Commit
•
5363dfa
1
Parent(s):
d88b184
update model card README.md
Browse files
README.md
CHANGED
|
@@ -13,8 +13,8 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 13 |
|
| 14 |
This model is a fine-tuned version of [Harveenchadha/vakyansh-wav2vec2-bengali-bnm-200](https://huggingface.co/Harveenchadha/vakyansh-wav2vec2-bengali-bnm-200) on the None dataset.
|
| 15 |
It achieves the following results on the evaluation set:
|
| 16 |
-
- Loss:
|
| 17 |
-
- Wer: 0.
|
| 18 |
|
| 19 |
## Model description
|
| 20 |
|
|
@@ -34,20 +34,23 @@ More information needed
|
|
| 34 |
|
| 35 |
The following hyperparameters were used during training:
|
| 36 |
- learning_rate: 0.0001
|
| 37 |
-
- train_batch_size:
|
| 38 |
- eval_batch_size: 8
|
| 39 |
- seed: 42
|
| 40 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 41 |
- lr_scheduler_type: linear
|
| 42 |
- lr_scheduler_warmup_steps: 1000
|
| 43 |
-
- num_epochs:
|
| 44 |
- mixed_precision_training: Native AMP
|
| 45 |
|
| 46 |
### Training results
|
| 47 |
|
| 48 |
| Training Loss | Epoch | Step | Validation Loss | Wer |
|
| 49 |
|:-------------:|:-----:|:----:|:---------------:|:------:|
|
| 50 |
-
|
|
|
|
|
|
|
|
|
|
|
| 51 |
|
| 52 |
|
| 53 |
### Framework versions
|
|
|
|
| 13 |
|
| 14 |
This model is a fine-tuned version of [Harveenchadha/vakyansh-wav2vec2-bengali-bnm-200](https://huggingface.co/Harveenchadha/vakyansh-wav2vec2-bengali-bnm-200) on the None dataset.
|
| 15 |
It achieves the following results on the evaluation set:
|
| 16 |
+
- Loss: 151.2085
|
| 17 |
+
- Wer: 0.4699
|
| 18 |
|
| 19 |
## Model description
|
| 20 |
|
|
|
|
| 34 |
|
| 35 |
The following hyperparameters were used during training:
|
| 36 |
- learning_rate: 0.0001
|
| 37 |
+
- train_batch_size: 16
|
| 38 |
- eval_batch_size: 8
|
| 39 |
- seed: 42
|
| 40 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 41 |
- lr_scheduler_type: linear
|
| 42 |
- lr_scheduler_warmup_steps: 1000
|
| 43 |
+
- num_epochs: 35
|
| 44 |
- mixed_precision_training: Native AMP
|
| 45 |
|
| 46 |
### Training results
|
| 47 |
|
| 48 |
| Training Loss | Epoch | Step | Validation Loss | Wer |
|
| 49 |
|:-------------:|:-----:|:----:|:---------------:|:------:|
|
| 50 |
+
| 2598.842 | 7.46 | 500 | 241.7446 | 0.6716 |
|
| 51 |
+
| 421.6865 | 14.93 | 1000 | 157.1855 | 0.5365 |
|
| 52 |
+
| 251.789 | 22.39 | 1500 | 153.7565 | 0.5045 |
|
| 53 |
+
| 160.1666 | 29.85 | 2000 | 151.2085 | 0.4699 |
|
| 54 |
|
| 55 |
|
| 56 |
### Framework versions
|