
|
18 |
+
type: DewiBrynJones/banc-trawsgrifiadau-bangor-clean
|
19 |
+
args: default
|
20 |
+
metrics:
|
21 |
+
- name: Wer
|
22 |
+
type: wer
|
23 |
+
value: 0.3064475414845421
|
24 |
---
|
25 |
|
26 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
|
|
28 |
|
29 |
# whisper-large-v3-ft-btb-cy
|
30 |
|
31 |
+
This model is a fine-tuned version of [openai/whisper-large-v3](https://huggingface.co/openai/whisper-large-v3) on the DewiBrynJones/banc-trawsgrifiadau-bangor-clean default dataset.
|
32 |
It achieves the following results on the evaluation set:
|
33 |
- Loss: 0.5776
|
34 |
- Wer: 0.3064
|
all_results.json
CHANGED
@@ -1,15 +1,15 @@
|
|
1 |
{
|
2 |
"epoch": 5.694760820045558,
|
3 |
-
"eval_loss": 0.
|
4 |
-
"eval_runtime":
|
5 |
"eval_samples": 7022,
|
6 |
-
"eval_samples_per_second": 2.
|
7 |
-
"eval_steps_per_second": 0.
|
8 |
-
"eval_wer": 0.
|
9 |
"total_flos": 5.43429854134272e+20,
|
10 |
-
"train_loss": 0.
|
11 |
-
"train_runtime":
|
12 |
"train_samples": 28086,
|
13 |
-
"train_samples_per_second": 2.
|
14 |
-
"train_steps_per_second": 0.
|
15 |
}
|
|
|
1 |
{
|
2 |
"epoch": 5.694760820045558,
|
3 |
+
"eval_loss": 0.5775743126869202,
|
4 |
+
"eval_runtime": 2885.0748,
|
5 |
"eval_samples": 7022,
|
6 |
+
"eval_samples_per_second": 2.434,
|
7 |
+
"eval_steps_per_second": 0.152,
|
8 |
+
"eval_wer": 0.3064475414845421,
|
9 |
"total_flos": 5.43429854134272e+20,
|
10 |
+
"train_loss": 0.25138797866106033,
|
11 |
+
"train_runtime": 67213.2375,
|
12 |
"train_samples": 28086,
|
13 |
+
"train_samples_per_second": 2.38,
|
14 |
+
"train_steps_per_second": 0.074
|
15 |
}
|
eval_results.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
"epoch": 5.694760820045558,
|
3 |
-
"eval_loss": 0.
|
4 |
-
"eval_runtime":
|
5 |
"eval_samples": 7022,
|
6 |
-
"eval_samples_per_second": 2.
|
7 |
-
"eval_steps_per_second": 0.
|
8 |
-
"eval_wer": 0.
|
9 |
}
|
|
|
1 |
{
|
2 |
"epoch": 5.694760820045558,
|
3 |
+
"eval_loss": 0.5775743126869202,
|
4 |
+
"eval_runtime": 2885.0748,
|
5 |
"eval_samples": 7022,
|
6 |
+
"eval_samples_per_second": 2.434,
|
7 |
+
"eval_steps_per_second": 0.152,
|
8 |
+
"eval_wer": 0.3064475414845421
|
9 |
}
|
runs/Aug22_06-11-06_202a94c24ee9/events.out.tfevents.1724374762.202a94c24ee9.32.1
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:543cedfd64104cd8eac1d9d6d69449fa278e7053f1e99196e7dd66c8be21d92d
|
3 |
+
size 406
|
train_results.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
"epoch": 5.694760820045558,
|
3 |
"total_flos": 5.43429854134272e+20,
|
4 |
-
"train_loss": 0.
|
5 |
-
"train_runtime":
|
6 |
"train_samples": 28086,
|
7 |
-
"train_samples_per_second": 2.
|
8 |
-
"train_steps_per_second": 0.
|
9 |
}
|
|
|
1 |
{
|
2 |
"epoch": 5.694760820045558,
|
3 |
"total_flos": 5.43429854134272e+20,
|
4 |
+
"train_loss": 0.25138797866106033,
|
5 |
+
"train_runtime": 67213.2375,
|
6 |
"train_samples": 28086,
|
7 |
+
"train_samples_per_second": 2.38,
|
8 |
+
"train_steps_per_second": 0.074
|
9 |
}
|
trainer_state.json
CHANGED
@@ -10,1457 +10,1457 @@
|
|
10 |
"log_history": [
|
11 |
{
|
12 |
"epoch": 0.02847380410022779,
|
13 |
-
"grad_norm": 8.
|
14 |
"learning_rate": 5.000000000000001e-07,
|
15 |
-
"loss": 1.
|
16 |
"step": 25
|
17 |
},
|
18 |
{
|
19 |
"epoch": 0.05694760820045558,
|
20 |
-
"grad_norm": 5.
|
21 |
"learning_rate": 1.0000000000000002e-06,
|
22 |
-
"loss": 1.
|
23 |
"step": 50
|
24 |
},
|
25 |
{
|
26 |
"epoch": 0.08542141230068337,
|
27 |
-
"grad_norm": 6.
|
28 |
"learning_rate": 1.5e-06,
|
29 |
-
"loss":
|
30 |
"step": 75
|
31 |
},
|
32 |
{
|
33 |
"epoch": 0.11389521640091116,
|
34 |
-
"grad_norm": 5.
|
35 |
"learning_rate": 2.0000000000000003e-06,
|
36 |
-
"loss": 0.
|
37 |
"step": 100
|
38 |
},
|
39 |
{
|
40 |
"epoch": 0.14236902050113895,
|
41 |
-
"grad_norm": 4.
|
42 |
"learning_rate": 2.5e-06,
|
43 |
-
"loss": 0.
|
44 |
"step": 125
|
45 |
},
|
46 |
{
|
47 |
"epoch": 0.17084282460136674,
|
48 |
-
"grad_norm": 5.
|
49 |
"learning_rate": 3e-06,
|
50 |
-
"loss": 0.
|
51 |
"step": 150
|
52 |
},
|
53 |
{
|
54 |
"epoch": 0.19931662870159453,
|
55 |
-
"grad_norm": 5.
|
56 |
"learning_rate": 3.5e-06,
|
57 |
-
"loss": 0.
|
58 |
"step": 175
|
59 |
},
|
60 |
{
|
61 |
"epoch": 0.22779043280182232,
|
62 |
-
"grad_norm": 4.
|
63 |
"learning_rate": 4.000000000000001e-06,
|
64 |
-
"loss": 0.
|
65 |
"step": 200
|
66 |
},
|
67 |
{
|
68 |
"epoch": 0.25626423690205014,
|
69 |
-
"grad_norm": 5.
|
70 |
"learning_rate": 4.5e-06,
|
71 |
-
"loss": 0.
|
72 |
"step": 225
|
73 |
},
|
74 |
{
|
75 |
"epoch": 0.2847380410022779,
|
76 |
-
"grad_norm": 4.
|
77 |
"learning_rate": 5e-06,
|
78 |
-
"loss": 0.
|
79 |
"step": 250
|
80 |
},
|
81 |
{
|
82 |
"epoch": 0.3132118451025057,
|
83 |
-
"grad_norm": 4.
|
84 |
"learning_rate": 5.500000000000001e-06,
|
85 |
-
"loss": 0.
|
86 |
"step": 275
|
87 |
},
|
88 |
{
|
89 |
"epoch": 0.3416856492027335,
|
90 |
-
"grad_norm": 4.
|
91 |
"learning_rate": 6e-06,
|
92 |
-
"loss": 0.
|
93 |
"step": 300
|
94 |
},
|
95 |
{
|
96 |
"epoch": 0.3701594533029613,
|
97 |
-
"grad_norm": 4.
|
98 |
"learning_rate": 6.5000000000000004e-06,
|
99 |
-
"loss": 0.
|
100 |
"step": 325
|
101 |
},
|
102 |
{
|
103 |
"epoch": 0.39863325740318906,
|
104 |
-
"grad_norm": 5.
|
105 |
"learning_rate": 7e-06,
|
106 |
-
"loss": 0.
|
107 |
"step": 350
|
108 |
},
|
109 |
{
|
110 |
"epoch": 0.4271070615034169,
|
111 |
-
"grad_norm": 4.
|
112 |
"learning_rate": 7.500000000000001e-06,
|
113 |
-
"loss": 0.
|
114 |
"step": 375
|
115 |
},
|
116 |
{
|
117 |
"epoch": 0.45558086560364464,
|
118 |
-
"grad_norm": 4.
|
119 |
"learning_rate": 8.000000000000001e-06,
|
120 |
-
"loss": 0.
|
121 |
"step": 400
|
122 |
},
|
123 |
{
|
124 |
"epoch": 0.48405466970387245,
|
125 |
-
"grad_norm": 4.
|
126 |
"learning_rate": 8.5e-06,
|
127 |
-
"loss": 0.
|
128 |
"step": 425
|
129 |
},
|
130 |
{
|
131 |
"epoch": 0.5125284738041003,
|
132 |
-
"grad_norm":
|
133 |
"learning_rate": 9e-06,
|
134 |
-
"loss": 0.
|
135 |
"step": 450
|
136 |
},
|
137 |
{
|
138 |
"epoch": 0.541002277904328,
|
139 |
-
"grad_norm": 4.
|
140 |
"learning_rate": 9.5e-06,
|
141 |
-
"loss": 0.
|
142 |
"step": 475
|
143 |
},
|
144 |
{
|
145 |
"epoch": 0.5694760820045558,
|
146 |
-
"grad_norm":
|
147 |
"learning_rate": 1e-05,
|
148 |
-
"loss": 0.
|
149 |
"step": 500
|
150 |
},
|
151 |
{
|
152 |
"epoch": 0.5979498861047836,
|
153 |
-
"grad_norm": 4.
|
154 |
"learning_rate": 9.944444444444445e-06,
|
155 |
-
"loss": 0.
|
156 |
"step": 525
|
157 |
},
|
158 |
{
|
159 |
"epoch": 0.6264236902050114,
|
160 |
-
"grad_norm": 4.
|
161 |
"learning_rate": 9.88888888888889e-06,
|
162 |
-
"loss": 0.
|
163 |
"step": 550
|
164 |
},
|
165 |
{
|
166 |
"epoch": 0.6548974943052391,
|
167 |
-
"grad_norm": 3.
|
168 |
"learning_rate": 9.833333333333333e-06,
|
169 |
-
"loss": 0.
|
170 |
"step": 575
|
171 |
},
|
172 |
{
|
173 |
"epoch": 0.683371298405467,
|
174 |
-
"grad_norm": 3.
|
175 |
"learning_rate": 9.777777777777779e-06,
|
176 |
-
"loss": 0.
|
177 |
"step": 600
|
178 |
},
|
179 |
{
|
180 |
"epoch": 0.7118451025056948,
|
181 |
-
"grad_norm": 3.
|
182 |
"learning_rate": 9.722222222222223e-06,
|
183 |
-
"loss": 0.
|
184 |
"step": 625
|
185 |
},
|
186 |
{
|
187 |
"epoch": 0.7403189066059226,
|
188 |
-
"grad_norm": 3.
|
189 |
"learning_rate": 9.666666666666667e-06,
|
190 |
-
"loss": 0.
|
191 |
"step": 650
|
192 |
},
|
193 |
{
|
194 |
"epoch": 0.7687927107061503,
|
195 |
-
"grad_norm":
|
196 |
"learning_rate": 9.611111111111112e-06,
|
197 |
-
"loss": 0.
|
198 |
"step": 675
|
199 |
},
|
200 |
{
|
201 |
"epoch": 0.7972665148063781,
|
202 |
-
"grad_norm":
|
203 |
"learning_rate": 9.555555555555556e-06,
|
204 |
-
"loss": 0.
|
205 |
"step": 700
|
206 |
},
|
207 |
{
|
208 |
"epoch": 0.8257403189066059,
|
209 |
-
"grad_norm": 3.
|
210 |
"learning_rate": 9.5e-06,
|
211 |
-
"loss": 0.
|
212 |
"step": 725
|
213 |
},
|
214 |
{
|
215 |
"epoch": 0.8542141230068337,
|
216 |
-
"grad_norm":
|
217 |
"learning_rate": 9.444444444444445e-06,
|
218 |
-
"loss": 0.
|
219 |
"step": 750
|
220 |
},
|
221 |
{
|
222 |
"epoch": 0.8826879271070615,
|
223 |
-
"grad_norm": 3.
|
224 |
"learning_rate": 9.38888888888889e-06,
|
225 |
-
"loss": 0.
|
226 |
"step": 775
|
227 |
},
|
228 |
{
|
229 |
"epoch": 0.9111617312072893,
|
230 |
-
"grad_norm": 3.
|
231 |
"learning_rate": 9.333333333333334e-06,
|
232 |
-
"loss": 0.
|
233 |
"step": 800
|
234 |
},
|
235 |
{
|
236 |
"epoch": 0.9396355353075171,
|
237 |
-
"grad_norm":
|
238 |
"learning_rate": 9.277777777777778e-06,
|
239 |
-
"loss": 0.
|
240 |
"step": 825
|
241 |
},
|
242 |
{
|
243 |
"epoch": 0.9681093394077449,
|
244 |
-
"grad_norm": 3.
|
245 |
"learning_rate": 9.222222222222224e-06,
|
246 |
-
"loss": 0.
|
247 |
"step": 850
|
248 |
},
|
249 |
{
|
250 |
"epoch": 0.9965831435079726,
|
251 |
-
"grad_norm": 3.
|
252 |
"learning_rate": 9.166666666666666e-06,
|
253 |
-
"loss": 0.
|
254 |
"step": 875
|
255 |
},
|
256 |
{
|
257 |
"epoch": 1.0250569476082005,
|
258 |
-
"grad_norm": 2.
|
259 |
"learning_rate": 9.111111111111112e-06,
|
260 |
-
"loss": 0.
|
261 |
"step": 900
|
262 |
},
|
263 |
{
|
264 |
"epoch": 1.0535307517084282,
|
265 |
-
"grad_norm": 3.
|
266 |
"learning_rate": 9.055555555555556e-06,
|
267 |
-
"loss": 0.
|
268 |
"step": 925
|
269 |
},
|
270 |
{
|
271 |
"epoch": 1.082004555808656,
|
272 |
-
"grad_norm":
|
273 |
"learning_rate": 9e-06,
|
274 |
-
"loss": 0.
|
275 |
"step": 950
|
276 |
},
|
277 |
{
|
278 |
"epoch": 1.1104783599088839,
|
279 |
-
"grad_norm": 3.
|
280 |
"learning_rate": 8.944444444444446e-06,
|
281 |
-
"loss": 0.
|
282 |
"step": 975
|
283 |
},
|
284 |
{
|
285 |
"epoch": 1.1389521640091116,
|
286 |
-
"grad_norm":
|
287 |
"learning_rate": 8.888888888888888e-06,
|
288 |
-
"loss": 0.
|
289 |
"step": 1000
|
290 |
},
|
291 |
{
|
292 |
"epoch": 1.1389521640091116,
|
293 |
-
"eval_loss": 0.
|
294 |
-
"eval_runtime":
|
295 |
-
"eval_samples_per_second": 2.
|
296 |
"eval_steps_per_second": 0.147,
|
297 |
-
"eval_wer": 0.
|
298 |
"step": 1000
|
299 |
},
|
300 |
{
|
301 |
"epoch": 1.1674259681093395,
|
302 |
-
"grad_norm": 3.
