<!-- 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-xlsr-53-ft-cy-cv13-btb
This model is a fine-tuned version of DewiBrynJones/wav2vec2-xlsr-ft-cv13-cy on the TECHIAITH/BANC-TRAWSGRIFIADAU-BANGOR - NA dataset. It achieves the following results on the evaluation set:
- Loss: 0.4176
- Wer: 0.3091
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: 0.0003
- train_batch_size: 16
- eval_batch_size: 128
- 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: 500
- num_epochs: 5.0
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Wer |
---|---|---|---|---|
No log | 0.21 | 100 | 3.1281 | 1.0 |
No log | 0.41 | 200 | 1.7209 | 0.7265 |
No log | 0.62 | 300 | 0.7829 | 0.5030 |
No log | 0.83 | 400 | 0.6802 | 0.4631 |
2.4313 | 1.03 | 500 | 0.6083 | 0.4256 |
2.4313 | 1.24 | 600 | 0.5712 | 0.4048 |
2.4313 | 1.44 | 700 | 0.5414 | 0.4058 |
2.4313 | 1.65 | 800 | 0.5271 | 0.4020 |
2.4313 | 1.86 | 900 | 0.5028 | 0.3759 |
0.6494 | 2.06 | 1000 | 0.4823 | 0.3643 |
0.6494 | 2.27 | 1100 | 0.4736 | 0.3633 |
0.6494 | 2.48 | 1200 | 0.4692 | 0.3592 |
0.6494 | 2.68 | 1300 | 0.4537 | 0.3570 |
0.6494 | 2.89 | 1400 | 0.4439 | 0.3450 |
0.4922 | 3.1 | 1500 | 0.4498 | 0.3453 |
0.4922 | 3.3 | 1600 | 0.4514 | 0.3470 |
0.4922 | 3.51 | 1700 | 0.4591 | 0.3389 |
0.4922 | 3.72 | 1800 | 0.4357 | 0.3284 |
0.4922 | 3.92 | 1900 | 0.4202 | 0.3216 |
0.3885 | 4.13 | 2000 | 0.4241 | 0.3199 |
0.3885 | 4.33 | 2100 | 0.4328 | 0.3189 |
0.3885 | 4.54 | 2200 | 0.4206 | 0.3163 |
0.3885 | 4.75 | 2300 | 0.4177 | 0.3141 |
0.3885 | 4.95 | 2400 | 0.4181 | 0.3092 |
Framework versions
- Transformers 4.29.2
- Pytorch 2.0.1+cu117
- Datasets 2.12.0
- Tokenizers 0.13.3