metadata
language:
- en
license: apache-2.0
base_model: openai/whisper-small
tags:
- generated_from_trainer
datasets:
- Jzuluaga/atcosim_corpus
metrics:
- wer
model-index:
- name: Whisper small - Whisper with atcosim
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: 'This is a dataset constructed from two datasets: ATCOSIM.'
type: Jzuluaga/atcosim_corpus
args: 'config: en, split: test small split [full]'
metrics:
- name: Wer
type: wer
value: 1.4177192827488738
Whisper small - Whisper with atcosim
This model is a fine-tuned version of openai/whisper-small on the This is a dataset constructed from two datasets: ATCOSIM. dataset. It achieves the following results on the evaluation set:
- Loss: 0.0385
- Wer: 1.4177
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: 50
- training_steps: 1000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.0349 | 0.2092 | 100 | 0.0974 | 6.4040 |
| 0.0493 | 0.4184 | 200 | 0.0664 | 3.2329 |
| 0.0464 | 0.6276 | 300 | 0.0519 | 2.8708 |
| 0.0394 | 0.8368 | 400 | 0.0474 | 2.3055 |
| 0.0177 | 1.0460 | 500 | 0.0429 | 1.7004 |
| 0.0054 | 1.2552 | 600 | 0.0416 | 1.5458 |
| 0.0182 | 1.4644 | 700 | 0.0411 | 1.5193 |
| 0.008 | 1.6736 | 800 | 0.0400 | 1.4663 |
| 0.0055 | 1.8828 | 900 | 0.0387 | 1.4619 |
| 0.0053 | 2.0921 | 1000 | 0.0385 | 1.4177 |
Framework versions
- Transformers 4.41.2
- Pytorch 2.3.0+cu121
- Datasets 2.20.0
- Tokenizers 0.19.1