metadata
language:
- ko
license: apache-2.0
base_model: openai/whisper-medium
tags:
- hf-asr-leaderboard
- generated_from_trainer
datasets:
- Marcusxx/chungnam_firestation
model-index:
- name: chungnam_firestation_model
results: []
chungnam_firestation_model
This model is a fine-tuned version of openai/whisper-medium on the Marcusxx/chungnam_firestation dataset. It achieves the following results on the evaluation set:
- Loss: 0.0714
- Cer: 22.6763
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: 16
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 250
- training_steps: 2000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Cer |
|---|---|---|---|---|
| 0.2819 | 1.3245 | 200 | 0.3365 | 276.3165 |
| 0.0471 | 2.6490 | 400 | 0.1175 | 51.0791 |
| 0.0182 | 3.9735 | 600 | 0.0832 | 52.0576 |
| 0.0059 | 5.2980 | 800 | 0.0678 | 24.5755 |
| 0.0015 | 6.6225 | 1000 | 0.0668 | 37.5540 |
| 0.0004 | 7.9470 | 1200 | 0.0673 | 23.5396 |
| 0.0002 | 9.2715 | 1400 | 0.0693 | 24.8058 |
| 0.0002 | 10.5960 | 1600 | 0.0704 | 23.7986 |
| 0.0002 | 11.9205 | 1800 | 0.0712 | 22.5899 |
| 0.0002 | 13.2450 | 2000 | 0.0714 | 22.6763 |
Framework versions
- Transformers 4.41.2
- Pytorch 2.2.2+cu121
- Datasets 3.2.0
- Tokenizers 0.19.1