metadata
license: apache-2.0
tags:
- generated_from_trainer
base_model: facebook/hubert-large-ll60k
metrics:
- wer
model-index:
- name: speech_ocean_hubert_mdd
results: []
speech_ocean_hubert_mdd
This model is a fine-tuned version of facebook/hubert-large-ll60k on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.1563
- Wer: 0.0612
- Cer: 0.0573
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: 8
- 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: 20
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
---|---|---|---|---|---|
24.692 | 0.9873 | 39 | 19.7376 | 0.9997 | 0.9999 |
6.7786 | 2.0 | 79 | 3.8382 | 1.0 | 1.0 |
3.77 | 2.9873 | 118 | 3.2211 | 1.0 | 1.0 |
3.1537 | 4.0 | 158 | 3.1178 | 1.0 | 1.0 |
3.0849 | 4.9873 | 197 | 3.0815 | 1.0 | 1.0 |
3.08 | 6.0 | 237 | 3.0583 | 1.0 | 1.0 |
3.0887 | 6.9873 | 276 | 3.0658 | 1.0 | 1.0 |
3.0528 | 8.0 | 316 | 3.0221 | 1.0 | 1.0 |
2.9694 | 8.9873 | 355 | 2.8183 | 1.0 | 1.0 |
2.6127 | 10.0 | 395 | 2.3266 | 0.9892 | 0.9935 |
2.0376 | 10.9873 | 434 | 1.6122 | 0.6557 | 0.6453 |
1.692 | 12.0 | 474 | 1.1974 | 0.4865 | 0.4708 |
1.245 | 12.9873 | 513 | 0.7927 | 0.2645 | 0.2425 |
0.9455 | 14.0 | 553 | 0.5426 | 0.1636 | 0.1371 |
0.7612 | 14.9873 | 592 | 0.3803 | 0.1214 | 0.1069 |
0.6122 | 16.0 | 632 | 0.2850 | 0.0912 | 0.0824 |
0.4748 | 16.9873 | 671 | 0.2247 | 0.0777 | 0.0705 |
0.4095 | 18.0 | 711 | 0.1796 | 0.0676 | 0.0626 |
0.348 | 18.9873 | 750 | 0.1623 | 0.0638 | 0.0595 |
0.3615 | 19.7468 | 780 | 0.1563 | 0.0612 | 0.0573 |
Framework versions
- Transformers 4.40.0
- Pytorch 2.2.1+cu121
- Datasets 2.19.0
- Tokenizers 0.19.1