metadata
base_model: openai/whisper-base
library_name: peft
license: apache-2.0
tags:
- generated_from_trainer
model-index:
- name: whisper-base-akan-v1
results: []
whisper-base-akan-v1
This model is a fine-tuned version of openai/whisper-base on the None dataset. It achieves the following results on the evaluation set:
- Loss: 1.0715
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-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: 50
- training_steps: 3000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 1.8659 | 5.0 | 250 | 1.6244 |
| 1.2252 | 10.0 | 500 | 1.3473 |
| 1.0708 | 15.0 | 750 | 1.2260 |
| 0.922 | 20.0 | 1000 | 1.1617 |
| 0.8532 | 25.0 | 1250 | 1.1217 |
| 0.7747 | 30.0 | 1500 | 1.0973 |
| 0.7356 | 35.0 | 1750 | 1.0864 |
| 0.6894 | 40.0 | 2000 | 1.0767 |
| 0.6671 | 45.0 | 2250 | 1.0746 |
| 0.642 | 50.0 | 2500 | 1.0719 |
| 0.6306 | 55.0 | 2750 | 1.0707 |
| 0.6225 | 60.0 | 3000 | 1.0715 |
Framework versions
- PEFT 0.13.2.dev0
- Transformers 4.44.2
- Pytorch 2.4.1+cu121
- Datasets 3.0.1
- Tokenizers 0.19.1