update model card README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,85 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
tags:
|
| 4 |
+
- generated_from_trainer
|
| 5 |
+
model-index:
|
| 6 |
+
- name: wav2vec2-xlsr-1b-fi
|
| 7 |
+
results: []
|
| 8 |
+
---
|
| 9 |
+
|
| 10 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 11 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 12 |
+
|
| 13 |
+
# wav2vec2-xlsr-1b-fi
|
| 14 |
+
|
| 15 |
+
This model is a fine-tuned version of [facebook/wav2vec2-xls-r-1b](https://huggingface.co/facebook/wav2vec2-xls-r-1b) on the None dataset.
|
| 16 |
+
It achieves the following results on the evaluation set:
|
| 17 |
+
- Loss: 0.1235
|
| 18 |
+
- Wer: 0.1533
|
| 19 |
+
|
| 20 |
+
## Model description
|
| 21 |
+
|
| 22 |
+
More information needed
|
| 23 |
+
|
| 24 |
+
## Intended uses & limitations
|
| 25 |
+
|
| 26 |
+
More information needed
|
| 27 |
+
|
| 28 |
+
## Training and evaluation data
|
| 29 |
+
|
| 30 |
+
More information needed
|
| 31 |
+
|
| 32 |
+
## Training procedure
|
| 33 |
+
|
| 34 |
+
### Training hyperparameters
|
| 35 |
+
|
| 36 |
+
The following hyperparameters were used during training:
|
| 37 |
+
- learning_rate: 5e-05
|
| 38 |
+
- train_batch_size: 32
|
| 39 |
+
- eval_batch_size: 8
|
| 40 |
+
- seed: 42
|
| 41 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 42 |
+
- lr_scheduler_type: linear
|
| 43 |
+
- lr_scheduler_warmup_steps: 500
|
| 44 |
+
- num_epochs: 5
|
| 45 |
+
- mixed_precision_training: Native AMP
|
| 46 |
+
|
| 47 |
+
### Training results
|
| 48 |
+
|
| 49 |
+
| Training Loss | Epoch | Step | Validation Loss | Wer |
|
| 50 |
+
|:-------------:|:-----:|:-----:|:---------------:|:------:|
|
| 51 |
+
| 0.968 | 0.18 | 500 | 0.4870 | 0.4720 |
|
| 52 |
+
| 0.6557 | 0.36 | 1000 | 0.2450 | 0.2931 |
|
| 53 |
+
| 0.647 | 0.54 | 1500 | 0.1818 | 0.2255 |
|
| 54 |
+
| 0.5297 | 0.72 | 2000 | 0.1698 | 0.2354 |
|
| 55 |
+
| 0.5802 | 0.9 | 2500 | 0.1581 | 0.2355 |
|
| 56 |
+
| 0.6351 | 1.07 | 3000 | 0.1689 | 0.2336 |
|
| 57 |
+
| 0.4626 | 1.25 | 3500 | 0.1719 | 0.3099 |
|
| 58 |
+
| 0.4526 | 1.43 | 4000 | 0.1434 | 0.2069 |
|
| 59 |
+
| 0.4692 | 1.61 | 4500 | 0.1645 | 0.2192 |
|
| 60 |
+
| 0.4584 | 1.79 | 5000 | 0.1483 | 0.1987 |
|
| 61 |
+
| 0.4234 | 1.97 | 5500 | 0.1499 | 0.2178 |
|
| 62 |
+
| 0.4243 | 2.15 | 6000 | 0.1345 | 0.2070 |
|
| 63 |
+
| 0.4108 | 2.33 | 6500 | 0.1383 | 0.1850 |
|
| 64 |
+
| 0.4048 | 2.51 | 7000 | 0.1338 | 0.1811 |
|
| 65 |
+
| 0.4085 | 2.69 | 7500 | 0.1290 | 0.1780 |
|
| 66 |
+
| 0.4026 | 2.87 | 8000 | 0.1239 | 0.1650 |
|
| 67 |
+
| 0.4033 | 3.04 | 8500 | 0.1346 | 0.1657 |
|
| 68 |
+
| 0.3986 | 3.22 | 9000 | 0.1310 | 0.1850 |
|
| 69 |
+
| 0.3867 | 3.4 | 9500 | 0.1273 | 0.1741 |
|
| 70 |
+
| 0.3658 | 3.58 | 10000 | 0.1219 | 0.1672 |
|
| 71 |
+
| 0.382 | 3.76 | 10500 | 0.1306 | 0.1698 |
|
| 72 |
+
| 0.3847 | 3.94 | 11000 | 0.1230 | 0.1577 |
|
| 73 |
+
| 0.3691 | 4.12 | 11500 | 0.1310 | 0.1615 |
|
| 74 |
+
| 0.3593 | 4.3 | 12000 | 0.1296 | 0.1622 |
|
| 75 |
+
| 0.3619 | 4.48 | 12500 | 0.1285 | 0.1601 |
|
| 76 |
+
| 0.3361 | 4.66 | 13000 | 0.1261 | 0.1569 |
|
| 77 |
+
| 0.3603 | 4.84 | 13500 | 0.1235 | 0.1533 |
|
| 78 |
+
|
| 79 |
+
|
| 80 |
+
### Framework versions
|
| 81 |
+
|
| 82 |
+
- Transformers 4.17.0.dev0
|
| 83 |
+
- Pytorch 1.10.2+cu102
|
| 84 |
+
- Datasets 1.18.2.dev0
|
| 85 |
+
- Tokenizers 0.11.0
|