Hugging Face's logo

MelanieKoe
/
w2v2-base-pretrained_lr5e-5_at0.8_da0.3

Automatic Speech Recognition
Transformers
Safetensors
wav2vec2
Generated from Trainer
Inference Endpoints
Model card Files Community
MelanieKoe commited on
Commit
b7835ca
•
1 Parent(s): d2bbe7b

Model save

Browse files
Files changed (2) hide show
  1. README.md +19 -19
  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: 1.9317
21
- - Wer: 0.3140
22
 
23
  ## Model description
24
 
@@ -43,7 +43,7 @@ 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: 500
47
  - training_steps: 4000
48
  - mixed_precision_training: Native AMP
49
 
@@ -51,22 +51,22 @@ The following hyperparameters were used during training:
51
 
52
  | Training Loss | Epoch | Step | Validation Loss | Wer |
53
  |:-------------:|:------:|:----:|:---------------:|:------:|
54
- | 17.666 | 17.86 | 250 | 3.5715 | 1.0 |
55
- | 3.1402 | 35.71 | 500 | 3.0649 | 1.0 |
56
- | 2.5877 | 53.57 | 750 | 1.9033 | 1.0 |
57
- | 0.5748 | 71.43 | 1000 | 1.3133 | 0.4827 |
58
- | 0.1178 | 89.29 | 1250 | 1.4879 | 0.3943 |
59
- | 0.0598 | 107.14 | 1500 | 1.6656 | 0.3810 |
60
- | 0.0423 | 125.0 | 1750 | 1.6371 | 0.3708 |
61
- | 0.0295 | 142.86 | 2000 | 1.8732 | 0.3635 |
62
- | 0.0228 | 160.71 | 2250 | 1.7997 | 0.3366 |
63
- | 0.018 | 178.57 | 2500 | 1.8825 | 0.3276 |
64
- | 0.014 | 196.43 | 2750 | 1.9048 | 0.3268 |
65
- | 0.0109 | 214.29 | 3000 | 1.9426 | 0.3101 |
66
- | 0.0093 | 232.14 | 3250 | 1.9842 | 0.3071 |
67
- | 0.008 | 250.0 | 3500 | 1.8719 | 0.3204 |
68
- | 0.0068 | 267.86 | 3750 | 1.9318 | 0.3063 |
69
- | 0.006 | 285.71 | 4000 | 1.9317 | 0.3140 |
70
 
71
 
72
  ### 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: 5.8781
21
+ - Wer: 0.7877
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: 4000
48
  - mixed_precision_training: Native AMP
49
 
 
51
 
52
  | Training Loss | Epoch | Step | Validation Loss | Wer |
53
  |:-------------:|:------:|:----:|:---------------:|:------:|
54
+ | 24.822 | 17.86 | 250 | 8.9601 | 1.0 |
55
+ | 6.2153 | 35.71 | 500 | 5.9697 | 1.0 |
56
+ | 3.5607 | 53.57 | 750 | 5.5112 | 1.0 |
57
+ | 3.2305 | 71.43 | 1000 | 5.6378 | 1.0 |
58
+ | 2.9631 | 89.29 | 1250 | 5.4385 | 1.0 |
59
+ | 2.5463 | 107.14 | 1500 | 5.2806 | 0.9953 |
60
+ | 2.1207 | 125.0 | 1750 | 5.3065 | 0.9492 |
61
+ | 1.7629 | 142.86 | 2000 | 5.4365 | 0.9158 |
62
+ | 1.4631 | 160.71 | 2250 | 5.1678 | 0.8667 |
63
+ | 1.2158 | 178.57 | 2500 | 5.3097 | 0.8428 |
64
+ | 1.0507 | 196.43 | 2750 | 5.6917 | 0.8279 |
65
+ | 0.9326 | 214.29 | 3000 | 5.7407 | 0.8197 |
66
+ | 0.8245 | 232.14 | 3250 | 5.5588 | 0.8039 |
67
+ | 0.7415 | 250.0 | 3500 | 5.7107 | 0.7860 |
68
+ | 0.694 | 267.86 | 3750 | 5.8551 | 0.7971 |
69
+ | 0.6634 | 285.71 | 4000 | 5.8781 | 0.7877 |
70
 
71
 
72
  ### Framework versions
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:55c84245d88b2a681cc8e0d83f47a8755509c0756a951983a57ea64c81385c29
3
  size 377629528
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7250fe476903c5ada0e5c3a069221948722bfddc2aae55db643299360df38d0d
3
  size 377629528