metadata
language:
- ko
license: apache-2.0
base_model: openai/whisper-base
tags:
- hf-asr-leaderboard
- generated_from_trainer
datasets:
- S000001
model-index:
- name: openai/whisper-base
results: []
openai/whisper-base
This model is a fine-tuned version of openai/whisper-base on the /nas/data/lowband_telephone/wav/training/D01/J01/S000001 dataset. It achieves the following results on the evaluation set:
- Loss: 1.3994
- Cer: 18.3333
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: 64
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- training_steps: 4000
Training results
| Training Loss | Epoch | Step | Validation Loss | Cer |
|---|---|---|---|---|
| 0.0 | 1000.0 | 1000 | 1.2835 | 17.5 |
| 0.0 | 2000.0 | 2000 | 1.3486 | 18.3333 |
| 0.0 | 3000.0 | 3000 | 1.3850 | 18.3333 |
| 0.0 | 4000.0 | 4000 | 1.3994 | 18.3333 |
Framework versions
- Transformers 4.36.0.dev0
- Pytorch 2.0.1+cu117
- Datasets 2.14.6
- Tokenizers 0.14.1