|
303 |
"learning_rate": 8.833333333333334e-06,
|
304 |
-
"loss": 0.
|
305 |
"step": 1025
|
306 |
},
|
307 |
{
|
308 |
"epoch": 1.1958997722095672,
|
309 |
-
"grad_norm": 3.
|
310 |
"learning_rate": 8.777777777777778e-06,
|
311 |
-
"loss": 0.
|
312 |
"step": 1050
|
313 |
},
|
314 |
{
|
315 |
"epoch": 1.224373576309795,
|
316 |
-
"grad_norm": 3.
|
317 |
"learning_rate": 8.722222222222224e-06,
|
318 |
-
"loss": 0.
|
319 |
"step": 1075
|
320 |
},
|
321 |
{
|
322 |
"epoch": 1.2528473804100229,
|
323 |
-
"grad_norm": 3.
|
324 |
"learning_rate": 8.666666666666668e-06,
|
325 |
-
"loss": 0.
|
326 |
"step": 1100
|
327 |
},
|
328 |
{
|
329 |
"epoch": 1.2813211845102506,
|
330 |
-
"grad_norm": 2.
|
331 |
"learning_rate": 8.611111111111112e-06,
|
332 |
-
"loss": 0.
|
333 |
"step": 1125
|
334 |
},
|
335 |
{
|
336 |
"epoch": 1.3097949886104785,
|
337 |
-
"grad_norm": 2.
|
338 |
"learning_rate": 8.555555555555556e-06,
|
339 |
-
"loss": 0.
|
340 |
"step": 1150
|
341 |
},
|
342 |
{
|
343 |
"epoch": 1.3382687927107062,
|
344 |
-
"grad_norm": 3.
|
345 |
"learning_rate": 8.5e-06,
|
346 |
-
"loss": 0.
|
347 |
"step": 1175
|
348 |
},
|
349 |
{
|
350 |
"epoch": 1.366742596810934,
|
351 |
-
"grad_norm": 3.
|
352 |
"learning_rate": 8.444444444444446e-06,
|
353 |
-
"loss": 0.
|
354 |
"step": 1200
|
355 |
},
|
356 |
{
|
357 |
"epoch": 1.3952164009111616,
|
358 |
-
"grad_norm": 2.
|
359 |
"learning_rate": 8.38888888888889e-06,
|
360 |
-
"loss": 0.
|
361 |
"step": 1225
|
362 |
},
|
363 |
{
|
364 |
"epoch": 1.4236902050113895,
|
365 |
-
"grad_norm": 2.
|
366 |
"learning_rate": 8.333333333333334e-06,
|
367 |
-
"loss": 0.
|
368 |
"step": 1250
|
369 |
},
|
370 |
{
|
371 |
"epoch": 1.4521640091116172,
|
372 |
-
"grad_norm":
|
373 |
"learning_rate": 8.277777777777778e-06,
|
374 |
-
"loss": 0.
|
375 |
"step": 1275
|
376 |
},
|
377 |
{
|
378 |
"epoch": 1.4806378132118452,
|
379 |
-
"grad_norm": 3.
|
380 |
"learning_rate": 8.222222222222222e-06,
|
381 |
-
"loss": 0.
|
382 |
"step": 1300
|
383 |
},
|
384 |
{
|
385 |
"epoch": 1.5091116173120729,
|
386 |
-
"grad_norm":
|
387 |
"learning_rate": 8.166666666666668e-06,
|
388 |
-
"loss": 0.
|
389 |
"step": 1325
|
390 |
},
|
391 |
{
|
392 |
"epoch": 1.5375854214123006,
|
393 |
-
"grad_norm": 2.
|
394 |
"learning_rate": 8.111111111111112e-06,
|
395 |
-
"loss": 0.
|
396 |
"step": 1350
|
397 |
},
|
398 |
{
|
399 |
"epoch": 1.5660592255125285,
|
400 |
-
"grad_norm":
|
401 |
"learning_rate": 8.055555555555557e-06,
|
402 |
-
"loss": 0.
|
403 |
"step": 1375
|
404 |
},
|
405 |
{
|
406 |
"epoch": 1.5945330296127562,
|
407 |
-
"grad_norm": 3.
|
408 |
"learning_rate": 8.000000000000001e-06,
|
409 |
-
"loss": 0.
|
410 |
"step": 1400
|
411 |
},
|
412 |
{
|
413 |
"epoch": 1.6230068337129842,
|
414 |
-
"grad_norm":
|
415 |
"learning_rate": 7.944444444444445e-06,
|
416 |
-
"loss": 0.
|
417 |
"step": 1425
|
418 |
},
|
419 |
{
|
420 |
"epoch": 1.6514806378132119,
|
421 |
-
"grad_norm": 3.
|
422 |
"learning_rate": 7.88888888888889e-06,
|
423 |
-
"loss": 0.
|
424 |
"step": 1450
|
425 |
},
|
426 |
{
|
427 |
"epoch": 1.6799544419134396,
|
428 |
-
"grad_norm": 2.
|
429 |
"learning_rate": 7.833333333333333e-06,
|
430 |
-
"loss": 0.
|
431 |
"step": 1475
|
432 |
},
|
433 |
{
|
434 |
"epoch": 1.7084282460136673,
|
435 |
-
"grad_norm": 3.
|
436 |
"learning_rate": 7.77777777777778e-06,
|
437 |
-
"loss": 0.
|
438 |
"step": 1500
|
439 |
},
|
440 |
{
|
441 |
"epoch": 1.7369020501138952,
|
442 |
-
"grad_norm": 3.
|
443 |
"learning_rate": 7.722222222222223e-06,
|
444 |
-
"loss": 0.
|
445 |
"step": 1525
|
446 |
},
|
447 |
{
|
448 |
"epoch": 1.7653758542141231,
|
449 |
-
"grad_norm": 3.
|
450 |
"learning_rate": 7.666666666666667e-06,
|
451 |
-
"loss": 0.
|
452 |
"step": 1550
|
453 |
},
|
454 |
{
|
455 |
"epoch": 1.7938496583143508,
|
456 |
-
"grad_norm": 2.
|
457 |
"learning_rate": 7.611111111111111e-06,
|
458 |
-
"loss": 0.
|
459 |
"step": 1575
|
460 |
},
|
461 |
{
|
462 |
"epoch": 1.8223234624145785,
|
463 |
-
"grad_norm": 2.
|
464 |
"learning_rate": 7.555555555555556e-06,
|
465 |
-
"loss": 0.
|
466 |
"step": 1600
|
467 |
},
|
468 |
{
|
469 |
"epoch": 1.8507972665148062,
|
470 |
-
"grad_norm": 3.
|
471 |
"learning_rate": 7.500000000000001e-06,
|
472 |
-
"loss": 0.
|
473 |
"step": 1625
|
474 |
},
|
475 |
{
|
476 |
"epoch": 1.8792710706150342,
|
477 |
-
"grad_norm": 3.
|
478 |
"learning_rate": 7.444444444444445e-06,
|
479 |
-
"loss": 0.
|
480 |
"step": 1650
|
481 |
},
|
482 |
{
|
483 |
"epoch": 1.907744874715262,
|
484 |
-
"grad_norm": 2.
|
485 |
"learning_rate": 7.38888888888889e-06,
|
486 |
-
"loss": 0.
|
487 |
"step": 1675
|
488 |
},
|
489 |
{
|
490 |
"epoch": 1.9362186788154898,
|
491 |
-
"grad_norm": 3.
|
492 |
"learning_rate": 7.333333333333333e-06,
|
493 |
-
"loss": 0.
|
494 |
"step": 1700
|
495 |
},
|
496 |
{
|
497 |
"epoch": 1.9646924829157175,
|
498 |
-
"grad_norm": 3.
|
499 |
"learning_rate": 7.277777777777778e-06,
|
500 |
-
"loss": 0.
|
501 |
"step": 1725
|
502 |
},
|
503 |
{
|
504 |
"epoch": 1.9931662870159452,
|
505 |
-
"grad_norm": 2.
|
506 |
"learning_rate": 7.222222222222223e-06,
|
507 |
-
"loss": 0.
|
508 |
"step": 1750
|
509 |
},
|
510 |
{
|
511 |
"epoch": 2.021640091116173,
|
512 |
-
"grad_norm": 2.
|
513 |
"learning_rate": 7.166666666666667e-06,
|
514 |
-
"loss": 0.
|
515 |
"step": 1775
|
516 |
},
|
517 |
{
|
518 |
"epoch": 2.050113895216401,
|
519 |
-
"grad_norm": 2.
|
520 |
"learning_rate": 7.111111111111112e-06,
|
521 |
-
"loss": 0.
|
522 |
"step": 1800
|
523 |
},
|
524 |
{
|
525 |
"epoch": 2.078587699316629,
|
526 |
-
"grad_norm": 2.
|
527 |
"learning_rate": 7.055555555555557e-06,
|
528 |
-
"loss": 0.
|
529 |
"step": 1825
|
530 |
},
|
531 |
{
|
532 |
"epoch": 2.1070615034168565,
|
533 |
-
"grad_norm": 2.
|
534 |
"learning_rate": 7e-06,
|
535 |
-
"loss": 0.
|
536 |
"step": 1850
|
537 |
},
|
538 |
{
|
539 |
"epoch": 2.135535307517084,
|
540 |
-
"grad_norm": 2.
|
541 |
"learning_rate": 6.944444444444445e-06,
|
542 |
-
"loss": 0.
|
543 |
"step": 1875
|
544 |
},
|
545 |
{
|
546 |
"epoch": 2.164009111617312,
|
547 |
-
"grad_norm": 2.
|
548 |
"learning_rate": 6.88888888888889e-06,
|
549 |
-
"loss": 0.
|
550 |
"step": 1900
|
551 |
},
|
552 |
{
|
553 |
"epoch": 2.19248291571754,
|
554 |
-
"grad_norm": 2.
|
555 |
"learning_rate": 6.833333333333334e-06,
|
556 |
-
"loss": 0.
|
557 |
"step": 1925
|
558 |
},
|
559 |
{
|
560 |
"epoch": 2.2209567198177678,
|
561 |
-
"grad_norm":
|
562 |
"learning_rate": 6.777777777777779e-06,
|
563 |
-
"loss": 0.
|
564 |
"step": 1950
|
565 |
},
|
566 |
{
|
567 |
"epoch": 2.2494305239179955,
|
568 |
-
"grad_norm":
|
569 |
"learning_rate": 6.7222222222222235e-06,
|
570 |
-
"loss": 0.
|
571 |
"step": 1975
|
572 |
},
|
573 |
{
|
574 |
"epoch": 2.277904328018223,
|
575 |
-
"grad_norm": 2.
|
576 |
"learning_rate": 6.666666666666667e-06,
|
577 |
-
"loss": 0.
|
578 |
"step": 2000
|
579 |
},
|
580 |
{
|
581 |
"epoch": 2.277904328018223,
|
582 |
-
"eval_loss": 0.
|
583 |
-
"eval_runtime":
|
584 |
-
"eval_samples_per_second": 2.
|
585 |
-
"eval_steps_per_second": 0.
|
586 |
-
"eval_wer": 0.
|
587 |
"step": 2000
|
588 |
},
|
589 |
{
|
590 |
"epoch": 2.306378132118451,
|
591 |
-
"grad_norm":
|
592 |
"learning_rate": 6.6111111111111115e-06,
|
593 |
-
"loss": 0.
|
594 |
"step": 2025
|
595 |
},
|
596 |
{
|
597 |
"epoch": 2.334851936218679,
|
598 |
-
"grad_norm": 2.
|
599 |
"learning_rate": 6.555555555555556e-06,
|
600 |
-
"loss": 0.
|
601 |
"step": 2050
|
602 |
},
|
603 |
{
|
604 |
"epoch": 2.3633257403189067,
|
605 |
-
"grad_norm": 2.
|
606 |
"learning_rate": 6.5000000000000004e-06,
|
607 |
-
"loss": 0.
|
608 |
"step": 2075
|
609 |
},
|
610 |
{
|
611 |
"epoch": 2.3917995444191344,
|
612 |
-
"grad_norm": 2.
|
613 |
"learning_rate": 6.444444444444445e-06,
|
614 |
-
"loss": 0.
|
615 |
"step": 2100
|
616 |
},
|
617 |
{
|
618 |
"epoch": 2.420273348519362,
|
619 |
-
"grad_norm": 2.
|
620 |
"learning_rate": 6.3888888888888885e-06,
|
621 |
-
"loss": 0.
|
622 |
"step": 2125
|
623 |
},
|
624 |
{
|
625 |
"epoch": 2.44874715261959,
|
626 |
-
"grad_norm": 2.
|
627 |
"learning_rate": 6.333333333333333e-06,
|
628 |
-
"loss": 0.
|
629 |
"step": 2150
|
630 |
},
|
631 |
{
|
632 |
"epoch": 2.477220956719818,
|
633 |
-
"grad_norm": 2.
|
634 |
"learning_rate": 6.277777777777778e-06,
|
635 |
-
"loss": 0.
|
636 |
"step": 2175
|
637 |
},
|
638 |
{
|
639 |
"epoch": 2.5056947608200457,
|
640 |
-
"grad_norm": 2.
|
641 |
"learning_rate": 6.222222222222223e-06,
|
642 |
-
"loss": 0.
|
643 |
"step": 2200
|
644 |
},
|
645 |
{
|
646 |
"epoch": 2.5341685649202734,
|
647 |
-
"grad_norm":
|
648 |
"learning_rate": 6.166666666666667e-06,
|
649 |
-
"loss": 0.
|
650 |
"step": 2225
|
651 |
},
|
652 |
{
|
653 |
"epoch": 2.562642369020501,
|
654 |
-
"grad_norm": 2.
|
655 |
"learning_rate": 6.111111111111112e-06,
|
656 |
-
"loss": 0.
|
657 |
"step": 2250
|
658 |
},
|
659 |
{
|
660 |
"epoch": 2.591116173120729,
|
661 |
-
"grad_norm": 2.
|
662 |
"learning_rate": 6.055555555555555e-06,
|
663 |
-
"loss": 0.
|
664 |
"step": 2275
|
665 |
},
|
666 |
{
|
667 |
"epoch": 2.619589977220957,
|
668 |
-
"grad_norm": 3.
|
669 |
"learning_rate": 6e-06,
|
670 |
-
"loss": 0.
|
671 |
"step": 2300
|
672 |
},
|
673 |
{
|
674 |
"epoch": 2.6480637813211843,
|
675 |
-
"grad_norm": 2.
|
676 |
"learning_rate": 5.944444444444445e-06,
|
677 |
-
"loss": 0.
|
678 |
"step": 2325
|
679 |
},
|
680 |
{
|
681 |
"epoch": 2.6765375854214124,
|
682 |
-
"grad_norm": 2.
|
683 |
"learning_rate": 5.88888888888889e-06,
|
684 |
-
"loss": 0.
|
685 |
"step": 2350
|
686 |
},
|
687 |
{
|
688 |
"epoch": 2.70501138952164,
|
689 |
-
"grad_norm": 2.
|
690 |
"learning_rate": 5.833333333333334e-06,
|
691 |
-
"loss": 0.
|
692 |
"step": 2375
|
693 |
},
|
694 |
{
|
695 |
"epoch": 2.733485193621868,
|
696 |
-
"grad_norm": 2.
|
697 |
"learning_rate": 5.777777777777778e-06,
|
698 |
-
"loss": 0.
