metadata
language:
- ko
license: apache-2.0
tags:
- hf-asr-leaderboard
- generated_from_trainer
base_model: openai/whisper-medium
datasets:
- Marcusxx/cheonanAddresses
model-index:
- name: cheonanAddresses_torchmodel_model
results: []
cheonanAddresses_torchmodel_model
This model is a fine-tuned version of openai/whisper-medium on the Marcusxx/cheonanAddresses dataset. It achieves the following results on the evaluation set:
- Loss: 0.0531
- Cer: 1.9003
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: 100
- training_steps: 10000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Cer |
|---|---|---|---|---|
| 0.0771 | 0.3101 | 1000 | 0.0758 | 28.1409 |
| 0.0713 | 0.6202 | 2000 | 0.0718 | 2.2685 |
| 0.0629 | 0.9302 | 3000 | 0.0682 | 2.3271 |
| 0.0583 | 1.2403 | 4000 | 0.0637 | 2.1723 |
| 0.0623 | 1.5504 | 5000 | 0.0613 | 2.1513 |
| 0.0568 | 1.8605 | 6000 | 0.0598 | 2.2237 |
| 0.051 | 2.1705 | 7000 | 0.0568 | 2.1897 |
| 0.0448 | 2.4806 | 8000 | 0.0561 | 2.9282 |
| 0.0517 | 2.7907 | 9000 | 0.0539 | 1.9350 |
| 0.0382 | 3.1008 | 10000 | 0.0531 | 1.9003 |
Framework versions
- Transformers 4.41.2
- Pytorch 2.2.2+cu121
- Datasets 2.19.2
- Tokenizers 0.19.1