update model card README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,86 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
tags:
|
| 4 |
+
- generated_from_trainer
|
| 5 |
+
datasets:
|
| 6 |
+
- common_voice
|
| 7 |
+
model-index:
|
| 8 |
+
- name: wav2vec2-large-xls-r-300m-russian-colab-beam_search_test
|
| 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-xls-r-300m-russian-colab-beam_search_test
|
| 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 |
+
It achieves the following results on the evaluation set:
|
| 19 |
+
- Loss: 0.7619
|
| 20 |
+
- Wer: 0.4680
|
| 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: 4
|
| 44 |
+
- total_train_batch_size: 64
|
| 45 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 46 |
+
- lr_scheduler_type: linear
|
| 47 |
+
- lr_scheduler_warmup_steps: 800
|
| 48 |
+
- num_epochs: 100
|
| 49 |
+
- mixed_precision_training: Native AMP
|
| 50 |
+
|
| 51 |
+
### Training results
|
| 52 |
+
|
| 53 |
+
| Training Loss | Epoch | Step | Validation Loss | Wer |
|
| 54 |
+
|:-------------:|:-----:|:----:|:---------------:|:------:|
|
| 55 |
+
| 10.0158 | 4.16 | 100 | 5.4134 | 1.0 |
|
| 56 |
+
| 4.0394 | 8.33 | 200 | 3.4304 | 1.0 |
|
| 57 |
+
| 3.2721 | 12.49 | 300 | 3.2273 | 1.0 |
|
| 58 |
+
| 3.1277 | 16.66 | 400 | 2.8023 | 0.9984 |
|
| 59 |
+
| 1.3791 | 20.82 | 500 | 0.9888 | 0.8546 |
|
| 60 |
+
| 0.3659 | 24.99 | 600 | 0.7602 | 0.6304 |
|
| 61 |
+
| 0.1858 | 29.16 | 700 | 0.7965 | 0.6156 |
|
| 62 |
+
| 0.1403 | 33.33 | 800 | 0.7998 | 0.5839 |
|
| 63 |
+
| 0.1173 | 37.49 | 900 | 0.8353 | 0.5941 |
|
| 64 |
+
| 0.0917 | 41.66 | 1000 | 0.8272 | 0.5522 |
|
| 65 |
+
| 0.0743 | 45.82 | 1100 | 0.8342 | 0.5471 |
|
| 66 |
+
| 0.063 | 49.99 | 1200 | 0.7988 | 0.5352 |
|
| 67 |
+
| 0.0528 | 54.16 | 1300 | 0.7740 | 0.5201 |
|
| 68 |
+
| 0.0456 | 58.33 | 1400 | 0.7636 | 0.5165 |
|
| 69 |
+
| 0.0389 | 62.49 | 1500 | 0.7922 | 0.5161 |
|
| 70 |
+
| 0.0329 | 66.66 | 1600 | 0.8035 | 0.5158 |
|
| 71 |
+
| 0.0283 | 70.82 | 1700 | 0.7873 | 0.4832 |
|
| 72 |
+
| 0.0255 | 74.99 | 1800 | 0.7853 | 0.4870 |
|
| 73 |
+
| 0.0236 | 79.16 | 1900 | 0.8236 | 0.5045 |
|
| 74 |
+
| 0.0202 | 83.33 | 2000 | 0.7661 | 0.4796 |
|
| 75 |
+
| 0.0165 | 87.49 | 2100 | 0.7584 | 0.4680 |
|
| 76 |
+
| 0.0156 | 91.66 | 2200 | 0.7685 | 0.4772 |
|
| 77 |
+
| 0.0149 | 95.82 | 2300 | 0.7519 | 0.4696 |
|
| 78 |
+
| 0.0126 | 99.99 | 2400 | 0.7619 | 0.4680 |
|
| 79 |
+
|
| 80 |
+
|
| 81 |
+
### Framework versions
|
| 82 |
+
|
| 83 |
+
- Transformers 4.11.3
|
| 84 |
+
- Pytorch 1.10.0+cu111
|
| 85 |
+
- Datasets 1.18.3
|
| 86 |
+
- Tokenizers 0.10.3
|