|
699 |
"step": 2400
|
700 |
},
|
701 |
{
|
702 |
"epoch": 2.7619589977220955,
|
703 |
-
"grad_norm":
|
704 |
"learning_rate": 5.722222222222222e-06,
|
705 |
-
"loss": 0.
|
706 |
"step": 2425
|
707 |
},
|
708 |
{
|
709 |
"epoch": 2.7904328018223232,
|
710 |
-
"grad_norm": 2.
|
711 |
"learning_rate": 5.666666666666667e-06,
|
712 |
-
"loss": 0.
|
713 |
"step": 2450
|
714 |
},
|
715 |
{
|
716 |
"epoch": 2.8189066059225514,
|
717 |
-
"grad_norm": 2.
|
718 |
"learning_rate": 5.611111111111112e-06,
|
719 |
-
"loss": 0.
|
720 |
"step": 2475
|
721 |
},
|
722 |
{
|
723 |
"epoch": 2.847380410022779,
|
724 |
-
"grad_norm": 2.
|
725 |
"learning_rate": 5.555555555555557e-06,
|
726 |
-
"loss": 0.
|
727 |
"step": 2500
|
728 |
},
|
729 |
{
|
730 |
"epoch": 2.875854214123007,
|
731 |
-
"grad_norm": 2.
|
732 |
"learning_rate": 5.500000000000001e-06,
|
733 |
-
"loss": 0.
|
734 |
"step": 2525
|
735 |
},
|
736 |
{
|
737 |
"epoch": 2.9043280182232345,
|
738 |
-
"grad_norm": 2.
|
739 |
"learning_rate": 5.444444444444445e-06,
|
740 |
-
"loss": 0.
|
741 |
"step": 2550
|
742 |
},
|
743 |
{
|
744 |
"epoch": 2.932801822323462,
|
745 |
-
"grad_norm": 2.
|
746 |
"learning_rate": 5.388888888888889e-06,
|
747 |
-
"loss": 0.
|
748 |
"step": 2575
|
749 |
},
|
750 |
{
|
751 |
"epoch": 2.9612756264236904,
|
752 |
-
"grad_norm":
|
753 |
"learning_rate": 5.333333333333334e-06,
|
754 |
-
"loss": 0.
|
755 |
"step": 2600
|
756 |
},
|
757 |
{
|
758 |
"epoch": 2.989749430523918,
|
759 |
-
"grad_norm": 2.
|
760 |
"learning_rate": 5.2777777777777785e-06,
|
761 |
-
"loss": 0.
|
762 |
"step": 2625
|
763 |
},
|
764 |
{
|
765 |
"epoch": 3.0182232346241458,
|
766 |
-
"grad_norm":
|
767 |
"learning_rate": 5.2222222222222226e-06,
|
768 |
-
"loss": 0.
|
769 |
"step": 2650
|
770 |
},
|
771 |
{
|
772 |
"epoch": 3.0466970387243735,
|
773 |
-
"grad_norm": 2.
|
774 |
"learning_rate": 5.1666666666666675e-06,
|
775 |
-
"loss": 0.
|
776 |
"step": 2675
|
777 |
},
|
778 |
{
|
779 |
"epoch": 3.075170842824601,
|
780 |
-
"grad_norm": 2.
|
781 |
"learning_rate": 5.1111111111111115e-06,
|
782 |
-
"loss": 0.
|
783 |
"step": 2700
|
784 |
},
|
785 |
{
|
786 |
"epoch": 3.1036446469248293,
|
787 |
-
"grad_norm":
|
788 |
"learning_rate": 5.0555555555555555e-06,
|
789 |
-
"loss": 0.
|
790 |
"step": 2725
|
791 |
},
|
792 |
{
|
793 |
"epoch": 3.132118451025057,
|
794 |
-
"grad_norm": 2.
|
795 |
"learning_rate": 5e-06,
|
796 |
-
"loss": 0.
|
797 |
"step": 2750
|
798 |
},
|
799 |
{
|
800 |
"epoch": 3.1605922551252847,
|
801 |
-
"grad_norm": 1.
|
802 |
"learning_rate": 4.944444444444445e-06,
|
803 |
-
"loss": 0.
|
804 |
"step": 2775
|
805 |
},
|
806 |
{
|
807 |
"epoch": 3.1890660592255125,
|
808 |
-
"grad_norm": 2.
|
809 |
"learning_rate": 4.888888888888889e-06,
|
810 |
-
"loss": 0.
|
811 |
"step": 2800
|
812 |
},
|
813 |
{
|
814 |
"epoch": 3.21753986332574,
|
815 |
-
"grad_norm": 2.
|
816 |
"learning_rate": 4.833333333333333e-06,
|
817 |
-
"loss": 0.
|
818 |
"step": 2825
|
819 |
},
|
820 |
{
|
821 |
"epoch": 3.2460136674259683,
|
822 |
-
"grad_norm": 2.
|
823 |
"learning_rate": 4.777777777777778e-06,
|
824 |
-
"loss": 0.
|
825 |
"step": 2850
|
826 |
},
|
827 |
{
|
828 |
"epoch": 3.274487471526196,
|
829 |
-
"grad_norm":
|
830 |
"learning_rate": 4.722222222222222e-06,
|
831 |
-
"loss": 0.
|
832 |
"step": 2875
|
833 |
},
|
834 |
{
|
835 |
"epoch": 3.3029612756264237,
|
836 |
-
"grad_norm": 2.
|
837 |
"learning_rate": 4.666666666666667e-06,
|
838 |
-
"loss": 0.
|
839 |
"step": 2900
|
840 |
},
|
841 |
{
|
842 |
"epoch": 3.3314350797266514,
|
843 |
-
"grad_norm": 2.
|
844 |
"learning_rate": 4.611111111111112e-06,
|
845 |
-
"loss": 0.
|
846 |
"step": 2925
|
847 |
},
|
848 |
{
|
849 |
"epoch": 3.359908883826879,
|
850 |
-
"grad_norm": 2.
|
851 |
"learning_rate": 4.555555555555556e-06,
|
852 |
-
"loss": 0.
|
853 |
"step": 2950
|
854 |
},
|
855 |
{
|
856 |
"epoch": 3.3883826879271073,
|
857 |
-
"grad_norm": 2.
|
858 |
"learning_rate": 4.5e-06,
|
859 |
-
"loss": 0.
|
860 |
"step": 2975
|
861 |
},
|
862 |
{
|
863 |
"epoch": 3.416856492027335,
|
864 |
-
"grad_norm": 2.
|
865 |
"learning_rate": 4.444444444444444e-06,
|
866 |
-
"loss": 0.
|
867 |
"step": 3000
|
868 |
},
|
869 |
{
|
870 |
"epoch": 3.416856492027335,
|
871 |
-
"eval_loss": 0.
|
872 |
-
"eval_runtime":
|
873 |
-
"eval_samples_per_second": 2.
|
874 |
-
"eval_steps_per_second": 0.
|
875 |
-
"eval_wer": 0.
|
876 |
"step": 3000
|
877 |
},
|
878 |
{
|
879 |
"epoch": 3.4453302961275627,
|
880 |
-
"grad_norm": 2.
|
881 |
"learning_rate": 4.388888888888889e-06,
|
882 |
-
"loss": 0.
|
883 |
"step": 3025
|
884 |
},
|
885 |
{
|
886 |
"epoch": 3.4738041002277904,
|
887 |
-
"grad_norm": 2.
|
888 |
"learning_rate": 4.333333333333334e-06,
|
889 |
-
"loss": 0.
|
890 |
"step": 3050
|
891 |
},
|
892 |
{
|
893 |
"epoch": 3.502277904328018,
|
894 |
-
"grad_norm": 2.
|
895 |
"learning_rate": 4.277777777777778e-06,
|
896 |
-
"loss": 0.
|
897 |
"step": 3075
|
898 |
},
|
899 |
{
|
900 |
"epoch": 3.5307517084282463,
|
901 |
-
"grad_norm": 2.
|
902 |
"learning_rate": 4.222222222222223e-06,
|
903 |
-
"loss": 0.
|
904 |
"step": 3100
|
905 |
},
|
906 |
{
|
907 |
"epoch": 3.559225512528474,
|
908 |
-
"grad_norm": 2.
|
909 |
"learning_rate": 4.166666666666667e-06,
|
910 |
-
"loss": 0.
|
911 |
"step": 3125
|
912 |
},
|
913 |
{
|
914 |
"epoch": 3.5876993166287017,
|
915 |
-
"grad_norm": 2.
|
916 |
"learning_rate": 4.111111111111111e-06,
|
917 |
-
"loss": 0.
|
918 |
"step": 3150
|
919 |
},
|
920 |
{
|
921 |
"epoch": 3.6161731207289294,
|
922 |
-
"grad_norm": 1.
|
923 |
"learning_rate": 4.055555555555556e-06,
|
924 |
-
"loss": 0.
|
925 |
"step": 3175
|
926 |
},
|
927 |
{
|
928 |
"epoch": 3.644646924829157,
|
929 |
-
"grad_norm": 2.
|
930 |
"learning_rate": 4.000000000000001e-06,
|
931 |
-
"loss": 0.
|
932 |
"step": 3200
|
933 |
},
|
934 |
{
|
935 |
"epoch": 3.6731207289293852,
|
936 |
-
"grad_norm":
|
937 |
"learning_rate": 3.944444444444445e-06,
|
938 |
-
"loss": 0.
|
939 |
"step": 3225
|
940 |
},
|
941 |
{
|
942 |
"epoch": 3.7015945330296125,
|
943 |
-
"grad_norm":
|
944 |
"learning_rate": 3.88888888888889e-06,
|
945 |
-
"loss": 0.
|
946 |
"step": 3250
|
947 |
},
|
948 |
{
|
949 |
"epoch": 3.7300683371298406,
|
950 |
-
"grad_norm": 1.
|
951 |
"learning_rate": 3.833333333333334e-06,
|
952 |
-
"loss": 0.
|
953 |
"step": 3275
|
954 |
},
|
955 |
{
|
956 |
"epoch": 3.7585421412300684,
|
957 |
-
"grad_norm": 2.
|
958 |
"learning_rate": 3.777777777777778e-06,
|
959 |
-
"loss": 0.
|
960 |
"step": 3300
|
961 |
},
|
962 |
{
|
963 |
"epoch": 3.787015945330296,
|
964 |
-
"grad_norm": 2.
|
965 |
"learning_rate": 3.7222222222222225e-06,
|
966 |
-
"loss": 0.
|
967 |
"step": 3325
|
968 |
},
|
969 |
{
|
970 |
"epoch": 3.8154897494305238,
|
971 |
-
"grad_norm":
|
972 |
"learning_rate": 3.6666666666666666e-06,
|
973 |
-
"loss": 0.
|
974 |
"step": 3350
|
975 |
},
|
976 |
{
|
977 |
"epoch": 3.8439635535307515,
|
978 |
-
"grad_norm": 2.
|
979 |
"learning_rate": 3.6111111111111115e-06,
|
980 |
-
"loss": 0.
|
981 |
"step": 3375
|
982 |
},
|
983 |
{
|
984 |
"epoch": 3.8724373576309796,
|
985 |
-
"grad_norm": 2.
|
986 |
"learning_rate": 3.555555555555556e-06,
|
987 |
-
"loss": 0.
|
988 |
"step": 3400
|
989 |
},
|
990 |
{
|
991 |
"epoch": 3.9009111617312073,
|
992 |
-
"grad_norm": 2.
|
993 |
"learning_rate": 3.5e-06,
|
994 |
-
"loss": 0.
|
995 |
"step": 3425
|
996 |
},
|
997 |
{
|
998 |
"epoch": 3.929384965831435,
|
999 |
-
"grad_norm": 2.
|
1000 |
"learning_rate": 3.444444444444445e-06,
|
1001 |
-
"loss": 0.
|
1002 |
"step": 3450
|
1003 |
},
|
1004 |
{
|
1005 |
"epoch": 3.9578587699316627,
|
1006 |
-
"grad_norm": 2.
|
1007 |
"learning_rate": 3.3888888888888893e-06,
|
1008 |
-
"loss": 0.
|
1009 |
"step": 3475
|
1010 |
},
|
1011 |
{
|
1012 |
"epoch": 3.9863325740318905,
|
1013 |
-
"grad_norm": 2.
|
1014 |
"learning_rate": 3.3333333333333333e-06,
|
1015 |
-
"loss": 0.
|
1016 |
"step": 3500
|
1017 |
},
|
1018 |
{
|
1019 |
"epoch": 4.014806378132119,
|
1020 |
-
"grad_norm": 1.
|
1021 |
"learning_rate": 3.277777777777778e-06,
|
1022 |
-
"loss": 0.
|
1023 |
"step": 3525
|
1024 |
},
|
1025 |
{
|
1026 |
"epoch": 4.043280182232346,
|
1027 |
-
"grad_norm": 1.
|
1028 |
"learning_rate": 3.2222222222222227e-06,
|
1029 |
-
"loss": 0.
|
1030 |
"step": 3550
|
1031 |
},
|
1032 |
{
|
1033 |
"epoch": 4.071753986332574,
|
1034 |
-
"grad_norm": 1.
|
1035 |
"learning_rate": 3.1666666666666667e-06,
|
1036 |
-
"loss": 0.
|
1037 |
"step": 3575
|
1038 |
},
|
1039 |
{
|
1040 |
"epoch": 4.100227790432802,
|
1041 |
-
"grad_norm": 1.
|
1042 |
"learning_rate": 3.1111111111111116e-06,
|
1043 |
-
"loss": 0.
|
1044 |
"step": 3600
|
1045 |
},
|
1046 |
{
|
1047 |
"epoch": 4.128701594533029,
|
1048 |
-
"grad_norm": 1.
|
1049 |
"learning_rate": 3.055555555555556e-06,
|
1050 |
-
"loss": 0.
|
1051 |
"step": 3625
|
1052 |
},
|
1053 |
{
|
1054 |
"epoch": 4.157175398633258,
|
1055 |
-
"grad_norm":
|
1056 |
"learning_rate": 3e-06,
|
1057 |
-
"loss": 0.
|
1058 |
"step": 3650
|
1059 |
},
|
1060 |
{
|
1061 |
"epoch": 4.185649202733485,
|
1062 |
-
"grad_norm": 1.
|
1063 |
"learning_rate": 2.944444444444445e-06,
|
1064 |
-
"loss": 0.
|
1065 |
"step": 3675
|
1066 |
},
|
1067 |
{
|
1068 |
"epoch": 4.214123006833713,
|
1069 |
-
"grad_norm": 2.
|
1070 |
"learning_rate": 2.888888888888889e-06,
|
1071 |
-
"loss": 0.
|
1072 |
"step": 3700
|
1073 |
},
|
1074 |
{
|
1075 |
"epoch": 4.242596810933941,
|
1076 |
-
"grad_norm": 1.
|
1077 |
"learning_rate": 2.8333333333333335e-06,
|
1078 |
-
"loss": 0.
|
1079 |
"step": 3725
|
1080 |
},
|
1081 |
{
|
1082 |
"epoch": 4.271070615034168,
|
1083 |
-
"grad_norm": 2.
|
1084 |
"learning_rate": 2.7777777777777783e-06,
|
1085 |
-
"loss": 0.
|
1086 |
"step": 3750
|
1087 |
},
|
1088 |
{
|
1089 |
"epoch": 4.2995444191343966,
|
1090 |
-
"grad_norm": 1.
|
1091 |
"learning_rate": 2.7222222222222224e-06,
|
1092 |
-
"loss": 0.
