update model card README.md
Browse files
README.md
CHANGED
@@ -14,8 +14,8 @@ should probably proofread and complete it, then remove this comment. -->
|
|
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.
|
18 |
-
- Wer: 0.
|
19 |
|
20 |
## Model description
|
21 |
|
@@ -34,7 +34,7 @@ More information needed
|
|
34 |
### Training hyperparameters
|
35 |
|
36 |
The following hyperparameters were used during training:
|
37 |
-
- learning_rate:
|
38 |
- train_batch_size: 16
|
39 |
- eval_batch_size: 8
|
40 |
- seed: 42
|
@@ -43,18 +43,22 @@ The following hyperparameters were used during training:
|
|
43 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
44 |
- lr_scheduler_type: linear
|
45 |
- lr_scheduler_warmup_steps: 500
|
46 |
-
- num_epochs:
|
47 |
- mixed_precision_training: Native AMP
|
48 |
|
49 |
### Training results
|
50 |
|
51 |
| Training Loss | Epoch | Step | Validation Loss | Wer |
|
52 |
|:-------------:|:-----:|:----:|:---------------:|:------:|
|
53 |
-
|
|
54 |
-
|
|
55 |
-
|
|
56 |
-
|
|
57 |
-
| 0.
|
|
|
|
|
|
|
|
|
58 |
|
59 |
|
60 |
### Framework versions
|
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: 0.6376
|
18 |
+
- Wer: 0.2137
|
19 |
|
20 |
## Model description
|
21 |
|
34 |
### Training hyperparameters
|
35 |
|
36 |
The following hyperparameters were used during training:
|
37 |
+
- learning_rate: 0.0005
|
38 |
- train_batch_size: 16
|
39 |
- eval_batch_size: 8
|
40 |
- seed: 42
|
43 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
44 |
- lr_scheduler_type: linear
|
45 |
- lr_scheduler_warmup_steps: 500
|
46 |
+
- num_epochs: 50
|
47 |
- mixed_precision_training: Native AMP
|
48 |
|
49 |
### Training results
|
50 |
|
51 |
| Training Loss | Epoch | Step | Validation Loss | Wer |
|
52 |
|:-------------:|:-----:|:----:|:---------------:|:------:|
|
53 |
+
| 0.7233 | 5.26 | 400 | 5.6115 | 0.4909 |
|
54 |
+
| 0.7224 | 10.52 | 800 | 1.1835 | 0.3629 |
|
55 |
+
| 0.5193 | 15.78 | 1200 | 0.5786 | 0.2994 |
|
56 |
+
| 0.4114 | 21.05 | 1600 | 0.6535 | 0.2823 |
|
57 |
+
| 0.3279 | 26.31 | 2000 | 0.6783 | 0.2709 |
|
58 |
+
| 0.2711 | 31.58 | 2400 | 0.6570 | 0.2490 |
|
59 |
+
| 0.2776 | 36.84 | 2800 | 0.6724 | 0.2358 |
|
60 |
+
| 0.1805 | 42.1 | 3200 | 0.6212 | 0.2241 |
|
61 |
+
| 0.1535 | 47.37 | 3600 | 0.6376 | 0.2137 |
|
62 |
|
63 |
|
64 |
### Framework versions
|