Commit
•
a4180c5
1
Parent(s):
d29dc73
update model card README.md
Browse files
README.md
CHANGED
|
@@ -1,10 +1,6 @@
|
|
| 1 |
---
|
| 2 |
-
language:
|
| 3 |
-
- ar
|
| 4 |
license: apache-2.0
|
| 5 |
tags:
|
| 6 |
-
- automatic-speech-recognition
|
| 7 |
-
- common_voice
|
| 8 |
- generated_from_trainer
|
| 9 |
datasets:
|
| 10 |
- common_voice
|
|
@@ -18,10 +14,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 18 |
|
| 19 |
# wav2vec2-xls-r-300m-ar
|
| 20 |
|
| 21 |
-
This model is a fine-tuned version of [facebook/wav2vec2-xls-r-300m](https://huggingface.co/facebook/wav2vec2-xls-r-300m) on the
|
| 22 |
-
It achieves the following results on the evaluation set:
|
| 23 |
-
- Loss: 6.9375
|
| 24 |
-
- Wer: 1.0
|
| 25 |
|
| 26 |
## Model description
|
| 27 |
|
|
@@ -40,41 +33,16 @@ More information needed
|
|
| 40 |
### Training hyperparameters
|
| 41 |
|
| 42 |
The following hyperparameters were used during training:
|
| 43 |
-
- learning_rate:
|
| 44 |
- train_batch_size: 64
|
| 45 |
- eval_batch_size: 64
|
| 46 |
- seed: 42
|
| 47 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 48 |
- lr_scheduler_type: linear
|
| 49 |
-
-
|
|
|
|
| 50 |
- mixed_precision_training: Native AMP
|
| 51 |
|
| 52 |
-
### Training results
|
| 53 |
-
|
| 54 |
-
| Training Loss | Epoch | Step | Validation Loss | Wer |
|
| 55 |
-
|:-------------:|:-----:|:----:|:---------------:|:---:|
|
| 56 |
-
| 19.0933 | 1.0 | 2 | 32.5547 | 1.0 |
|
| 57 |
-
| 18.6757 | 2.0 | 4 | 31.0408 | 1.0 |
|
| 58 |
-
| 17.1649 | 3.0 | 6 | 29.5776 | 1.0 |
|
| 59 |
-
| 14.7415 | 4.0 | 8 | 22.9183 | 1.0 |
|
| 60 |
-
| 11.8071 | 5.0 | 10 | 17.5073 | 1.0 |
|
| 61 |
-
| 9.7675 | 6.0 | 12 | 14.1743 | 1.0 |
|
| 62 |
-
| 8.4193 | 7.0 | 14 | 12.1225 | 1.0 |
|
| 63 |
-
| 7.4746 | 8.0 | 16 | 10.7446 | 1.0 |
|
| 64 |
-
| 6.8442 | 9.0 | 18 | 9.7949 | 1.0 |
|
| 65 |
-
| 6.3765 | 10.0 | 20 | 9.1148 | 1.0 |
|
| 66 |
-
| 6.0321 | 11.0 | 22 | 8.5934 | 1.0 |
|
| 67 |
-
| 5.7783 | 12.0 | 24 | 8.1983 | 1.0 |
|
| 68 |
-
| 5.5827 | 13.0 | 26 | 7.8780 | 1.0 |
|
| 69 |
-
| 5.4249 | 14.0 | 28 | 7.6288 | 1.0 |
|
| 70 |
-
| 5.3088 | 15.0 | 30 | 7.4279 | 1.0 |
|
| 71 |
-
| 5.2078 | 16.0 | 32 | 7.2688 | 1.0 |
|
| 72 |
-
| 5.1289 | 17.0 | 34 | 7.1459 | 1.0 |
|
| 73 |
-
| 5.0697 | 18.0 | 36 | 7.0528 | 1.0 |
|
| 74 |
-
| 5.0227 | 19.0 | 38 | 6.9835 | 1.0 |
|
| 75 |
-
| 4.9853 | 20.0 | 40 | 6.9375 | 1.0 |
|
| 76 |
-
|
| 77 |
-
|
| 78 |
### Framework versions
|
| 79 |
|
| 80 |
- Transformers 4.17.0.dev0
|
|
|
|
| 1 |
---
|
|
|
|
|
|
|
| 2 |
license: apache-2.0
|
| 3 |
tags:
|
|
|
|
|
|
|
| 4 |
- generated_from_trainer
|
| 5 |
datasets:
|
| 6 |
- common_voice
|
|
|
|
| 14 |
|
| 15 |
# wav2vec2-xls-r-300m-ar
|
| 16 |
|
| 17 |
+
This model is a fine-tuned version of [facebook/wav2vec2-xls-r-300m](https://huggingface.co/facebook/wav2vec2-xls-r-300m) on the common_voice dataset.
|
|
|
|
|
|
|
|
|
|
| 18 |
|
| 19 |
## Model description
|
| 20 |
|
|
|
|
| 33 |
### Training hyperparameters
|
| 34 |
|
| 35 |
The following hyperparameters were used during training:
|
| 36 |
+
- learning_rate: 0.0005
|
| 37 |
- train_batch_size: 64
|
| 38 |
- eval_batch_size: 64
|
| 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: 2000
|
| 43 |
+
- num_epochs: 5
|
| 44 |
- mixed_precision_training: Native AMP
|
| 45 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 46 |
### Framework versions
|
| 47 |
|
| 48 |
- Transformers 4.17.0.dev0
|