update model card README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,87 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
tags:
|
| 4 |
+
- generated_from_trainer
|
| 5 |
+
datasets:
|
| 6 |
+
- common_voice
|
| 7 |
+
model-index:
|
| 8 |
+
- name: wav2vec2-large-xlsr-53-W2V2-TATAR-SMALL
|
| 9 |
+
results: []
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 13 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 14 |
+
|
| 15 |
+
# wav2vec2-large-xlsr-53-W2V2-TATAR-SMALL
|
| 16 |
+
|
| 17 |
+
This model is a fine-tuned version of [facebook/wav2vec2-large-xlsr-53](https://huggingface.co/facebook/wav2vec2-large-xlsr-53) on the common_voice dataset.
|
| 18 |
+
It achieves the following results on the evaluation set:
|
| 19 |
+
- Loss: 0.4714
|
| 20 |
+
- Wer: 0.5316
|
| 21 |
+
|
| 22 |
+
## Model description
|
| 23 |
+
|
| 24 |
+
More information needed
|
| 25 |
+
|
| 26 |
+
## Intended uses & limitations
|
| 27 |
+
|
| 28 |
+
More information needed
|
| 29 |
+
|
| 30 |
+
## Training and evaluation data
|
| 31 |
+
|
| 32 |
+
More information needed
|
| 33 |
+
|
| 34 |
+
## Training procedure
|
| 35 |
+
|
| 36 |
+
### Training hyperparameters
|
| 37 |
+
|
| 38 |
+
The following hyperparameters were used during training:
|
| 39 |
+
- learning_rate: 0.0003
|
| 40 |
+
- train_batch_size: 16
|
| 41 |
+
- eval_batch_size: 8
|
| 42 |
+
- seed: 42
|
| 43 |
+
- gradient_accumulation_steps: 2
|
| 44 |
+
- total_train_batch_size: 32
|
| 45 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 46 |
+
- lr_scheduler_type: linear
|
| 47 |
+
- lr_scheduler_warmup_steps: 500
|
| 48 |
+
- num_epochs: 30
|
| 49 |
+
- mixed_precision_training: Native AMP
|
| 50 |
+
|
| 51 |
+
### Training results
|
| 52 |
+
|
| 53 |
+
| Training Loss | Epoch | Step | Validation Loss | Wer |
|
| 54 |
+
|:-------------:|:-----:|:-----:|:---------------:|:------:|
|
| 55 |
+
| 6.2446 | 1.17 | 400 | 3.2621 | 1.0 |
|
| 56 |
+
| 1.739 | 2.35 | 800 | 0.5832 | 0.7688 |
|
| 57 |
+
| 0.4718 | 3.52 | 1200 | 0.4785 | 0.6824 |
|
| 58 |
+
| 0.3574 | 4.69 | 1600 | 0.4814 | 0.6792 |
|
| 59 |
+
| 0.2946 | 5.86 | 2000 | 0.4484 | 0.6506 |
|
| 60 |
+
| 0.2674 | 7.04 | 2400 | 0.4612 | 0.6225 |
|
| 61 |
+
| 0.2349 | 8.21 | 2800 | 0.4600 | 0.6050 |
|
| 62 |
+
| 0.2206 | 9.38 | 3200 | 0.4772 | 0.6048 |
|
| 63 |
+
| 0.2072 | 10.56 | 3600 | 0.4676 | 0.6106 |
|
| 64 |
+
| 0.1984 | 11.73 | 4000 | 0.4816 | 0.6079 |
|
| 65 |
+
| 0.1793 | 12.9 | 4400 | 0.4616 | 0.5836 |
|
| 66 |
+
| 0.172 | 14.08 | 4800 | 0.4808 | 0.5860 |
|
| 67 |
+
| 0.1624 | 15.25 | 5200 | 0.4854 | 0.5820 |
|
| 68 |
+
| 0.156 | 16.42 | 5600 | 0.4609 | 0.5656 |
|
| 69 |
+
| 0.1448 | 17.59 | 6000 | 0.4926 | 0.5817 |
|
| 70 |
+
| 0.1406 | 18.77 | 6400 | 0.4638 | 0.5654 |
|
| 71 |
+
| 0.1337 | 19.94 | 6800 | 0.4731 | 0.5652 |
|
| 72 |
+
| 0.1317 | 21.11 | 7200 | 0.4861 | 0.5639 |
|
| 73 |
+
| 0.1179 | 22.29 | 7600 | 0.4766 | 0.5521 |
|
| 74 |
+
| 0.1197 | 23.46 | 8000 | 0.4824 | 0.5584 |
|
| 75 |
+
| 0.1096 | 24.63 | 8400 | 0.5006 | 0.5559 |
|
| 76 |
+
| 0.1038 | 25.81 | 8800 | 0.4994 | 0.5440 |
|
| 77 |
+
| 0.0992 | 26.98 | 9200 | 0.4867 | 0.5405 |
|
| 78 |
+
| 0.0984 | 28.15 | 9600 | 0.4798 | 0.5361 |
|
| 79 |
+
| 0.0943 | 29.33 | 10000 | 0.4714 | 0.5316 |
|
| 80 |
+
|
| 81 |
+
|
| 82 |
+
### Framework versions
|
| 83 |
+
|
| 84 |
+
- Transformers 4.11.3
|
| 85 |
+
- Pytorch 1.10.0+cu111
|
| 86 |
+
- Datasets 1.14.0
|
| 87 |
+
- Tokenizers 0.10.3
|