|
1093 |
"step": 3775
|
1094 |
},
|
1095 |
{
|
1096 |
"epoch": 4.328018223234624,
|
1097 |
-
"grad_norm": 1.
|
1098 |
"learning_rate": 2.666666666666667e-06,
|
1099 |
-
"loss": 0.
|
1100 |
"step": 3800
|
1101 |
},
|
1102 |
{
|
1103 |
"epoch": 4.356492027334852,
|
1104 |
-
"grad_norm":
|
1105 |
"learning_rate": 2.6111111111111113e-06,
|
1106 |
-
"loss": 0.
|
1107 |
"step": 3825
|
1108 |
},
|
1109 |
{
|
1110 |
"epoch": 4.38496583143508,
|
1111 |
-
"grad_norm":
|
1112 |
"learning_rate": 2.5555555555555557e-06,
|
1113 |
-
"loss": 0.
|
1114 |
"step": 3850
|
1115 |
},
|
1116 |
{
|
1117 |
"epoch": 4.413439635535307,
|
1118 |
-
"grad_norm":
|
1119 |
"learning_rate": 2.5e-06,
|
1120 |
-
"loss": 0.
|
1121 |
"step": 3875
|
1122 |
},
|
1123 |
{
|
1124 |
"epoch": 4.4419134396355355,
|
1125 |
-
"grad_norm": 1.
|
1126 |
-
"learning_rate": 2.
|
1127 |
-
"loss": 0.
|
1128 |
"step": 3900
|
1129 |
},
|
1130 |
{
|
1131 |
"epoch": 4.470387243735763,
|
1132 |
-
"grad_norm": 1.
|
1133 |
-
"learning_rate": 2.
|
1134 |
-
"loss": 0.
|
1135 |
"step": 3925
|
1136 |
},
|
1137 |
{
|
1138 |
"epoch": 4.498861047835991,
|
1139 |
-
"grad_norm":
|
1140 |
-
"learning_rate": 2.
|
1141 |
-
"loss": 0.
|
1142 |
"step": 3950
|
1143 |
},
|
1144 |
{
|
1145 |
"epoch": 4.527334851936219,
|
1146 |
-
"grad_norm": 1.
|
1147 |
-
"learning_rate": 2.
|
1148 |
-
"loss": 0.
|
1149 |
"step": 3975
|
1150 |
},
|
1151 |
{
|
1152 |
"epoch": 4.555808656036446,
|
1153 |
-
"grad_norm":
|
1154 |
-
"learning_rate": 2.
|
1155 |
-
"loss": 0.
|
1156 |
"step": 4000
|
1157 |
},
|
1158 |
{
|
1159 |
"epoch": 4.555808656036446,
|
1160 |
-
"eval_loss": 0.
|
1161 |
-
"eval_runtime":
|
1162 |
-
"eval_samples_per_second": 2.
|
1163 |
"eval_steps_per_second": 0.148,
|
1164 |
-
"eval_wer": 0.
|
1165 |
"step": 4000
|
1166 |
},
|
1167 |
{
|
1168 |
"epoch": 4.5842824601366745,
|
1169 |
-
"grad_norm": 2.
|
1170 |
-
"learning_rate": 2.
|
1171 |
-
"loss": 0.
|
1172 |
"step": 4025
|
1173 |
},
|
1174 |
{
|
1175 |
"epoch": 4.612756264236902,
|
1176 |
-
"grad_norm": 1.
|
1177 |
-
"learning_rate": 2.
|
1178 |
-
"loss": 0.
|
1179 |
"step": 4050
|
1180 |
},
|
1181 |
{
|
1182 |
"epoch": 4.64123006833713,
|
1183 |
-
"grad_norm": 2.
|
1184 |
-
"learning_rate": 2.
|
1185 |
-
"loss": 0.
|
1186 |
"step": 4075
|
1187 |
},
|
1188 |
{
|
1189 |
"epoch": 4.669703872437358,
|
1190 |
-
"grad_norm":
|
1191 |
-
"learning_rate": 2.
|
1192 |
-
"loss": 0.
|
1193 |
"step": 4100
|
1194 |
},
|
1195 |
{
|
1196 |
"epoch": 4.698177676537585,
|
1197 |
-
"grad_norm": 1.
|
1198 |
-
"learning_rate": 1.
|
1199 |
-
"loss": 0.
|
1200 |
"step": 4125
|
1201 |
},
|
1202 |
{
|
1203 |
"epoch": 4.7266514806378135,
|
1204 |
-
"grad_norm": 1.
|
1205 |
-
"learning_rate": 1.
|
1206 |
-
"loss": 0.
|
1207 |
"step": 4150
|
1208 |
},
|
1209 |
{
|
1210 |
"epoch": 4.755125284738041,
|
1211 |
-
"grad_norm": 1.
|
1212 |
-
"learning_rate": 1.
|
1213 |
-
"loss": 0.
|
1214 |
"step": 4175
|
1215 |
},
|
1216 |
{
|
1217 |
"epoch": 4.783599088838269,
|
1218 |
-
"grad_norm": 1.
|
1219 |
-
"learning_rate": 1.
|
1220 |
-
"loss": 0.
|
1221 |
"step": 4200
|
1222 |
},
|
1223 |
{
|
1224 |
"epoch": 4.812072892938497,
|
1225 |
-
"grad_norm": 2.
|
1226 |
-
"learning_rate": 1.
|
1227 |
-
"loss": 0.
|
1228 |
"step": 4225
|
1229 |
},
|
1230 |
{
|
1231 |
"epoch": 4.840546697038724,
|
1232 |
-
"grad_norm": 2.
|
1233 |
-
"learning_rate": 1.
|
1234 |
-
"loss": 0.
|
1235 |
"step": 4250
|
1236 |
},
|
1237 |
{
|
1238 |
"epoch": 4.8690205011389525,
|
1239 |
-
"grad_norm":
|
1240 |
-
"learning_rate": 1.
|
1241 |
-
"loss": 0.
|
1242 |
"step": 4275
|
1243 |
},
|
1244 |
{
|
1245 |
"epoch": 4.89749430523918,
|
1246 |
-
"grad_norm":
|
1247 |
-
"learning_rate": 1.
|
1248 |
-
"loss": 0.
|
1249 |
"step": 4300
|
1250 |
},
|
1251 |
{
|
1252 |
"epoch": 4.925968109339408,
|
1253 |
-
"grad_norm": 1.
|
1254 |
-
"learning_rate": 1.
|
1255 |
-
"loss": 0.
|
1256 |
"step": 4325
|
1257 |
},
|
1258 |
{
|
1259 |
"epoch": 4.954441913439636,
|
1260 |
-
"grad_norm":
|
1261 |
-
"learning_rate": 1.
|
1262 |
-
"loss": 0.
|
1263 |
"step": 4350
|
1264 |
},
|
1265 |
{
|
1266 |
"epoch": 4.982915717539863,
|
1267 |
-
"grad_norm": 1.
|
1268 |
-
"learning_rate": 1.
|
1269 |
-
"loss": 0.
|
1270 |
"step": 4375
|
1271 |
},
|
1272 |
{
|
1273 |
"epoch": 5.011389521640091,
|
1274 |
-
"grad_norm": 1.
|
1275 |
-
"learning_rate": 1.
|
1276 |
-
"loss": 0.
|
1277 |
"step": 4400
|
1278 |
},
|
1279 |
{
|
1280 |
"epoch": 5.039863325740319,
|
1281 |
-
"grad_norm": 0.
|
1282 |
-
"learning_rate": 1.
|
1283 |
-
"loss": 0.
|
1284 |
"step": 4425
|
1285 |
},
|
1286 |
{
|
1287 |
"epoch": 5.068337129840547,
|
1288 |
-
"grad_norm":
|
1289 |
-
"learning_rate": 1.
|
1290 |
-
"loss": 0.
|
1291 |
"step": 4450
|
1292 |
},
|
1293 |
{
|
1294 |
"epoch": 5.096810933940774,
|
1295 |
-
"grad_norm": 1.
|
1296 |
-
"learning_rate": 1.
|
1297 |
-
"loss": 0.
|
1298 |
"step": 4475
|
1299 |
},
|
1300 |
{
|
1301 |
"epoch": 5.125284738041002,
|
1302 |
-
"grad_norm":
|
1303 |
-
"learning_rate": 1.
|
1304 |
-
"loss": 0.
|
1305 |
"step": 4500
|
1306 |
},
|
1307 |
{
|
1308 |
"epoch": 5.15375854214123,
|
1309 |
-
"grad_norm":
|
1310 |
-
"learning_rate": 1.
|
1311 |
-
"loss": 0.
|
1312 |
"step": 4525
|
1313 |
},
|
1314 |
{
|
1315 |
"epoch": 5.182232346241458,
|
1316 |
-
"grad_norm": 1.
|
1317 |
-
"learning_rate": 1.
|
1318 |
-
"loss": 0.
|
1319 |
"step": 4550
|
1320 |
},
|
1321 |
{
|
1322 |
"epoch": 5.210706150341686,
|
1323 |
-
"grad_norm": 1.
|
1324 |
-
"learning_rate": 9.
|
1325 |
-
"loss": 0.
|
1326 |
"step": 4575
|
1327 |
},
|
1328 |
{
|
1329 |
"epoch": 5.239179954441913,
|
1330 |
-
"grad_norm": 1.
|
1331 |
-
"learning_rate": 8.
|
1332 |
-
"loss": 0.
|
1333 |
"step": 4600
|
1334 |
},
|
1335 |
{
|
1336 |
"epoch": 5.267653758542141,
|
1337 |
-
"grad_norm":
|
1338 |
-
"learning_rate": 8.
|
1339 |
-
"loss": 0.
|
1340 |
"step": 4625
|
1341 |
},
|
1342 |
{
|
1343 |
"epoch": 5.296127562642369,
|
1344 |
-
"grad_norm":
|
1345 |
-
"learning_rate": 7.
|
1346 |
-
"loss": 0.
|
1347 |
"step": 4650
|
1348 |
},
|
1349 |
{
|
1350 |
"epoch": 5.324601366742597,
|
1351 |
-
"grad_norm":
|
1352 |
-
"learning_rate": 7.
|
1353 |
-
"loss": 0.
|
1354 |
"step": 4675
|
1355 |
},
|
1356 |
{
|
1357 |
"epoch": 5.353075170842825,
|
1358 |
-
"grad_norm":
|
1359 |
-
"learning_rate": 6.
|
1360 |
-
"loss": 0.
|
1361 |
"step": 4700
|
1362 |
},
|
1363 |
{
|
1364 |
"epoch": 5.381548974943052,
|
1365 |
-
"grad_norm": 1.
|
1366 |
-
"learning_rate": 6.
|
1367 |
-
"loss": 0.
|
1368 |
"step": 4725
|
1369 |
},
|
1370 |
{
|
1371 |
"epoch": 5.41002277904328,
|
1372 |
-
"grad_norm": 1.
|
1373 |
-
"learning_rate": 5.
|
1374 |
-
"loss": 0.
|
1375 |
"step": 4750
|
1376 |
},
|
1377 |
{
|
1378 |
"epoch": 5.438496583143508,
|
1379 |
-
"grad_norm": 1.
|
1380 |
-
"learning_rate": 5.
|
1381 |
-
"loss": 0.
|
1382 |
"step": 4775
|
1383 |
},
|
1384 |
{
|
1385 |
"epoch": 5.466970387243736,
|
1386 |
-
"grad_norm": 1.
|
1387 |
-
"learning_rate": 4.
|
1388 |
-
"loss": 0.
|
1389 |
"step": 4800
|
1390 |
},
|
1391 |
{
|
1392 |
"epoch": 5.495444191343964,
|
1393 |
-
"grad_norm": 1.
|
1394 |
-
"learning_rate": 3.
|
1395 |
-
"loss": 0.
|
1396 |
"step": 4825
|
1397 |
},
|
1398 |
{
|
1399 |
"epoch": 5.523917995444191,
|
1400 |
-
"grad_norm": 1.
|
1401 |
-
"learning_rate": 3.
|
1402 |
-
"loss": 0.
|
1403 |
"step": 4850
|
1404 |
},
|
1405 |
{
|
1406 |
"epoch": 5.552391799544419,
|
1407 |
-
"grad_norm": 0.
|
1408 |
-
"learning_rate": 2.
|
1409 |
-
"loss": 0.
|
1410 |
"step": 4875
|
1411 |
},
|
1412 |
{
|
1413 |
"epoch": 5.5808656036446465,
|
1414 |
-
"grad_norm":
|
1415 |
-
"learning_rate": 2.
|
1416 |
-
"loss": 0.
|
1417 |
"step": 4900
|
1418 |
},
|
1419 |
{
|
1420 |
"epoch": 5.609339407744875,
|
1421 |
-
"grad_norm":
|
1422 |
-
"learning_rate": 1.
|
1423 |
-
"loss": 0.
|
1424 |
"step": 4925
|
1425 |
},
|
1426 |
{
|
1427 |
"epoch": 5.637813211845103,
|
1428 |
-
"grad_norm": 1.
|
1429 |
-
"learning_rate": 1.
|
1430 |
-
"loss": 0.
|
1431 |
"step": 4950
|
1432 |
},
|
1433 |
{
|
1434 |
"epoch": 5.66628701594533,
|
1435 |
-
"grad_norm": 0.
|
1436 |
-
"learning_rate": 5.
|
1437 |
-
"loss": 0.
|
1438 |
"step": 4975
|
1439 |
},
|
1440 |
{
|
1441 |
"epoch": 5.694760820045558,
|
1442 |
-
"grad_norm": 0.
|
1443 |
-
"learning_rate":
|
1444 |
-
"loss": 0.
|
1445 |
"step": 5000
|
1446 |
},
|
1447 |
{
|
1448 |
"epoch": 5.694760820045558,
|
1449 |
-
"eval_loss": 0.
|
1450 |
-
"eval_runtime":
|
1451 |
-
"eval_samples_per_second": 2.
|
1452 |
-
"eval_steps_per_second": 0.
|
1453 |
-
"eval_wer": 0.
|
1454 |
"step": 5000
|
1455 |
},
|
1456 |
{
|
1457 |
"epoch": 5.694760820045558,
|
1458 |
"step": 5000,
|
1459 |
"total_flos": 5.43429854134272e+20,
|
1460 |
-
"train_loss": 0.
|
1461 |
-
"train_runtime":
|
1462 |
-
"train_samples_per_second": 2.
|
1463 |
-
"train_steps_per_second": 0.
