<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
wav2vec2-10epochs-3e3
This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53 on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.7024
- Wer: 0.6481
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: 0.003
- train_batch_size: 4
- eval_batch_size: 4
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 8
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 10
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Wer |
---|---|---|---|---|
0.0938 | 0.36 | 100 | 0.4842 | 0.3227 |
0.67 | 0.72 | 200 | 0.7219 | 0.5669 |
0.7133 | 1.08 | 300 | 1.0698 | 0.7080 |
1.0312 | 1.44 | 400 | 1.2692 | 0.8953 |
1.2162 | 1.8 | 500 | 1.4763 | 1.0443 |
1.2401 | 2.16 | 600 | 1.4906 | 0.8694 |
1.2022 | 2.52 | 700 | 1.3686 | 0.9518 |
1.154 | 2.88 | 800 | 1.1618 | 0.9109 |
1.0467 | 3.24 | 900 | 1.2007 | 0.8602 |
1.1785 | 3.6 | 1000 | 1.2000 | 0.9160 |
0.979 | 3.96 | 1100 | 1.1464 | 0.8852 |
1.1421 | 4.32 | 1200 | 1.1117 | 0.9018 |
0.9622 | 4.68 | 1300 | 1.0976 | 0.8602 |
1.0939 | 5.04 | 1400 | 1.1126 | 0.8831 |
0.9414 | 5.4 | 1500 | 1.0134 | 0.8448 |
0.9433 | 5.76 | 1600 | 0.9320 | 0.7977 |
0.8389 | 6.12 | 1700 | 0.9013 | 0.7742 |
0.8838 | 6.47 | 1800 | 0.9088 | 0.7509 |
0.7907 | 6.83 | 1900 | 0.8581 | 0.7382 |
0.7704 | 7.19 | 2000 | 0.8300 | 0.7481 |
0.667 | 7.55 | 2100 | 0.8221 | 0.7349 |
0.6111 | 7.91 | 2200 | 0.7803 | 0.7102 |
0.5555 | 8.27 | 2300 | 0.8198 | 0.7314 |
0.4947 | 8.63 | 2400 | 0.8127 | 0.7036 |
0.4697 | 8.99 | 2500 | 0.7514 | 0.6805 |
0.402 | 9.35 | 2600 | 0.7348 | 0.6606 |
0.3682 | 9.71 | 2700 | 0.7024 | 0.6481 |
Framework versions
- Transformers 4.11.3
- Pytorch 1.10.1
- Datasets 2.9.0
- Tokenizers 0.10.3