metadata
license: apache-2.0
tags:
- generated_from_trainer
datasets:
- common_voice
model-index:
- name: wav2vec2-large-xls-r-300m-ia
results: []
wav2vec2-large-xls-r-300m-ia
This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the common_voice dataset. It achieves the following results on the evaluation set:
- Loss: 0.1452
- Wer: 0.1253
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 3e-05
- train_batch_size: 16
- eval_batch_size: 4
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 400
- num_epochs: 30
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 7.432 | 1.87 | 400 | 2.9636 | 1.0 |
| 2.6922 | 3.74 | 800 | 2.2111 | 0.9977 |
| 1.2581 | 5.61 | 1200 | 0.4864 | 0.4028 |
| 0.6232 | 7.48 | 1600 | 0.2807 | 0.2413 |
| 0.4479 | 9.35 | 2000 | 0.2219 | 0.1885 |
| 0.3654 | 11.21 | 2400 | 0.1886 | 0.1606 |
| 0.323 | 13.08 | 2800 | 0.1716 | 0.1444 |
| 0.2935 | 14.95 | 3200 | 0.1687 | 0.1443 |
| 0.2707 | 16.82 | 3600 | 0.1632 | 0.1382 |
| 0.2559 | 18.69 | 4000 | 0.1507 | 0.1337 |
| 0.2433 | 20.56 | 4400 | 0.1572 | 0.1358 |
| 0.2338 | 22.43 | 4800 | 0.1489 | 0.1305 |
| 0.2258 | 24.3 | 5200 | 0.1485 | 0.1278 |
| 0.2218 | 26.17 | 5600 | 0.1470 | 0.1272 |
| 0.2169 | 28.04 | 6000 | 0.1470 | 0.1270 |
| 0.2117 | 29.91 | 6400 | 0.1452 | 0.1253 |
Framework versions
- Transformers 4.17.0.dev0
- Pytorch 1.10.0+cu111
- Datasets 1.18.3
- Tokenizers 0.11.0