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: ATCO2-ASR and
ATCOSIM.
type: Jzuluaga/atcosim_corpus
args: 'config: en, split: test small split [3000]'
metrics:
- name: Wer
type: wer
value: 8.947972793922798
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: ATCO2-ASR and ATCOSIM. dataset. It achieves the following results on the evaluation set:
- Loss: 0.2504
- Wer: 8.9480
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 |
|---|---|---|---|---|
| 0.0569 | 0.5319 | 100 | 0.3078 | 11.7922 |
| 0.0307 | 1.0638 | 200 | 0.2728 | 10.8515 |
| 0.0253 | 1.5957 | 300 | 0.2762 | 10.7190 |
| 0.0066 | 2.1277 | 400 | 0.2551 | 9.0761 |
| 0.0061 | 2.6596 | 500 | 0.2526 | 9.5795 |
| 0.002 | 3.1915 | 600 | 0.2504 | 9.0010 |
| 0.0019 | 3.7234 | 700 | 0.2561 | 9.2880 |
| 0.0007 | 4.2553 | 800 | 0.2535 | 9.1026 |
| 0.001 | 4.7872 | 900 | 0.2495 | 8.9259 |
| 0.0003 | 5.3191 | 1000 | 0.2504 | 8.9480 |
Framework versions
- Transformers 4.41.2
- Pytorch 2.3.0+cu121
- Datasets 2.19.2
- Tokenizers 0.19.1