metadata
language:
- es
license: apache-2.0
base_model: openai/whisper-medium
tags:
- generated_from_trainer
datasets:
- Mezosky/es_clinical_assistance_10k
metrics:
- wer
model-index:
- name: Whisper Chilean Spanish Medium
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: Mezosky/es_clinical_assistance_10k
type: Mezosky/es_clinical_assistance_10k
metrics:
- name: Wer
type: wer
value: 7.774513918030494
Whisper Chilean Spanish Medium
This model is a fine-tuned version of openai/whisper-medium on the Mezosky/es_clinical_assistance_10k dataset. It achieves the following results on the evaluation set:
- Loss: 0.1058
- Wer: 7.7745
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: 500
- training_steps: 1000
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.6275 | 0.17 | 100 | 0.5455 | 13.3333 |
| 0.185 | 0.34 | 200 | 0.1782 | 10.7316 |
| 0.1523 | 0.51 | 300 | 0.1539 | 10.9106 |
| 0.1373 | 0.69 | 400 | 0.1399 | 10.1329 |
| 0.1538 | 0.86 | 500 | 0.1322 | 17.5493 |
| 0.1007 | 1.03 | 600 | 0.1238 | 8.4963 |
| 0.0782 | 1.2 | 700 | 0.1187 | 8.4599 |
| 0.0722 | 1.37 | 800 | 0.1128 | 7.8137 |
| 0.0715 | 1.54 | 900 | 0.1081 | 7.6934 |
| 0.0927 | 1.72 | 1000 | 0.1058 | 7.7745 |
Framework versions
- Transformers 4.39.3
- Pytorch 2.2.2+cu121
- Datasets 2.18.0
- Tokenizers 0.15.2