Hugging Face's logo

MelanieKoe
/
w2v2-base-pretrained_lr1e-4_at0.8_da0.2

Automatic Speech Recognition
Transformers
Safetensors
wav2vec2
Generated from Trainer
Inference Endpoints
Model card Files Community
MelanieKoe commited on
Commit
290cd78
•
1 Parent(s): d192532

Model save

Browse files
Files changed (2) hide show
  1. README.md +18 -24
  2. model.safetensors +1 -1
README.md CHANGED
@@ -17,8 +17,8 @@ should probably proofread and complete it, then remove this comment. -->
17
 
18
  This model is a fine-tuned version of [facebook/wav2vec2-base](https://huggingface.co/facebook/wav2vec2-base) on the None dataset.
19
  It achieves the following results on the evaluation set:
20
- - Loss: 1.3297
21
- - Wer: 0.2683
22
 
23
  ## Model description
24
 
@@ -43,34 +43,28 @@ The following hyperparameters were used during training:
43
  - seed: 42
44
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
45
  - lr_scheduler_type: linear
46
- - lr_scheduler_warmup_steps: 100
47
- - num_epochs: 200
48
  - mixed_precision_training: Native AMP
49
 
50
  ### Training results
51
 
52
  | Training Loss | Epoch | Step | Validation Loss | Wer |
53
  |:-------------:|:-----:|:----:|:---------------:|:------:|
54
- | 23.6917 | 10.0 | 100 | 3.6026 | 1.0 |
55
- | 3.259 | 20.0 | 200 | 3.1696 | 1.0 |
56
- | 3.1152 | 30.0 | 300 | 3.1344 | 1.0 |
57
- | 3.0799 | 40.0 | 400 | 3.0976 | 1.0 |
58
- | 3.0793 | 50.0 | 500 | 3.0977 | 1.0 |
59
- | 3.0692 | 60.0 | 600 | 3.0992 | 1.0 |
60
- | 3.0604 | 70.0 | 700 | 3.1350 | 1.0 |
61
- | 3.0397 | 80.0 | 800 | 3.0537 | 1.0 |
62
- | 2.9847 | 90.0 | 900 | 2.9905 | 1.0 |
63
- | 2.5926 | 100.0 | 1000 | 2.2350 | 1.0077 |
64
- | 0.9017 | 110.0 | 1100 | 1.2152 | 0.6301 |
65
- | 0.2326 | 120.0 | 1200 | 1.2279 | 0.4524 |
66
- | 0.1364 | 130.0 | 1300 | 1.2103 | 0.4238 |
67
- | 0.0935 | 140.0 | 1400 | 1.1953 | 0.3926 |
68
- | 0.0759 | 150.0 | 1500 | 1.3237 | 0.3516 |
69
- | 0.0599 | 160.0 | 1600 | 1.3929 | 0.3050 |
70
- | 0.0438 | 170.0 | 1700 | 1.3132 | 0.2717 |
71
- | 0.0389 | 180.0 | 1800 | 1.3469 | 0.2666 |
72
- | 0.0355 | 190.0 | 1900 | 1.3029 | 0.2691 |
73
- | 0.03 | 200.0 | 2000 | 1.3297 | 0.2683 |
74
 
75
 
76
  ### Framework versions
 
17
 
18
  This model is a fine-tuned version of [facebook/wav2vec2-base](https://huggingface.co/facebook/wav2vec2-base) on the None dataset.
19
  It achieves the following results on the evaluation set:
20
+ - Loss: 2.2354
21
+ - Wer: 0.1781
22
 
23
  ## Model description
24
 
 
43
  - seed: 42
44
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
45
  - lr_scheduler_type: linear
46
+ - lr_scheduler_warmup_steps: 1000
47
+ - training_steps: 3500
48
  - mixed_precision_training: Native AMP
49
 
50
  ### Training results
51
 
52
  | Training Loss | Epoch | Step | Validation Loss | Wer |
53
  |:-------------:|:-----:|:----:|:---------------:|:------:|
54
+ | 18.7165 | 25.0 | 250 | 3.6383 | 1.0 |
55
+ | 3.2049 | 50.0 | 500 | 3.1037 | 1.0 |
56
+ | 2.5797 | 75.0 | 750 | 1.3392 | 0.9987 |
57
+ | 0.2099 | 100.0 | 1000 | 1.5991 | 0.2947 |
58
+ | 0.0635 | 125.0 | 1250 | 1.7346 | 0.2119 |
59
+ | 0.0354 | 150.0 | 1500 | 2.0431 | 0.1939 |
60
+ | 0.0184 | 175.0 | 1750 | 1.9597 | 0.1935 |
61
+ | 0.0129 | 200.0 | 2000 | 2.2028 | 0.1991 |
62
+ | 0.0101 | 225.0 | 2250 | 1.9856 | 0.1820 |
63
+ | 0.0061 | 250.0 | 2500 | 1.9899 | 0.1897 |
64
+ | 0.0064 | 275.0 | 2750 | 2.1246 | 0.1811 |
65
+ | 0.0037 | 300.0 | 3000 | 2.2400 | 0.1803 |
66
+ | 0.0033 | 325.0 | 3250 | 2.1766 | 0.1786 |
67
+ | 0.0029 | 350.0 | 3500 | 2.2354 | 0.1781 |
 
 
 
 
 
 
68
 
69
 
70
  ### Framework versions
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:e13d1612b75cc6586cd0f433bf7f138ef260733306dae13bfa608f98883c7339
3
  size 377629528
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fa6a2ba23790452c4fc308c3cefb22a1cb184a94b56d52923a8457ac4c486d39
3
  size 377629528