./whisper-large-cit-synth-do015-wd0-lr5e-06-1000
This model is a fine-tuned version of openai/whisper-large-v3 on the SF 1000 dataset. It achieves the following results on the evaluation set:
- Loss: 0.4526
- Wer: 20.3899
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: 5e-06
- train_batch_size: 4
- eval_batch_size: 8
- seed: 42
- distributed_type: multi-GPU
- gradient_accumulation_steps: 4
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 100
- training_steps: 500
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Wer |
---|---|---|---|---|
0.7187 | 0.8889 | 50 | 0.4062 | 24.2105 |
0.4122 | 1.7778 | 100 | 0.3523 | 22.3782 |
0.2917 | 2.6667 | 150 | 0.3494 | 23.5867 |
0.2242 | 3.5556 | 200 | 0.3618 | 23.0019 |
0.1529 | 4.4444 | 250 | 0.3770 | 22.3392 |
0.1322 | 5.3333 | 300 | 0.3906 | 21.2476 |
0.0987 | 6.2222 | 350 | 0.4133 | 20.9747 |
0.0798 | 7.1111 | 400 | 0.4302 | 23.8986 |
0.0613 | 8.0 | 450 | 0.4438 | 20.5848 |
0.0545 | 8.8889 | 500 | 0.4526 | 20.3899 |
Framework versions
- Transformers 4.42.3
- Pytorch 1.13.1+cu117
- Datasets 2.20.0
- Tokenizers 0.19.1
- Downloads last month
- 0