|
1464 |
}
|
1465 |
],
|
1466 |
"logging_steps": 25,
|
|
|
10 |
"log_history": [
|
11 |
{
|
12 |
"epoch": 0.02847380410022779,
|
13 |
+
"grad_norm": 8.29788875579834,
|
14 |
"learning_rate": 5.000000000000001e-07,
|
15 |
+
"loss": 1.7669,
|
16 |
"step": 25
|
17 |
},
|
18 |
{
|
19 |
"epoch": 0.05694760820045558,
|
20 |
+
"grad_norm": 5.953624725341797,
|
21 |
"learning_rate": 1.0000000000000002e-06,
|
22 |
+
"loss": 1.347,
|
23 |
"step": 50
|
24 |
},
|
25 |
{
|
26 |
"epoch": 0.08542141230068337,
|
27 |
+
"grad_norm": 6.313845157623291,
|
28 |
"learning_rate": 1.5e-06,
|
29 |
+
"loss": 1.0011,
|
30 |
"step": 75
|
31 |
},
|
32 |
{
|
33 |
"epoch": 0.11389521640091116,
|
34 |
+
"grad_norm": 5.495311260223389,
|
35 |
"learning_rate": 2.0000000000000003e-06,
|
36 |
+
"loss": 0.8525,
|
37 |
"step": 100
|
38 |
},
|
39 |
{
|
40 |
"epoch": 0.14236902050113895,
|
41 |
+
"grad_norm": 4.961582183837891,
|
42 |
"learning_rate": 2.5e-06,
|
43 |
+
"loss": 0.7632,
|
44 |
"step": 125
|
45 |
},
|
46 |
{
|
47 |
"epoch": 0.17084282460136674,
|
48 |
+
"grad_norm": 5.068117141723633,
|
49 |
"learning_rate": 3e-06,
|
50 |
+
"loss": 0.7467,
|
51 |
"step": 150
|
52 |
},
|
53 |
{
|
54 |
"epoch": 0.19931662870159453,
|
55 |
+
"grad_norm": 5.016373634338379,
|
56 |
"learning_rate": 3.5e-06,
|
57 |
+
"loss": 0.7178,
|
58 |
"step": 175
|
59 |
},
|
60 |
{
|
61 |
"epoch": 0.22779043280182232,
|
62 |
+
"grad_norm": 4.4371137619018555,
|
63 |
"learning_rate": 4.000000000000001e-06,
|
64 |
+
"loss": 0.6761,
|
65 |
"step": 200
|
66 |
},
|
67 |
{
|
68 |
"epoch": 0.25626423690205014,
|
69 |
+
"grad_norm": 5.090580940246582,
|
70 |
"learning_rate": 4.5e-06,
|
71 |
+
"loss": 0.6484,
|
72 |
"step": 225
|
73 |
},
|
74 |
{
|
75 |
"epoch": 0.2847380410022779,
|
76 |
+
"grad_norm": 4.779529094696045,
|
77 |
"learning_rate": 5e-06,
|
78 |
+
"loss": 0.677,
|
79 |
"step": 250
|
80 |
},
|
81 |
{
|
82 |
"epoch": 0.3132118451025057,
|
83 |
+
"grad_norm": 4.2773518562316895,
|
84 |
"learning_rate": 5.500000000000001e-06,
|
85 |
+
"loss": 0.6384,
|
86 |
"step": 275
|
87 |
},
|
88 |
{
|
89 |
"epoch": 0.3416856492027335,
|
90 |
+
"grad_norm": 4.421960830688477,
|
91 |
"learning_rate": 6e-06,
|
92 |
+
"loss": 0.6291,
|
93 |
"step": 300
|
94 |
},
|
95 |
{
|
96 |
"epoch": 0.3701594533029613,
|
97 |
+
"grad_norm": 4.473905563354492,
|
98 |
"learning_rate": 6.5000000000000004e-06,
|
99 |
+
"loss": 0.6275,
|
100 |
"step": 325
|
101 |
},
|
102 |
{
|
103 |
"epoch": 0.39863325740318906,
|
104 |
+
"grad_norm": 5.127504825592041,
|
105 |
"learning_rate": 7e-06,
|
106 |
+
"loss": 0.5969,
|
107 |
"step": 350
|
108 |
},
|
109 |
{
|
110 |
"epoch": 0.4271070615034169,
|
111 |
+
"grad_norm": 4.176600933074951,
|
112 |
"learning_rate": 7.500000000000001e-06,
|
113 |
+
"loss": 0.5706,
|
114 |
"step": 375
|
115 |
},
|
116 |
{
|
117 |
"epoch": 0.45558086560364464,
|
118 |
+
"grad_norm": 4.631809234619141,
|
119 |
"learning_rate": 8.000000000000001e-06,
|
120 |
+
"loss": 0.6083,
|
121 |
"step": 400
|
122 |
},
|
123 |
{
|
124 |
"epoch": 0.48405466970387245,
|
125 |
+
"grad_norm": 4.151340484619141,
|
126 |
"learning_rate": 8.5e-06,
|
127 |
+
"loss": 0.5777,
|
128 |
"step": 425
|
129 |
},
|
130 |
{
|
131 |
"epoch": 0.5125284738041003,
|
132 |
+
"grad_norm": 4.039161205291748,
|
133 |
"learning_rate": 9e-06,
|
134 |
+
"loss": 0.5529,
|
135 |
"step": 450
|
136 |
},
|
137 |
{
|
138 |
"epoch": 0.541002277904328,
|
139 |
+
"grad_norm": 4.89629602432251,
|
140 |
"learning_rate": 9.5e-06,
|
141 |
+
"loss": 0.5581,
|
142 |
"step": 475
|
143 |
},
|
144 |
{
|
145 |
"epoch": 0.5694760820045558,
|
146 |
+
"grad_norm": 11.599489212036133,
|
147 |
"learning_rate": 1e-05,
|
148 |
+
"loss": 0.5621,
|
149 |
"step": 500
|
150 |
},
|
151 |
{
|
152 |
"epoch": 0.5979498861047836,
|
153 |
+
"grad_norm": 4.748031139373779,
|
154 |
"learning_rate": 9.944444444444445e-06,
|
155 |
+
"loss": 0.5785,
|
156 |
"step": 525
|
157 |
},
|
158 |
{
|
159 |
"epoch": 0.6264236902050114,
|
160 |
+
"grad_norm": 4.0705485343933105,
|
161 |
"learning_rate": 9.88888888888889e-06,
|
162 |
+
"loss": 0.5625,
|
163 |
"step": 550
|
164 |
},
|
165 |
{
|
166 |
"epoch": 0.6548974943052391,
|
167 |
+
"grad_norm": 3.605722427368164,
|
168 |
"learning_rate": 9.833333333333333e-06,
|
169 |
+
"loss": 0.5613,
|
170 |
"step": 575
|
171 |
},
|
172 |
{
|
173 |
"epoch": 0.683371298405467,
|
174 |
+
"grad_norm": 3.7756550312042236,
|
175 |
"learning_rate": 9.777777777777779e-06,
|
176 |
+
"loss": 0.5521,
|
177 |
"step": 600
|
178 |
},
|
179 |
{
|
180 |
"epoch": 0.7118451025056948,
|
181 |
+
"grad_norm": 3.743063449859619,
|
182 |
"learning_rate": 9.722222222222223e-06,
|
183 |
+
"loss": 0.5521,
|
184 |
"step": 625
|
185 |
},
|
186 |
{
|
187 |
"epoch": 0.7403189066059226,
|
188 |
+
"grad_norm": 3.786343812942505,
|
189 |
"learning_rate": 9.666666666666667e-06,
|
190 |
+
"loss": 0.5374,
|
191 |
"step": 650
|
192 |
},
|
193 |
{
|
194 |
"epoch": 0.7687927107061503,
|
195 |
+
"grad_norm": 3.285397529602051,
|
196 |
"learning_rate": 9.611111111111112e-06,
|
197 |
+
"loss": 0.5155,
|
198 |
"step": 675
|
199 |
},
|
200 |
{
|
201 |
"epoch": 0.7972665148063781,
|
202 |
+
"grad_norm": 4.082733154296875,
|
203 |
"learning_rate": 9.555555555555556e-06,
|
204 |
+
"loss": 0.5266,
|
205 |
"step": 700
|
206 |
},
|
207 |
{
|
208 |
"epoch": 0.8257403189066059,
|
209 |
+
"grad_norm": 3.823092222213745,
|
210 |
"learning_rate": 9.5e-06,
|
211 |
+
"loss": 0.5022,
|
212 |
"step": 725
|
213 |
},
|
214 |
{
|
215 |
"epoch": 0.8542141230068337,
|
216 |
+
"grad_norm": 3.905947685241699,
|
217 |
"learning_rate": 9.444444444444445e-06,
|
218 |
+
"loss": 0.4924,
|
219 |
"step": 750
|
220 |
},
|
221 |
{
|
222 |
"epoch": 0.8826879271070615,
|
223 |
+
"grad_norm": 3.707880735397339,
|
224 |
"learning_rate": 9.38888888888889e-06,
|
225 |
+
"loss": 0.496,
|
226 |
"step": 775
|
227 |
},
|
228 |
{
|
229 |
"epoch": 0.9111617312072893,
|
230 |
+
"grad_norm": 3.18892502784729,
|
231 |
"learning_rate": 9.333333333333334e-06,
|
232 |
+
"loss": 0.5012,
|
233 |
"step": 800
|
234 |
},
|
235 |
{
|
236 |
"epoch": 0.9396355353075171,
|
237 |
+
"grad_norm": 3.8103792667388916,
|
238 |
"learning_rate": 9.277777777777778e-06,
|
239 |
+
"loss": 0.4767,
|
240 |
"step": 825
|
241 |
},
|
242 |
{
|
243 |
"epoch": 0.9681093394077449,
|
244 |
+
"grad_norm": 3.4749033451080322,
|
245 |
"learning_rate": 9.222222222222224e-06,
|
246 |
+
"loss": 0.5078,
|
247 |
"step": 850
|
248 |
},
|
249 |
{
|
250 |
"epoch": 0.9965831435079726,
|
251 |
+
"grad_norm": 3.836411476135254,
|
252 |
"learning_rate": 9.166666666666666e-06,
|
253 |
+
"loss": 0.481,
|
254 |
"step": 875
|
255 |
},
|
256 |
{
|
257 |
"epoch": 1.0250569476082005,
|
258 |
+
"grad_norm": 2.716602325439453,
|
259 |
"learning_rate": 9.111111111111112e-06,
|
260 |
+
"loss": 0.3908,
|
261 |
"step": 900
|
262 |
},
|
263 |
{
|
264 |
"epoch": 1.0535307517084282,
|
265 |
+
"grad_norm": 3.388803482055664,
|
266 |
"learning_rate": 9.055555555555556e-06,
|
267 |
+
"loss": 0.3558,
|
268 |
"step": 925
|
269 |
},
|
270 |
{
|
271 |
"epoch": 1.082004555808656,
|
272 |
+
"grad_norm": 3.0076682567596436,
|
273 |
"learning_rate": 9e-06,
|
274 |
+
"loss": 0.3697,
|
275 |
"step": 950
|
276 |
},
|
277 |
{
|
278 |
"epoch": 1.1104783599088839,
|
279 |
+
"grad_norm": 3.24129056930542,
|
280 |
"learning_rate": 8.944444444444446e-06,
|
281 |
+
"loss": 0.3739,
|
282 |
"step": 975
|
283 |
},
|
284 |
{
|
285 |
"epoch": 1.1389521640091116,
|
286 |
+
"grad_norm": 2.9381356239318848,
|
287 |
"learning_rate": 8.888888888888888e-06,
|
288 |
+
"loss": 0.3674,
|
289 |
"step": 1000
|
290 |
},
|
291 |
{
|
292 |
"epoch": 1.1389521640091116,
|
293 |
+
"eval_loss": 0.4708998501300812,
|
294 |
+
"eval_runtime": 2991.8311,
|
295 |
+
"eval_samples_per_second": 2.347,
|
296 |
"eval_steps_per_second": 0.147,
|
297 |
+
"eval_wer": 0.34162802747479043,
|
298 |
"step": 1000
|
299 |
},
|
300 |
{
|
301 |
"epoch": 1.1674259681093395,
|
302 |
+
"grad_norm": 3.604189872741699,
|
303 |
"learning_rate": 8.833333333333334e-06,
|
304 |
+
"loss": 0.3498,
|
305 |
"step": 1025
|
306 |
},
|
307 |
{
|
308 |
"epoch": 1.1958997722095672,
|
309 |
+
"grad_norm": 3.277430772781372,
|
310 |
"learning_rate": 8.777777777777778e-06,
|
311 |
+
"loss": 0.3628,
|
312 |
"step": 1050
|
313 |
},
|
314 |
{
|
315 |
"epoch": 1.224373576309795,
|
316 |
+
"grad_norm": 3.1814355850219727,
|
317 |
"learning_rate": 8.722222222222224e-06,
|
318 |
+
"loss": 0.3506,
|
319 |
"step": 1075
|
320 |
},
|
321 |
{
|
322 |
"epoch": 1.2528473804100229,
|
323 |
+
"grad_norm": 3.2664480209350586,
|
324 |
"learning_rate": 8.666666666666668e-06,
|
325 |
+
"loss": 0.3559,
|
326 |
"step": 1100
|
327 |
},
|
328 |
{
|
329 |
"epoch": 1.2813211845102506,
|
330 |
+
"grad_norm": 2.933966875076294,
|
331 |
"learning_rate": 8.611111111111112e-06,
|
332 |
+
"loss": 0.3697,
|
333 |
"step": 1125
|
334 |
},
|
335 |
{
|
336 |
"epoch": 1.3097949886104785,
|
337 |
+
"grad_norm": 2.8139538764953613,
|
338 |
"learning_rate": 8.555555555555556e-06,
|
339 |
+
"loss": 0.3674,
|
340 |
"step": 1150
|
341 |
},
|
342 |
{
|
343 |
"epoch": 1.3382687927107062,
|
344 |
+
"grad_norm": 3.1578900814056396,
|
345 |
"learning_rate": 8.5e-06,
|
346 |
+
"loss": 0.3411,
|
347 |
"step": 1175
|
348 |
},
|
349 |
{
|
350 |
"epoch": 1.366742596810934,
|
351 |
+
"grad_norm": 3.069774866104126,
|
352 |
"learning_rate": 8.444444444444446e-06,
|
353 |
+
"loss": 0.3655,
|
354 |
"step": 1200
|
355 |
},
|
356 |
{
|
357 |
"epoch": 1.3952164009111616,
|
358 |
+
"grad_norm": 2.6920642852783203,
|
359 |
"learning_rate": 8.38888888888889e-06,
|
360 |
+
"loss": 0.3492,
|
361 |
"step": 1225
|
362 |
},
|
363 |
{
|
364 |
"epoch": 1.4236902050113895,
|
365 |
+
"grad_norm": 2.7643768787384033,
|
366 |
"learning_rate": 8.333333333333334e-06,
|
367 |
+
"loss": 0.3534,
|
368 |
"step": 1250
|
369 |
},
|
370 |
{
|
371 |
"epoch": 1.4521640091116172,
|
372 |
+
"grad_norm": 3.1626651287078857,
|
373 |
"learning_rate": 8.277777777777778e-06,
|
374 |
+
"loss": 0.3317,
|
375 |
"step": 1275
|
376 |
},
|
377 |
{
|
378 |
"epoch": 1.4806378132118452,
|
379 |
+
"grad_norm": 3.417654514312744,
|
380 |
"learning_rate": 8.222222222222222e-06,
|
381 |
+
"loss": 0.3828,
|
382 |
"step": 1300
|
383 |
},
|
384 |
{
|
385 |
"epoch": 1.5091116173120729,
|
386 |
+
