update model card README.md
Browse files
README.md
CHANGED
@@ -16,8 +16,8 @@ should probably proofread and complete it, then remove this comment. -->
|
|
16 |
|
17 |
This model is a fine-tuned version of [facebook/wav2vec2-base-100k-voxpopuli](https://huggingface.co/facebook/wav2vec2-base-100k-voxpopuli) on the None dataset.
|
18 |
It achieves the following results on the evaluation set:
|
19 |
-
- Loss:
|
20 |
-
- Accuracy: 0.
|
21 |
|
22 |
## Model description
|
23 |
|
@@ -36,27 +36,42 @@ More information needed
|
|
36 |
### Training hyperparameters
|
37 |
|
38 |
The following hyperparameters were used during training:
|
39 |
-
- learning_rate:
|
40 |
- train_batch_size: 2
|
41 |
- eval_batch_size: 2
|
42 |
- seed: 42
|
43 |
-
- gradient_accumulation_steps:
|
44 |
-
- total_train_batch_size:
|
45 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
46 |
- lr_scheduler_type: linear
|
47 |
- lr_scheduler_warmup_ratio: 0.1
|
48 |
-
- num_epochs:
|
49 |
- mixed_precision_training: Native AMP
|
50 |
|
51 |
### Training results
|
52 |
|
53 |
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|
54 |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
|
55 |
-
| No log | 1.0 |
|
56 |
-
|
|
57 |
-
|
|
58 |
-
|
|
59 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
60 |
|
61 |
|
62 |
### Framework versions
|
|
|
16 |
|
17 |
This model is a fine-tuned version of [facebook/wav2vec2-base-100k-voxpopuli](https://huggingface.co/facebook/wav2vec2-base-100k-voxpopuli) on the None dataset.
|
18 |
It achieves the following results on the evaluation set:
|
19 |
+
- Loss: 0.9408
|
20 |
+
- Accuracy: 0.86
|
21 |
|
22 |
## Model description
|
23 |
|
|
|
36 |
### Training hyperparameters
|
37 |
|
38 |
The following hyperparameters were used during training:
|
39 |
+
- learning_rate: 5e-05
|
40 |
- train_batch_size: 2
|
41 |
- eval_batch_size: 2
|
42 |
- seed: 42
|
43 |
+
- gradient_accumulation_steps: 2
|
44 |
+
- total_train_batch_size: 4
|
45 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
46 |
- lr_scheduler_type: linear
|
47 |
- lr_scheduler_warmup_ratio: 0.1
|
48 |
+
- num_epochs: 20
|
49 |
- mixed_precision_training: Native AMP
|
50 |
|
51 |
### Training results
|
52 |
|
53 |
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|
54 |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
|
55 |
+
| No log | 1.0 | 225 | 2.1672 | 0.3 |
|
56 |
+
| 2.1675 | 2.0 | 450 | 2.0095 | 0.29 |
|
57 |
+
| 2.1675 | 3.0 | 675 | 1.7326 | 0.29 |
|
58 |
+
| 1.7199 | 4.0 | 900 | 1.4980 | 0.49 |
|
59 |
+
| 1.7199 | 5.0 | 1125 | 1.4088 | 0.37 |
|
60 |
+
| 1.3585 | 6.0 | 1350 | 1.2238 | 0.54 |
|
61 |
+
| 1.3585 | 7.0 | 1575 | 1.3579 | 0.52 |
|
62 |
+
| 1.0599 | 8.0 | 1800 | 0.9954 | 0.62 |
|
63 |
+
| 1.0599 | 9.0 | 2025 | 0.9543 | 0.73 |
|
64 |
+
| 0.8337 | 10.0 | 2250 | 0.9428 | 0.76 |
|
65 |
+
| 0.8337 | 11.0 | 2475 | 0.8810 | 0.78 |
|
66 |
+
| 0.5861 | 12.0 | 2700 | 0.7753 | 0.76 |
|
67 |
+
| 0.5861 | 13.0 | 2925 | 0.9981 | 0.74 |
|
68 |
+
| 0.3662 | 14.0 | 3150 | 1.1597 | 0.77 |
|
69 |
+
| 0.3662 | 15.0 | 3375 | 1.0466 | 0.79 |
|
70 |
+
| 0.277 | 16.0 | 3600 | 1.0763 | 0.81 |
|
71 |
+
| 0.277 | 17.0 | 3825 | 0.8407 | 0.87 |
|
72 |
+
| 0.1731 | 18.0 | 4050 | 0.9317 | 0.86 |
|
73 |
+
| 0.1731 | 19.0 | 4275 | 0.8545 | 0.87 |
|
74 |
+
| 0.1489 | 20.0 | 4500 | 0.9408 | 0.86 |
|
75 |
|
76 |
|
77 |
### Framework versions
|