Giseok commited on
Commit
3a8337c
·
1 Parent(s): 92fee6a

update model card README.md

Browse files
Files changed (1) hide show
  1. README.md +32 -32
README.md CHANGED
@@ -12,10 +12,10 @@ should probably proofread and complete it, then remove this comment. -->
12
 
13
  # wav2vec2-base-STTTest
14
 
15
- This model is a fine-tuned version of [facebook/wav2vec2-base-960h](https://huggingface.co/facebook/wav2vec2-base-960h) on the None dataset.
16
  It achieves the following results on the evaluation set:
17
- - Loss: 0.2188
18
- - Wer: 0.2310
19
 
20
  ## Model description
21
 
@@ -48,35 +48,35 @@ The following hyperparameters were used during training:
48
 
49
  | Training Loss | Epoch | Step | Validation Loss | Wer |
50
  |:-------------:|:-----:|:-----:|:---------------:|:------:|
51
- | 5.6272 | 1.0 | 500 | 4.1587 | 1.0 |
52
- | 2.6402 | 2.01 | 1000 | 0.3609 | 0.3751 |
53
- | 0.6464 | 3.01 | 1500 | 0.2678 | 0.2673 |
54
- | 0.5417 | 4.02 | 2000 | 0.2225 | 0.2591 |
55
- | 0.3863 | 5.02 | 2500 | 0.2243 | 0.2735 |
56
- | 0.3231 | 6.02 | 3000 | 0.2294 | 0.2626 |
57
- | 0.3165 | 7.03 | 3500 | 0.2287 | 0.2594 |
58
- | 0.2057 | 8.03 | 4000 | 0.2325 | 0.2582 |
59
- | 0.1966 | 9.04 | 4500 | 0.2180 | 0.2608 |
60
- | 0.1604 | 10.04 | 5000 | 0.2000 | 0.2458 |
61
- | 0.1923 | 11.04 | 5500 | 0.2375 | 0.2573 |
62
- | 0.1657 | 12.05 | 6000 | 0.2533 | 0.2696 |
63
- | 0.1621 | 13.05 | 6500 | 0.2522 | 0.2529 |
64
- | 0.1285 | 14.06 | 7000 | 0.2275 | 0.2573 |
65
- | 0.1254 | 15.06 | 7500 | 0.2319 | 0.2482 |
66
- | 0.1074 | 16.06 | 8000 | 0.2331 | 0.2465 |
67
- | 0.147 | 17.07 | 8500 | 0.2290 | 0.2470 |
68
- | 0.1028 | 18.07 | 9000 | 0.2315 | 0.2507 |
69
- | 0.0916 | 19.08 | 9500 | 0.2456 | 0.2496 |
70
- | 0.0918 | 20.08 | 10000 | 0.2361 | 0.2467 |
71
- | 0.0889 | 21.08 | 10500 | 0.2291 | 0.2422 |
72
- | 0.0721 | 22.09 | 11000 | 0.2452 | 0.2425 |
73
- | 0.0767 | 23.09 | 11500 | 0.2366 | 0.2403 |
74
- | 0.0779 | 24.1 | 12000 | 0.2310 | 0.2400 |
75
- | 0.06 | 25.1 | 12500 | 0.2213 | 0.2344 |
76
- | 0.0657 | 26.1 | 13000 | 0.2211 | 0.2322 |
77
- | 0.0644 | 27.11 | 13500 | 0.2228 | 0.2316 |
78
- | 0.0758 | 28.11 | 14000 | 0.2206 | 0.2325 |
79
- | 0.0624 | 29.12 | 14500 | 0.2188 | 0.2310 |
80
 
81
 
82
  ### Framework versions
 
12
 
13
  # wav2vec2-base-STTTest
14
 
15
+ This model is a fine-tuned version of [facebook/wav2vec2-base](https://huggingface.co/facebook/wav2vec2-base) on the None dataset.
16
  It achieves the following results on the evaluation set:
17
+ - Loss: 0.5198
18
+ - Wer: 0.3393
19
 
20
  ## Model description
21
 
 
48
 
49
  | Training Loss | Epoch | Step | Validation Loss | Wer |
50
  |:-------------:|:-----:|:-----:|:---------------:|:------:|
51
+ | 0.231 | 1.0 | 500 | 0.4337 | 0.4100 |
52
+ | 0.1845 | 2.01 | 1000 | 0.4296 | 0.3931 |
53
+ | 0.1551 | 3.01 | 1500 | 0.4397 | 0.3770 |
54
+ | 0.1479 | 4.02 | 2000 | 0.4524 | 0.3827 |
55
+ | 0.1186 | 5.02 | 2500 | 0.5182 | 0.3795 |
56
+ | 0.1079 | 6.02 | 3000 | 0.4799 | 0.3737 |
57
+ | 0.0974 | 7.03 | 3500 | 0.4966 | 0.3860 |
58
+ | 0.0878 | 8.03 | 4000 | 0.4993 | 0.3699 |
59
+ | 0.0788 | 9.04 | 4500 | 0.5183 | 0.3678 |
60
+ | 0.0732 | 10.04 | 5000 | 0.5064 | 0.3635 |
61
+ | 0.0664 | 11.04 | 5500 | 0.5330 | 0.3663 |
62
+ | 0.0596 | 12.05 | 6000 | 0.5147 | 0.3516 |
63
+ | 0.0538 | 13.05 | 6500 | 0.5254 | 0.3581 |
64
+ | 0.0535 | 14.06 | 7000 | 0.4902 | 0.3534 |
65
+ | 0.0492 | 15.06 | 7500 | 0.5115 | 0.3488 |
66
+ | 0.0455 | 16.06 | 8000 | 0.5250 | 0.3472 |
67
+ | 0.0434 | 17.07 | 8500 | 0.5338 | 0.3515 |
68
+ | 0.0351 | 18.07 | 9000 | 0.5365 | 0.3444 |
69
+ | 0.0341 | 19.08 | 9500 | 0.4886 | 0.3439 |
70
+ | 0.0332 | 20.08 | 10000 | 0.5234 | 0.3475 |
71
+ | 0.0289 | 21.08 | 10500 | 0.5375 | 0.3464 |
72
+ | 0.028 | 22.09 | 11000 | 0.5395 | 0.3478 |
73
+ | 0.0225 | 23.09 | 11500 | 0.5236 | 0.3428 |
74
+ | 0.0244 | 24.1 | 12000 | 0.5122 | 0.3402 |
75
+ | 0.0246 | 25.1 | 12500 | 0.5212 | 0.3390 |
76
+ | 0.0214 | 26.1 | 13000 | 0.5198 | 0.3393 |
77
+ | 0.0179 | 27.11 | 13500 | 0.5198 | 0.3393 |
78
+ | 0.0194 | 28.11 | 14000 | 0.5198 | 0.3393 |
79
+ | 0.0193 | 29.12 | 14500 | 0.5198 | 0.3393 |
80
 
81
 
82
  ### Framework versions