"grad_norm": 3.388624668121338,
|
387 |
"learning_rate": 8.166666666666668e-06,
|
388 |
+
"loss": 0.3611,
|
389 |
"step": 1325
|
390 |
},
|
391 |
{
|
392 |
"epoch": 1.5375854214123006,
|
393 |
+
"grad_norm": 2.685516834259033,
|
394 |
"learning_rate": 8.111111111111112e-06,
|
395 |
+
"loss": 0.3562,
|
396 |
"step": 1350
|
397 |
},
|
398 |
{
|
399 |
"epoch": 1.5660592255125285,
|
400 |
+
"grad_norm": 3.0524353981018066,
|
401 |
"learning_rate": 8.055555555555557e-06,
|
402 |
+
"loss": 0.3633,
|
403 |
"step": 1375
|
404 |
},
|
405 |
{
|
406 |
"epoch": 1.5945330296127562,
|
407 |
+
"grad_norm": 3.1776881217956543,
|
408 |
"learning_rate": 8.000000000000001e-06,
|
409 |
+
"loss": 0.353,
|
410 |
"step": 1400
|
411 |
},
|
412 |
{
|
413 |
"epoch": 1.6230068337129842,
|
414 |
+
"grad_norm": 3.046795129776001,
|
415 |
"learning_rate": 7.944444444444445e-06,
|
416 |
+
"loss": 0.3556,
|
417 |
"step": 1425
|
418 |
},
|
419 |
{
|
420 |
"epoch": 1.6514806378132119,
|
421 |
+
"grad_norm": 3.1991679668426514,
|
422 |
"learning_rate": 7.88888888888889e-06,
|
423 |
+
"loss": 0.3407,
|
424 |
"step": 1450
|
425 |
},
|
426 |
{
|
427 |
"epoch": 1.6799544419134396,
|
428 |
+
"grad_norm": 2.918931007385254,
|
429 |
"learning_rate": 7.833333333333333e-06,
|
430 |
+
"loss": 0.3608,
|
431 |
"step": 1475
|
432 |
},
|
433 |
{
|
434 |
"epoch": 1.7084282460136673,
|
435 |
+
"grad_norm": 3.0339889526367188,
|
436 |
"learning_rate": 7.77777777777778e-06,
|
437 |
+
"loss": 0.3539,
|
438 |
"step": 1500
|
439 |
},
|
440 |
{
|
441 |
"epoch": 1.7369020501138952,
|
442 |
+
"grad_norm": 3.138218402862549,
|
443 |
"learning_rate": 7.722222222222223e-06,
|
444 |
+
"loss": 0.345,
|
445 |
"step": 1525
|
446 |
},
|
447 |
{
|
448 |
"epoch": 1.7653758542141231,
|
449 |
+
"grad_norm": 3.70744252204895,
|
450 |
"learning_rate": 7.666666666666667e-06,
|
451 |
+
"loss": 0.347,
|
452 |
"step": 1550
|
453 |
},
|
454 |
{
|
455 |
"epoch": 1.7938496583143508,
|
456 |
+
"grad_norm": 2.8581366539001465,
|
457 |
"learning_rate": 7.611111111111111e-06,
|
458 |
+
"loss": 0.3541,
|
459 |
"step": 1575
|
460 |
},
|
461 |
{
|
462 |
"epoch": 1.8223234624145785,
|
463 |
+
"grad_norm": 2.9355995655059814,
|
464 |
"learning_rate": 7.555555555555556e-06,
|
465 |
+
"loss": 0.356,
|
466 |
"step": 1600
|
467 |
},
|
468 |
{
|
469 |
"epoch": 1.8507972665148062,
|
470 |
+
"grad_norm": 3.1599183082580566,
|
471 |
"learning_rate": 7.500000000000001e-06,
|
472 |
+
"loss": 0.3677,
|
473 |
"step": 1625
|
474 |
},
|
475 |
{
|
476 |
"epoch": 1.8792710706150342,
|
477 |
+
"grad_norm": 3.0059103965759277,
|
478 |
"learning_rate": 7.444444444444445e-06,
|
479 |
+
"loss": 0.3443,
|
480 |
"step": 1650
|
481 |
},
|
482 |
{
|
483 |
"epoch": 1.907744874715262,
|
484 |
+
"grad_norm": 2.8699076175689697,
|
485 |
"learning_rate": 7.38888888888889e-06,
|
486 |
+
"loss": 0.3593,
|
487 |
"step": 1675
|
488 |
},
|
489 |
{
|
490 |
"epoch": 1.9362186788154898,
|
491 |
+
"grad_norm": 3.279315233230591,
|
492 |
"learning_rate": 7.333333333333333e-06,
|
493 |
+
"loss": 0.3386,
|
494 |
"step": 1700
|
495 |
},
|
496 |
{
|
497 |
"epoch": 1.9646924829157175,
|
498 |
+
"grad_norm": 3.5267879962921143,
|
499 |
"learning_rate": 7.277777777777778e-06,
|
500 |
+
"loss": 0.3486,
|
501 |
"step": 1725
|
502 |
},
|
503 |
{
|
504 |
"epoch": 1.9931662870159452,
|
505 |
+
"grad_norm": 2.794961929321289,
|
506 |
"learning_rate": 7.222222222222223e-06,
|
507 |
+
"loss": 0.3372,
|
508 |
"step": 1750
|
509 |
},
|
510 |
{
|
511 |
"epoch": 2.021640091116173,
|
512 |
+
"grad_norm": 2.9796462059020996,
|
513 |
"learning_rate": 7.166666666666667e-06,
|
514 |
+
"loss": 0.252,
|
515 |
"step": 1775
|
516 |
},
|
517 |
{
|
518 |
"epoch": 2.050113895216401,
|
519 |
+
"grad_norm": 2.33910870552063,
|
520 |
"learning_rate": 7.111111111111112e-06,
|
521 |
+
"loss": 0.2207,
|
522 |
"step": 1800
|
523 |
},
|
524 |
{
|
525 |
"epoch": 2.078587699316629,
|
526 |
+
"grad_norm": 2.6431703567504883,
|
527 |
"learning_rate": 7.055555555555557e-06,
|
528 |
+
"loss": 0.2158,
|
529 |
"step": 1825
|
530 |
},
|
531 |
{
|
532 |
"epoch": 2.1070615034168565,
|
533 |
+
"grad_norm": 2.63135027885437,
|
534 |
"learning_rate": 7e-06,
|
535 |
+
"loss": 0.2172,
|
536 |
"step": 1850
|
537 |
},
|
538 |
{
|
539 |
"epoch": 2.135535307517084,
|
540 |
+
"grad_norm": 2.7133092880249023,
|
541 |
"learning_rate": 6.944444444444445e-06,
|
542 |
+
"loss": 0.2095,
|
543 |
"step": 1875
|
544 |
},
|
545 |
{
|
546 |
"epoch": 2.164009111617312,
|
547 |
+
"grad_norm": 2.2241263389587402,
|
548 |
"learning_rate": 6.88888888888889e-06,
|
549 |
+
"loss": 0.2118,
|
550 |
"step": 1900
|
551 |
},
|
552 |
{
|
553 |
"epoch": 2.19248291571754,
|
554 |
+
"grad_norm": 2.509427070617676,
|
555 |
"learning_rate": 6.833333333333334e-06,
|
556 |
+
"loss": 0.217,
|
557 |
"step": 1925
|
558 |
},
|
559 |
{
|
560 |
"epoch": 2.2209567198177678,
|
561 |
+
"grad_norm": 3.031168222427368,
|
562 |
"learning_rate": 6.777777777777779e-06,
|
563 |
+
"loss": 0.2239,
|
564 |
"step": 1950
|
565 |
},
|
566 |
{
|
567 |
"epoch": 2.2494305239179955,
|
568 |
+
"grad_norm": 2.869800567626953,
|
569 |
"learning_rate": 6.7222222222222235e-06,
|
570 |
+
"loss": 0.2247,
|
571 |
"step": 1975
|
572 |
},
|
573 |
{
|
574 |
"epoch": 2.277904328018223,
|
575 |
+
"grad_norm": 2.5018348693847656,
|
576 |
"learning_rate": 6.666666666666667e-06,
|
577 |
+
"loss": 0.2228,
|
578 |
"step": 2000
|
579 |
},
|
580 |
{
|
581 |
"epoch": 2.277904328018223,
|
582 |
+
"eval_loss": 0.4358457922935486,
|
583 |
+
"eval_runtime": 3078.2421,
|
584 |
+
"eval_samples_per_second": 2.281,
|
585 |
+
"eval_steps_per_second": 0.143,
|
586 |
+
"eval_wer": 0.3116688145185933,
|
587 |
"step": 2000
|
588 |
},
|
589 |
{
|
590 |
"epoch": 2.306378132118451,
|
591 |
+
"grad_norm": 3.0430192947387695,
|
592 |
"learning_rate": 6.6111111111111115e-06,
|
593 |
+
"loss": 0.2138,
|
594 |
"step": 2025
|
595 |
},
|
596 |
{
|
597 |
"epoch": 2.334851936218679,
|
598 |
+
"grad_norm": 2.651179075241089,
|
599 |
"learning_rate": 6.555555555555556e-06,
|
600 |
+
"loss": 0.233,
|
601 |
"step": 2050
|
602 |
},
|
603 |
{
|
604 |
"epoch": 2.3633257403189067,
|
605 |
+
"grad_norm": 2.935361623764038,
|
606 |
"learning_rate": 6.5000000000000004e-06,
|
607 |
+
"loss": 0.207,
|
608 |
"step": 2075
|
609 |
},
|
610 |
{
|
611 |
"epoch": 2.3917995444191344,
|
612 |
+
"grad_norm": 2.7219200134277344,
|
613 |
"learning_rate": 6.444444444444445e-06,
|
614 |
+
"loss": 0.2291,
|
615 |
"step": 2100
|
616 |
},
|
617 |
{
|
618 |
"epoch": 2.420273348519362,
|
619 |
+
"grad_norm": 2.854238510131836,
|
620 |
"learning_rate": 6.3888888888888885e-06,
|
621 |
+
"loss": 0.2372,
|
622 |
"step": 2125
|
623 |
},
|
624 |
{
|
625 |
"epoch": 2.44874715261959,
|
626 |
+
"grad_norm": 2.149994373321533,
|
627 |
"learning_rate": 6.333333333333333e-06,
|
628 |
+
"loss": 0.2304,
|
629 |
"step": 2150
|
630 |
},
|
631 |
{
|
632 |
"epoch": 2.477220956719818,
|
633 |
+
"grad_norm": 2.346008062362671,
|
634 |
"learning_rate": 6.277777777777778e-06,
|
635 |
+
"loss": 0.2181,
|
636 |
"step": 2175
|
637 |
},
|
638 |
{
|
639 |
"epoch": 2.5056947608200457,
|
640 |
+
"grad_norm": 2.859161138534546,
|
641 |
"learning_rate": 6.222222222222223e-06,
|
642 |
+
"loss": 0.2235,
|
643 |
"step": 2200
|
644 |
},
|
645 |
{
|
646 |
"epoch": 2.5341685649202734,
|
647 |
+
"grad_norm": 2.6515421867370605,
|
648 |
"learning_rate": 6.166666666666667e-06,
|
649 |
+
"loss": 0.2218,
|
650 |
"step": 2225
|
651 |
},
|
652 |
{
|
653 |
"epoch": 2.562642369020501,
|
654 |
+
"grad_norm": 2.8457796573638916,
|
655 |
"learning_rate": 6.111111111111112e-06,
|
656 |
+
"loss": 0.2197,
|
657 |
"step": 2250
|
658 |
},
|
659 |
{
|
660 |
"epoch": 2.591116173120729,
|
661 |
+
"grad_norm": 2.9670958518981934,
|
662 |
"learning_rate": 6.055555555555555e-06,
|
663 |
+
"loss": 0.2179,
|
664 |
"step": 2275
|
665 |
},
|
666 |
{
|
667 |
"epoch": 2.619589977220957,
|
668 |
+
"grad_norm": 3.2194952964782715,
|
669 |
"learning_rate": 6e-06,
|
670 |
+
"loss": 0.2201,
|
671 |
"step": 2300
|
672 |
},
|
673 |
{
|
674 |
"epoch": 2.6480637813211843,
|
675 |
+
"grad_norm": 2.424835205078125,
|
676 |
"learning_rate": 5.944444444444445e-06,
|
677 |
+
"loss": 0.2295,
|
678 |
"step": 2325
|
679 |
},
|
680 |
{
|
681 |
"epoch": 2.6765375854214124,
|
682 |
+
"grad_norm": 2.3368935585021973,
|
683 |
"learning_rate": 5.88888888888889e-06,
|
684 |
+
"loss": 0.211,
|
685 |
"step": 2350
|
686 |
},
|
687 |
{
|
688 |
"epoch": 2.70501138952164,
|
689 |
+
"grad_norm": 2.0860483646392822,
|
690 |
"learning_rate": 5.833333333333334e-06,
|
691 |
+
"loss": 0.2171,
|
692 |
"step": 2375
|
693 |
},
|
694 |
{
|
695 |
"epoch": 2.733485193621868,
|
696 |
+
"grad_norm": 2.6217784881591797,
|
697 |
"learning_rate": 5.777777777777778e-06,
|
698 |
+
"loss": 0.2047,
|
699 |
"step": 2400
|
700 |
},
|
701 |
{
|
702 |
"epoch": 2.7619589977220955,
|
703 |
+
"grad_norm": 3.0176894664764404,
|
704 |
"learning_rate": 5.722222222222222e-06,
|
705 |
+
"loss": 0.2248,
|
706 |
"step": 2425
|
707 |
},
|
708 |
{
|
709 |
"epoch": 2.7904328018223232,
|
710 |
+
"grad_norm": 2.360590696334839,
|
711 |
"learning_rate": 5.666666666666667e-06,
|
712 |
+
"loss": 0.2101,
|
713 |
"step": 2450
|
714 |
},
|
715 |
{
|
716 |
"epoch": 2.8189066059225514,
|
717 |
+
"grad_norm": 2.451885223388672,
|
718 |
"learning_rate": 5.611111111111112e-06,
|
719 |
+
"loss": 0.2275,
|
720 |
"step": 2475
|
721 |
},
|
722 |
{
|
723 |
"epoch": 2.847380410022779,
|
724 |
+
"grad_norm": 2.3589582443237305,
|
725 |
"learning_rate": 5.555555555555557e-06,
|
726 |
+
"loss": 0.217,
|
727 |
"step": 2500
|
728 |
},
|
729 |
{
|
730 |
"epoch": 2.875854214123007,
|
731 |
+
"grad_norm": 2.6570801734924316,
|
732 |
"learning_rate": 5.500000000000001e-06,
|
733 |
+
"loss": 0.2304,
|
734 |
"step": 2525
|
735 |
},
|
736 |
{
|
737 |
"epoch": 2.9043280182232345,
|
738 |
+
"grad_norm": 2.6280150413513184,
|
739 |
"learning_rate": 5.444444444444445e-06,
|
740 |
+
"loss": 0.2105,
|
741 |
"step": 2550
|
742 |
},
|
743 |
{
|
744 |
"epoch": 2.932801822323462,
|
745 |
+
"grad_norm": 2.4569592475891113,
|
746 |
"learning_rate": 5.388888888888889e-06,
|
747 |
+
"loss": 0.2027,
|
748 |
"step": 2575
|
749 |
},
|
750 |
{
|
751 |
"epoch": 2.9612756264236904,
|
752 |
+
"grad_norm": 3.0375263690948486,
|
753 |
