update model card README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,83 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
language:
|
| 3 |
+
- mr
|
| 4 |
+
license: apache-2.0
|
| 5 |
+
tags:
|
| 6 |
+
- automatic-speech-recognition
|
| 7 |
+
- mozilla-foundation/common_voice_8_0
|
| 8 |
+
- generated_from_trainer
|
| 9 |
+
datasets:
|
| 10 |
+
- common_voice
|
| 11 |
+
model-index:
|
| 12 |
+
- name: ''
|
| 13 |
+
results: []
|
| 14 |
+
---
|
| 15 |
+
|
| 16 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 17 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 18 |
+
|
| 19 |
+
#
|
| 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 MOZILLA-FOUNDATION/COMMON_VOICE_8_0 - MR dataset.
|
| 22 |
+
It achieves the following results on the evaluation set:
|
| 23 |
+
- Loss: 0.5951
|
| 24 |
+
- Wer: 0.5435
|
| 25 |
+
|
| 26 |
+
## Model description
|
| 27 |
+
|
| 28 |
+
More information needed
|
| 29 |
+
|
| 30 |
+
## Intended uses & limitations
|
| 31 |
+
|
| 32 |
+
More information needed
|
| 33 |
+
|
| 34 |
+
## Training and evaluation data
|
| 35 |
+
|
| 36 |
+
More information needed
|
| 37 |
+
|
| 38 |
+
## Training procedure
|
| 39 |
+
|
| 40 |
+
### Training hyperparameters
|
| 41 |
+
|
| 42 |
+
The following hyperparameters were used during training:
|
| 43 |
+
- learning_rate: 7.5e-05
|
| 44 |
+
- train_batch_size: 8
|
| 45 |
+
- eval_batch_size: 8
|
| 46 |
+
- seed: 42
|
| 47 |
+
- gradient_accumulation_steps: 4
|
| 48 |
+
- total_train_batch_size: 32
|
| 49 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 50 |
+
- lr_scheduler_type: linear
|
| 51 |
+
- lr_scheduler_warmup_steps: 2000
|
| 52 |
+
- num_epochs: 400.0
|
| 53 |
+
- mixed_precision_training: Native AMP
|
| 54 |
+
|
| 55 |
+
### Training results
|
| 56 |
+
|
| 57 |
+
| Training Loss | Epoch | Step | Validation Loss | Wer |
|
| 58 |
+
|:-------------:|:------:|:----:|:---------------:|:------:|
|
| 59 |
+
| 4.2706 | 22.73 | 500 | 4.0174 | 1.0 |
|
| 60 |
+
| 3.2492 | 45.45 | 1000 | 3.2309 | 0.9908 |
|
| 61 |
+
| 1.9709 | 68.18 | 1500 | 1.0651 | 0.8440 |
|
| 62 |
+
| 1.4088 | 90.91 | 2000 | 0.5765 | 0.6550 |
|
| 63 |
+
| 1.1326 | 113.64 | 2500 | 0.4842 | 0.5760 |
|
| 64 |
+
| 0.9709 | 136.36 | 3000 | 0.4785 | 0.6013 |
|
| 65 |
+
| 0.8433 | 159.09 | 3500 | 0.5048 | 0.5419 |
|
| 66 |
+
| 0.7404 | 181.82 | 4000 | 0.5052 | 0.5339 |
|
| 67 |
+
| 0.6589 | 204.55 | 4500 | 0.5237 | 0.5897 |
|
| 68 |
+
| 0.5831 | 227.27 | 5000 | 0.5166 | 0.5447 |
|
| 69 |
+
| 0.5375 | 250.0 | 5500 | 0.5292 | 0.5487 |
|
| 70 |
+
| 0.4784 | 272.73 | 6000 | 0.5480 | 0.5596 |
|
| 71 |
+
| 0.4421 | 295.45 | 6500 | 0.5682 | 0.5467 |
|
| 72 |
+
| 0.4047 | 318.18 | 7000 | 0.5681 | 0.5447 |
|
| 73 |
+
| 0.3779 | 340.91 | 7500 | 0.5783 | 0.5347 |
|
| 74 |
+
| 0.3525 | 363.64 | 8000 | 0.5856 | 0.5367 |
|
| 75 |
+
| 0.3393 | 386.36 | 8500 | 0.5960 | 0.5359 |
|
| 76 |
+
|
| 77 |
+
|
| 78 |
+
### Framework versions
|
| 79 |
+
|
| 80 |
+
- Transformers 4.16.0.dev0
|
| 81 |
+
- Pytorch 1.10.1+cu113
|
| 82 |
+
- Datasets 1.18.1.dev0
|
| 83 |
+
- Tokenizers 0.11.0
|