metadata
language:
- hi
license: apache-2.0
base_model: openai/whisper-small
tags:
- hf-asr-leaderboard
- generated_from_trainer
datasets:
- aihub_elder
model-index:
- name: whisper-small-ko-E50_Y_freq_speed
results: []
whisper-small-ko-E50_Y_freq_speed
This model is a fine-tuned version of openai/whisper-small on the aihub Y dialogue dataset. It achieves the following results on the evaluation set:
- Loss: 0.1746
- Cer: 5.4570
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: 1e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- 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: 50
- num_epochs: 2
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Cer |
---|---|---|---|---|
0.4066 | 0.13 | 100 | 0.2624 | 6.2676 |
0.2544 | 0.26 | 200 | 0.2160 | 5.8036 |
0.2379 | 0.39 | 300 | 0.2100 | 5.7507 |
0.2078 | 0.52 | 400 | 0.1967 | 6.1325 |
0.1842 | 0.64 | 500 | 0.1921 | 5.4570 |
0.1653 | 0.77 | 600 | 0.1847 | 5.8564 |
0.1703 | 0.9 | 700 | 0.1809 | 5.7683 |
0.0863 | 1.03 | 800 | 0.1799 | 5.6743 |
0.0718 | 1.16 | 900 | 0.1829 | 5.1339 |
0.0763 | 1.29 | 1000 | 0.1772 | 5.7801 |
0.0709 | 1.42 | 1100 | 0.1792 | 5.6215 |
0.0661 | 1.55 | 1200 | 0.1748 | 4.9930 |
0.068 | 1.68 | 1300 | 0.1743 | 5.4100 |
0.0595 | 1.81 | 1400 | 0.1749 | 5.4864 |
0.0624 | 1.93 | 1500 | 0.1746 | 5.4570 |
Framework versions
- Transformers 4.37.0.dev0
- Pytorch 2.1.0+cu121
- Datasets 2.16.0
- Tokenizers 0.15.0