"learning_rate": 5.333333333333334e-06,
|
754 |
+
"loss": 0.2164,
|
755 |
"step": 2600
|
756 |
},
|
757 |
{
|
758 |
"epoch": 2.989749430523918,
|
759 |
+
"grad_norm": 2.2923061847686768,
|
760 |
"learning_rate": 5.2777777777777785e-06,
|
761 |
+
"loss": 0.2194,
|
762 |
"step": 2625
|
763 |
},
|
764 |
{
|
765 |
"epoch": 3.0182232346241458,
|
766 |
+
"grad_norm": 2.045480966567993,
|
767 |
"learning_rate": 5.2222222222222226e-06,
|
768 |
+
"loss": 0.1584,
|
769 |
"step": 2650
|
770 |
},
|
771 |
{
|
772 |
"epoch": 3.0466970387243735,
|
773 |
+
"grad_norm": 2.0926032066345215,
|
774 |
"learning_rate": 5.1666666666666675e-06,
|
775 |
+
"loss": 0.1257,
|
776 |
"step": 2675
|
777 |
},
|
778 |
{
|
779 |
"epoch": 3.075170842824601,
|
780 |
+
"grad_norm": 2.2969894409179688,
|
781 |
"learning_rate": 5.1111111111111115e-06,
|
782 |
+
"loss": 0.1211,
|
783 |
"step": 2700
|
784 |
},
|
785 |
{
|
786 |
"epoch": 3.1036446469248293,
|
787 |
+
"grad_norm": 2.1000475883483887,
|
788 |
"learning_rate": 5.0555555555555555e-06,
|
789 |
+
"loss": 0.1225,
|
790 |
"step": 2725
|
791 |
},
|
792 |
{
|
793 |
"epoch": 3.132118451025057,
|
794 |
+
"grad_norm": 2.593099594116211,
|
795 |
"learning_rate": 5e-06,
|
796 |
+
"loss": 0.1289,
|
797 |
"step": 2750
|
798 |
},
|
799 |
{
|
800 |
"epoch": 3.1605922551252847,
|
801 |
+
"grad_norm": 1.8630212545394897,
|
802 |
"learning_rate": 4.944444444444445e-06,
|
803 |
+
"loss": 0.129,
|
804 |
"step": 2775
|
805 |
},
|
806 |
{
|
807 |
"epoch": 3.1890660592255125,
|
808 |
+
"grad_norm": 2.0034408569335938,
|
809 |
"learning_rate": 4.888888888888889e-06,
|
810 |
+
"loss": 0.1222,
|
811 |
"step": 2800
|
812 |
},
|
813 |
{
|
814 |
"epoch": 3.21753986332574,
|
815 |
+
"grad_norm": 2.4184627532958984,
|
816 |
"learning_rate": 4.833333333333333e-06,
|
817 |
+
"loss": 0.1382,
|
818 |
"step": 2825
|
819 |
},
|
820 |
{
|
821 |
"epoch": 3.2460136674259683,
|
822 |
+
"grad_norm": 2.2710211277008057,
|
823 |
"learning_rate": 4.777777777777778e-06,
|
824 |
+
"loss": 0.1185,
|
825 |
"step": 2850
|
826 |
},
|
827 |
{
|
828 |
"epoch": 3.274487471526196,
|
829 |
+
"grad_norm": 2.108302116394043,
|
830 |
"learning_rate": 4.722222222222222e-06,
|
831 |
+
"loss": 0.1256,
|
832 |
"step": 2875
|
833 |
},
|
834 |
{
|
835 |
"epoch": 3.3029612756264237,
|
836 |
+
"grad_norm": 2.66013503074646,
|
837 |
"learning_rate": 4.666666666666667e-06,
|
838 |
+
"loss": 0.126,
|
839 |
"step": 2900
|
840 |
},
|
841 |
{
|
842 |
"epoch": 3.3314350797266514,
|
843 |
+
"grad_norm": 2.502146005630493,
|
844 |
"learning_rate": 4.611111111111112e-06,
|
845 |
+
"loss": 0.1188,
|
846 |
"step": 2925
|
847 |
},
|
848 |
{
|
849 |
"epoch": 3.359908883826879,
|
850 |
+
"grad_norm": 2.9654130935668945,
|
851 |
"learning_rate": 4.555555555555556e-06,
|
852 |
+
"loss": 0.1265,
|
853 |
"step": 2950
|
854 |
},
|
855 |
{
|
856 |
"epoch": 3.3883826879271073,
|
857 |
+
"grad_norm": 2.2393388748168945,
|
858 |
"learning_rate": 4.5e-06,
|
859 |
+
"loss": 0.1197,
|
860 |
"step": 2975
|
861 |
},
|
862 |
{
|
863 |
"epoch": 3.416856492027335,
|
864 |
+
"grad_norm": 2.5502378940582275,
|
865 |
"learning_rate": 4.444444444444444e-06,
|
866 |
+
"loss": 0.1266,
|
867 |
"step": 3000
|
868 |
},
|
869 |
{
|
870 |
"epoch": 3.416856492027335,
|
871 |
+
"eval_loss": 0.45878028869628906,
|
872 |
+
"eval_runtime": 3136.0137,
|
873 |
+
"eval_samples_per_second": 2.239,
|
874 |
+
"eval_steps_per_second": 0.14,
|
875 |
+
"eval_wer": 0.3070321114934435,
|
876 |
"step": 3000
|
877 |
},
|
878 |
{
|
879 |
"epoch": 3.4453302961275627,
|
880 |
+
"grad_norm": 2.695366382598877,
|
881 |
"learning_rate": 4.388888888888889e-06,
|
882 |
+
"loss": 0.1261,
|
883 |
"step": 3025
|
884 |
},
|
885 |
{
|
886 |
"epoch": 3.4738041002277904,
|
887 |
+
"grad_norm": 2.2533366680145264,
|
888 |
"learning_rate": 4.333333333333334e-06,
|
889 |
+
"loss": 0.1153,
|
890 |
"step": 3050
|
891 |
},
|
892 |
{
|
893 |
"epoch": 3.502277904328018,
|
894 |
+
"grad_norm": 2.5580222606658936,
|
895 |
"learning_rate": 4.277777777777778e-06,
|
896 |
+
"loss": 0.1275,
|
897 |
"step": 3075
|
898 |
},
|
899 |
{
|
900 |
"epoch": 3.5307517084282463,
|
901 |
+
"grad_norm": 2.1148972511291504,
|
902 |
"learning_rate": 4.222222222222223e-06,
|
903 |
+
"loss": 0.1254,
|
904 |
"step": 3100
|
905 |
},
|
906 |
{
|
907 |
"epoch": 3.559225512528474,
|
908 |
+
"grad_norm": 2.1242175102233887,
|
909 |
"learning_rate": 4.166666666666667e-06,
|
910 |
+
"loss": 0.1265,
|
911 |
"step": 3125
|
912 |
},
|
913 |
{
|
914 |
"epoch": 3.5876993166287017,
|
915 |
+
"grad_norm": 2.1025946140289307,
|
916 |
"learning_rate": 4.111111111111111e-06,
|
917 |
+
"loss": 0.1193,
|
918 |
"step": 3150
|
919 |
},
|
920 |
{
|
921 |
"epoch": 3.6161731207289294,
|
922 |
+
"grad_norm": 1.925126314163208,
|
923 |
"learning_rate": 4.055555555555556e-06,
|
924 |
+
"loss": 0.1243,
|
925 |
"step": 3175
|
926 |
},
|
927 |
{
|
928 |
"epoch": 3.644646924829157,
|
929 |
+
"grad_norm": 2.383075714111328,
|
930 |
"learning_rate": 4.000000000000001e-06,
|
931 |
+
"loss": 0.1192,
|
932 |
"step": 3200
|
933 |
},
|
934 |
{
|
935 |
"epoch": 3.6731207289293852,
|
936 |
+
"grad_norm": 1.961838960647583,
|
937 |
"learning_rate": 3.944444444444445e-06,
|
938 |
+
"loss": 0.1192,
|
939 |
"step": 3225
|
940 |
},
|
941 |
{
|
942 |
"epoch": 3.7015945330296125,
|
943 |
+
"grad_norm": 2.1979856491088867,
|
944 |
"learning_rate": 3.88888888888889e-06,
|
945 |
+
"loss": 0.1276,
|
946 |
"step": 3250
|
947 |
},
|
948 |
{
|
949 |
"epoch": 3.7300683371298406,
|
950 |
+
"grad_norm": 1.8826966285705566,
|
951 |
"learning_rate": 3.833333333333334e-06,
|
952 |
+
"loss": 0.1184,
|
953 |
"step": 3275
|
954 |
},
|
955 |
{
|
956 |
"epoch": 3.7585421412300684,
|
957 |
+
"grad_norm": 2.3589093685150146,
|
958 |
"learning_rate": 3.777777777777778e-06,
|
959 |
+
"loss": 0.1193,
|
960 |
"step": 3300
|
961 |
},
|
962 |
{
|
963 |
"epoch": 3.787015945330296,
|
964 |
+
"grad_norm": 2.299064874649048,
|
965 |
"learning_rate": 3.7222222222222225e-06,
|
966 |
+
"loss": 0.1218,
|
967 |
"step": 3325
|
968 |
},
|
969 |
{
|
970 |
"epoch": 3.8154897494305238,
|
971 |
+
"grad_norm": 2.088062047958374,
|
972 |
"learning_rate": 3.6666666666666666e-06,
|
973 |
+
"loss": 0.1241,
|
974 |
"step": 3350
|
975 |
},
|
976 |
{
|
977 |
"epoch": 3.8439635535307515,
|
978 |
+
"grad_norm": 2.695964813232422,
|
979 |
"learning_rate": 3.6111111111111115e-06,
|
980 |
+
"loss": 0.1297,
|
981 |
"step": 3375
|
982 |
},
|
983 |
{
|
984 |
"epoch": 3.8724373576309796,
|
985 |
+
"grad_norm": 2.341107130050659,
|
986 |
"learning_rate": 3.555555555555556e-06,
|
987 |
+
"loss": 0.1287,
|
988 |
"step": 3400
|
989 |
},
|
990 |
{
|
991 |
"epoch": 3.9009111617312073,
|
992 |
+
"grad_norm": 2.7177109718322754,
|
993 |
"learning_rate": 3.5e-06,
|
994 |
+
"loss": 0.1222,
|
995 |
"step": 3425
|
996 |
},
|
997 |
{
|
998 |
"epoch": 3.929384965831435,
|
999 |
+
"grad_norm": 2.182626962661743,
|
1000 |
"learning_rate": 3.444444444444445e-06,
|
1001 |
+
"loss": 0.1257,
|
1002 |
"step": 3450
|
1003 |
},
|
1004 |
{
|
1005 |
"epoch": 3.9578587699316627,
|
1006 |
+
"grad_norm": 2.2234408855438232,
|
1007 |
"learning_rate": 3.3888888888888893e-06,
|
1008 |
+
"loss": 0.121,
|
1009 |
"step": 3475
|
1010 |
},
|
1011 |
{
|
1012 |
"epoch": 3.9863325740318905,
|
1013 |
+
"grad_norm": 2.3420960903167725,
|
1014 |
"learning_rate": 3.3333333333333333e-06,
|
1015 |
+
"loss": 0.1232,
|
1016 |
"step": 3500
|
1017 |
},
|
1018 |
{
|
1019 |
"epoch": 4.014806378132119,
|
1020 |
+
"grad_norm": 1.4395020008087158,
|
1021 |
"learning_rate": 3.277777777777778e-06,
|
1022 |
+
"loss": 0.0916,
|
1023 |
"step": 3525
|
1024 |
},
|
1025 |
{
|
1026 |
"epoch": 4.043280182232346,
|
1027 |
+
"grad_norm": 1.6081222295761108,
|
1028 |
"learning_rate": 3.2222222222222227e-06,
|
1029 |
+
"loss": 0.0605,
|
1030 |
"step": 3550
|
1031 |
},
|
1032 |
{
|
1033 |
"epoch": 4.071753986332574,
|
1034 |
+
"grad_norm": 1.7407175302505493,
|
1035 |
"learning_rate": 3.1666666666666667e-06,
|
1036 |
+
"loss": 0.0585,
|
1037 |
"step": 3575
|
1038 |
},
|
1039 |
{
|
1040 |
"epoch": 4.100227790432802,
|
1041 |
+
"grad_norm": 1.4883263111114502,
|
1042 |
"learning_rate": 3.1111111111111116e-06,
|
1043 |
+
"loss": 0.0605,
|
1044 |
"step": 3600
|
1045 |
},
|
1046 |
{
|
1047 |
"epoch": 4.128701594533029,
|
1048 |
+
"grad_norm": 1.864080786705017,
|
1049 |
"learning_rate": 3.055555555555556e-06,
|
1050 |
+
"loss": 0.0583,
|
1051 |
"step": 3625
|
1052 |
},
|
1053 |
{
|
1054 |
"epoch": 4.157175398633258,
|
1055 |
+
"grad_norm": 2.384763479232788,
|
1056 |
"learning_rate": 3e-06,
|
1057 |
+
"loss": 0.0608,
|
1058 |
"step": 3650
|
1059 |
},
|
1060 |
{
|
1061 |
"epoch": 4.185649202733485,
|
1062 |
+
"grad_norm": 1.896714448928833,
|
1063 |
"learning_rate": 2.944444444444445e-06,
|
1064 |
+
"loss": 0.0637,
|
1065 |
"step": 3675
|
1066 |
},
|
1067 |
{
|
1068 |
"epoch": 4.214123006833713,
|
1069 |
+
"grad_norm": 2.230041265487671,
|
1070 |
"learning_rate": 2.888888888888889e-06,
|
1071 |
+
"loss": 0.0651,
|
1072 |
"step": 3700
|
1073 |
},
|
1074 |
{
|
1075 |
"epoch": 4.242596810933941,
|
1076 |
+
"grad_norm": 1.9410207271575928,
|
1077 |
"learning_rate": 2.8333333333333335e-06,
|
1078 |
+
"loss": 0.0594,
|
1079 |
"step": 3725
|
1080 |
},
|
1081 |
{
|
1082 |
"epoch": 4.271070615034168,
|
1083 |
+
"grad_norm": 2.4352822303771973,
|
1084 |
"learning_rate": 2.7777777777777783e-06,
|
1085 |
+
"loss": 0.0581,
|
1086 |
"step": 3750
|
1087 |
},
|
1088 |
{
|
1089 |
"epoch": 4.2995444191343966,
|
1090 |
+
"grad_norm": 1.998757243156433,
|
1091 |
"learning_rate": 2.7222222222222224e-06,
|
1092 |
+
"loss": 0.0627,
|
1093 |
"step": 3775
|
1094 |
},
|
1095 |
{
|
1096 |
"epoch": 4.328018223234624,
|
1097 |
+
"grad_norm": 1.773343563079834,
|
1098 |
"learning_rate": 2.666666666666667e-06,
|
1099 |
+
"loss": 0.0639,
|
1100 |
"step": 3800
|
1101 |
},
|
1102 |
{
|
1103 |
"epoch": 4.356492027334852,
|
1104 |
+
"grad_norm": 2.35654616355896,
|
1105 |
"learning_rate": 2.6111111111111113e-06,
|
1106 |
+
"loss": 0.0578,
|
1107 |
"step": 3825
|
1108 |
},
|
1109 |
{
|
1110 |
"epoch": 4.38496583143508,
|
1111 |
+
"grad_norm": 2.2221155166625977,
|
1112 |
"learning_rate": 2.5555555555555557e-06,
|
1113 |
+
"loss": 0.0585,
|
1114 |
"step": 3850
|
1115 |
},
|
1116 |
{
|
1117 |
"epoch": 4.413439635535307,
|
1118 |
+
"grad_norm": 1.8902034759521484,
|
1119 |
"learning_rate": 2.5e-06,
|
1120 |
+
"loss": 0.0598,
|
1121 |
"step": 3875
|
1122 |
},
|
1123 |
{
|
1124 |
"epoch": 4.4419134396355355,
|
1125 |
+
"grad_norm": 1.73625910282135,
|
1126 |
+
"learning_rate": 2.446666666666667e-06,
|
1127 |
+
"loss": 0.0616,
|
1128 |
"step": 3900
|
1129 |
},
|
1130 |
{
|
1131 |
"epoch": 4.470387243735763,
|
1132 |
+
"grad_norm": 1.6135858297348022,
|
1133 |
+
"learning_rate": 2.3911111111111113e-06,
|
1134 |
+
"loss": 0.0592,
|
1135 |
"step": 3925
|
1136 |
},
|
1137 |
{
|
1138 |
"epoch": 4.498861047835991,
|
1139 |
+
"grad_norm": 2.070269823074341,
|
1140 |
+
"learning_rate": 2.3355555555555557e-06,
|
1141 |
+
"loss": 0.0573,
|
1142 |
"step": 3950
|
1143 |
},
|
1144 |
{
|
1145 |
"epoch": 4.527334851936219,
|
1146 |
+
"grad_norm": 1.600358247756958,
|
1147 |
+
"learning_rate": 2.28e-06,
|
1148 |
+
"loss": 0.0607,
|
1149 |
"step": 3975
|
1150 |
},
|
1151 |
{
|
1152 |
"epoch": 4.555808656036446,
|
1153 |
+
"grad_norm": 2.013101577758789,
|
1154 |
+
"learning_rate": 2.2244444444444447e-06,
|
1155 |
+
"loss": 0.0611,
|
1156 |
"step": 4000
|
1157 |
},
|
1158 |
{
|
1159 |
"epoch": 4.555808656036446,
|
1160 |
+
"eval_loss": 0.5140041708946228,
|
1161 |
+
"eval_runtime": 2974.0961,
|
1162 |
+
"eval_samples_per_second": 2.361,
|
1163 |
"eval_steps_per_second": 0.148,
|
1164 |
+
"eval_wer": 0.30393654758267014,
|
1165 |
"step": 4000
|
1166 |
},
|
1167 |
{
|
1168 |
"epoch": 4.5842824601366745,
|
1169 |
+
"grad_norm": 2.0340895652770996,
|
1170 |
+
"learning_rate": 2.168888888888889e-06,
|
1171 |
+
"loss": 0.0579,
|
1172 |
"step": 4025
|
1173 |
},
|
1174 |
{
|
1175 |
"epoch": 4.612756264236902,
|
1176 |
+
"grad_norm": 1.8350932598114014,
|
1177 |
+
"learning_rate": 2.1133333333333336e-06,
|
1178 |
+
"loss": 0.0567,
|
1179 |
"step": 4050
|
1180 |
},
|
1181 |
{
|
1182 |
"epoch": 4.64123006833713,
|
1183 |
+
"grad_norm": 2.042941093444824,
|
1184 |
+
"learning_rate": 2.057777777777778e-06,
|
1185 |
+
"loss": 0.0688,
|
1186 |
"step": 4075
|
1187 |
},
|
1188 |
{
|
1189 |
"epoch": 4.669703872437358,
|
1190 |
+
"grad_norm": 2.5973660945892334,
|
1191 |
+
"learning_rate": 2.0022222222222225e-06,
|
1192 |
+
"loss": 0.057,
|
1193 |
"step": 4100
|
1194 |
},
|
1195 |
{
|
1196 |
"epoch": 4.698177676537585,
|
1197 |
+
"grad_norm": 1.658544659614563,
|
1198 |
+
"learning_rate": 1.9466666666666665e-06,
|
1199 |
+
"loss": 0.0573,
|
1200 |
"step": 4125
|
1201 |
},
|
1202 |
{
|
1203 |
"epoch": 4.7266514806378135,
|
1204 |
+
"grad_norm": 1.918231725692749,
|
1205 |
+
"learning_rate": 1.8911111111111114e-06,
|
1206 |
+
"loss": 0.0591,
|
1207 |
"step": 4150
|
1208 |
},
|
1209 |
{
|
1210 |
"epoch": 4.755125284738041,
|
1211 |
+
"grad_norm": 1.6860567331314087,
|
1212 |
+
"learning_rate": 1.8355555555555557e-06,
|
1213 |
+
"loss": 0.0598,
|
1214 |
"step": 4175
|
1215 |
},
|
1216 |
{
|
1217 |
"epoch": 4.783599088838269,
|
1218 |
+
"grad_norm": 1.7576823234558105,
|
1219 |
+
"learning_rate": 1.7800000000000001e-06,
|
1220 |
+
"loss": 0.0606,
|
1221 |
"step": 4200
|
1222 |
},
|
1223 |
{
|
1224 |
"epoch": 4.812072892938497,
|
1225 |
+
"grad_norm": 2.130852699279785,
|
1226 |
+
"learning_rate": 1.7244444444444448e-06,
|
1227 |
+
"loss": 0.0584,
|
1228 |
"step": 4225
|
1229 |
},
|
1230 |
{
|
1231 |
"epoch": 4.840546697038724,
|
1232 |
+
"grad_norm": 2.2588632106781006,
|
1233 |
+
"learning_rate": 1.668888888888889e-06,
|
1234 |
+
"loss": 0.0594,
|
1235 |
"step": 4250
|
1236 |
},
|
1237 |
{
|
1238 |
"epoch": 4.8690205011389525,
|
1239 |
+
"grad_norm": 1.6730296611785889,
|
1240 |
+
"learning_rate": 1.6133333333333335e-06,
|
1241 |
+
"loss": 0.0542,
|
1242 |
"step": 4275
|
1243 |
},
|
1244 |
{
|
1245 |
"epoch": 4.89749430523918,
|
1246 |
+
"grad_norm": 2.031381607055664,
|
1247 |
+
"learning_rate": 1.5577777777777777e-06,
|
1248 |
+
"loss": 0.0534,
|
1249 |
"step": 4300
|
1250 |
},
|
1251 |
{
|
1252 |
"epoch": 4.925968109339408,
|
1253 |
+
"grad_norm": 1.2967132329940796,
|
1254 |
+
"learning_rate": 1.5022222222222224e-06,
|
1255 |
+
"loss": 0.053,
|
1256 |
"step": 4325
|
1257 |
},
|
1258 |
{
|
1259 |
"epoch": 4.954441913439636,
|
1260 |
+
"grad_norm": 2.0789716243743896,
|
1261 |
+
"learning_rate": 1.4466666666666669e-06,
|
1262 |
+
"loss": 0.0546,
|
1263 |
"step": 4350
|
1264 |
},
|
1265 |
{
|
1266 |
"epoch": 4.982915717539863,
|
1267 |
+
"grad_norm": 1.8398990631103516,
|
1268 |
+
"learning_rate": 1.3911111111111111e-06,
|
1269 |
+
"loss": 0.0548,
|
1270 |
"step": 4375
|
1271 |
},
|
1272 |
{
|
1273 |
"epoch": 5.011389521640091,
|
1274 |
+
"grad_norm": 1.2099305391311646,
|
1275 |
+
"learning_rate": 1.3355555555555558e-06,
|
1276 |
+
"loss": 0.0464,
|
1277 |
"step": 4400
|
1278 |
},
|
1279 |
{
|
1280 |
"epoch": 5.039863325740319,
|
1281 |
+
"grad_norm": 0.9306023120880127,
|
1282 |
+
"learning_rate": 1.28e-06,
|
1283 |
+
"loss": 0.0286,
|
1284 |
"step": 4425
|
1285 |
},
|
1286 |
{
|
1287 |
"epoch": 5.068337129840547,
|
1288 |
+
"grad_norm": 1.1420648097991943,
|
1289 |
+
"learning_rate": 1.2244444444444445e-06,
|
1290 |
+
"loss": 0.0252,
|
1291 |
"step": 4450
|
1292 |
},
|
1293 |
{
|
1294 |
"epoch": 5.096810933940774,
|
1295 |
+
"grad_norm": 1.3366564512252808,
|
1296 |
+
"learning_rate": 1.168888888888889e-06,
|
1297 |
+
"loss": 0.0251,
|
1298 |
"step": 4475
|
1299 |
},
|
1300 |
{
|
1301 |
"epoch": 5.125284738041002,
|
1302 |
+
"grad_norm": 1.437525749206543,
|
1303 |
+
"learning_rate": 1.1133333333333334e-06,
|
1304 |
+
"loss": 0.0252,
|
1305 |
"step": 4500
|
1306 |
},
|
1307 |
{
|
1308 |
"epoch": 5.15375854214123,
|
1309 |
+
"grad_norm": 0.8598686456680298,
|
1310 |
+
"learning_rate": 1.0577777777777779e-06,
|
1311 |
+
"loss": 0.0249,
|
1312 |
"step": 4525
|
1313 |
},
|
1314 |
{
|
1315 |
"epoch": 5.182232346241458,
|
1316 |
+
"grad_norm": 1.447130560874939,
|
1317 |
+
"learning_rate": 1.0022222222222223e-06,
|
1318 |
+
"loss": 0.0243,
|
1319 |
"step": 4550
|
1320 |
},
|
1321 |
{
|
1322 |
"epoch": 5.210706150341686,
|
1323 |
+
"grad_norm": 1.1469416618347168,
|
1324 |
+
"learning_rate": 9.466666666666667e-07,
|
1325 |
+
"loss": 0.0276,
|
1326 |
"step": 4575
|
1327 |
},
|
1328 |
{
|
1329 |
"epoch": 5.239179954441913,
|
1330 |
+
"grad_norm": 1.7906535863876343,
|
1331 |
+
"learning_rate": 8.911111111111112e-07,
|
1332 |
+
"loss": 0.026,
|
1333 |
"step": 4600
|
1334 |
},
|
1335 |
{
|
1336 |
"epoch": 5.267653758542141,
|
1337 |
+
"grad_norm": 1.0731186866760254,
|
1338 |
+
"learning_rate": 8.355555555555556e-07,
|
1339 |
+
"loss": 0.0253,
|
1340 |
"step": 4625
|
1341 |
},
|
1342 |
{
|
1343 |
"epoch": 5.296127562642369,
|
1344 |
+
"grad_norm": 2.2479116916656494,
|
1345 |
+
"learning_rate": 7.8e-07,
|
1346 |
+
"loss": 0.0261,
|
1347 |
"step": 4650
|
1348 |
},
|
1349 |
{
|
1350 |
"epoch": 5.324601366742597,
|
1351 |
+
"grad_norm": 1.143470287322998,
|
1352 |
+
"learning_rate": 7.244444444444446e-07,
|
1353 |
+
"loss": 0.0249,
|
1354 |
"step": 4675
|
1355 |
},
|
1356 |
{
|
1357 |
"epoch": 5.353075170842825,
|
1358 |
+
"grad_norm": 1.1293048858642578,
|
1359 |
+
"learning_rate": 6.68888888888889e-07,
|
1360 |
+
"loss": 0.0236,
|
1361 |
"step": 4700
|
1362 |
},
|
1363 |
{
|
1364 |
"epoch": 5.381548974943052,
|
1365 |
+
"grad_norm": 1.7358245849609375,
|
1366 |
+
"learning_rate": 6.133333333333333e-07,
|
1367 |
+
"loss": 0.023,
|
1368 |
"step": 4725
|
1369 |
},
|
1370 |
{
|
1371 |
"epoch": 5.41002277904328,
|
1372 |
+
"grad_norm": 1.031139612197876,
|
1373 |
+
"learning_rate": 5.577777777777779e-07,
|
1374 |
+
"loss": 0.0234,
|
1375 |
"step": 4750
|
1376 |
},
|
1377 |
{
|
1378 |
"epoch": 5.438496583143508,
|
1379 |
+
"grad_norm": 1.8313310146331787,
|
1380 |
+
"learning_rate": 5.022222222222222e-07,
|
1381 |
+
"loss": 0.0261,
|
1382 |
"step": 4775
|
1383 |
},
|
1384 |
{
|
1385 |
"epoch": 5.466970387243736,
|
1386 |
+
"grad_norm": 1.6976441144943237,
|
1387 |
+
"learning_rate": 4.466666666666667e-07,
|
1388 |
+
"loss": 0.0237,
|
1389 |
"step": 4800
|
1390 |
},
|
1391 |
{
|
1392 |
"epoch": 5.495444191343964,
|
1393 |
+
"grad_norm": 1.3694329261779785,
|
1394 |
+
"learning_rate": 3.9111111111111115e-07,
|
1395 |
+
"loss": 0.0257,
|
1396 |
"step": 4825
|
1397 |
},
|
1398 |
{
|
1399 |
"epoch": 5.523917995444191,
|
1400 |
+
"grad_norm": 1.0080304145812988,
|
1401 |
+
"learning_rate": 3.3555555555555556e-07,
|
1402 |
+
"loss": 0.0246,
|
1403 |
"step": 4850
|
1404 |
},
|
1405 |
{
|
1406 |
"epoch": 5.552391799544419,
|
1407 |
+
"grad_norm": 0.9774746298789978,
|
1408 |
+
"learning_rate": 2.8e-07,
|
1409 |
+
"loss": 0.0254,
|
1410 |
"step": 4875
|
1411 |
},
|
1412 |
{
|
1413 |
"epoch": 5.5808656036446465,
|
1414 |
+
"grad_norm": 1.1044853925704956,
|
1415 |
+
"learning_rate": 2.2444444444444445e-07,
|
1416 |
+
"loss": 0.026,
|
1417 |
"step": 4900
|
1418 |
},
|
1419 |
{
|
1420 |
"epoch": 5.609339407744875,
|
1421 |
+
"grad_norm": 1.2919448614120483,
|
1422 |
+
"learning_rate": 1.6888888888888888e-07,
|
1423 |
+
"loss": 0.0236,
|
1424 |
"step": 4925
|
1425 |
},
|
1426 |
{
|
1427 |
"epoch": 5.637813211845103,
|
1428 |
+
"grad_norm": 1.2227028608322144,
|
1429 |
+
"learning_rate": 1.1333333333333336e-07,
|
1430 |
+
"loss": 0.0273,
|
1431 |
"step": 4950
|
1432 |
},
|
1433 |
{
|
1434 |
"epoch": 5.66628701594533,
|
1435 |
+
"grad_norm": 0.887917697429657,
|
1436 |
+
"learning_rate": 5.777777777777778e-08,
|
1437 |
+
"loss": 0.0249,
|
1438 |
"step": 4975
|
1439 |
},
|
1440 |
{
|
1441 |
"epoch": 5.694760820045558,
|
1442 |
+
"grad_norm": 0.8368929624557495,
|
1443 |
+
"learning_rate": 2.2222222222222225e-09,
|
1444 |
+
"loss": 0.0245,
|
1445 |
"step": 5000
|
1446 |
},
|
1447 |
{
|
1448 |
"epoch": 5.694760820045558,
|
1449 |
+
"eval_loss": 0.5775743126869202,
|
1450 |
+
"eval_runtime": 2967.3924,
|
1451 |
+
"eval_samples_per_second": 2.366,
|
1452 |
+
"eval_steps_per_second": 0.148,
|
1453 |
+
"eval_wer": 0.3064475414845421,
|
1454 |
"step": 5000
|
1455 |
},
|
1456 |
{
|
1457 |
"epoch": 5.694760820045558,
|
1458 |
"step": 5000,
|
1459 |
"total_flos": 5.43429854134272e+20,
|
1460 |
+
"train_loss": 0.25138797866106033,
|
1461 |
+
"train_runtime": 67213.2375,
|
1462 |
+
"train_samples_per_second": 2.38,
|
1463 |
+
"train_steps_per_second": 0.074
|
1464 |
}
|
1465 |
],
|
1466 |
"logging_steps": 25,
|