metadata
license: apache-2.0
tags:
- generated_from_trainer
datasets:
- fleurs
metrics:
- wer
model-index:
- name: openai/whisper-tiny
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: fleurs
type: fleurs
config: en_us
split: validation
args: en_us
metrics:
- name: Wer
type: wer
value: 19.3465805193222
openai/whisper-tiny
This model is a fine-tuned version of openai/whisper-tiny on the fleurs dataset. It achieves the following results on the evaluation set:
- Loss: 0.5568
- Wer: 19.3466
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_ratio: 0.2
- training_steps: 407
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 1.1599 | 0.1 | 40 | 1.1427 | 15.2139 |
| 0.4655 | 1.1 | 80 | 0.5613 | 17.5911 |
| 0.2753 | 2.09 | 120 | 0.5241 | 17.2132 |
| 0.2077 | 3.09 | 160 | 0.5242 | 17.2620 |
| 0.1636 | 4.09 | 200 | 0.5290 | 17.6643 |
| 0.1322 | 5.09 | 240 | 0.5351 | 18.2128 |
| 0.123 | 6.08 | 280 | 0.5429 | 18.9077 |
| 0.1074 | 7.08 | 320 | 0.5500 | 19.0540 |
| 0.1007 | 8.08 | 360 | 0.5553 | 19.3100 |
| 0.0876 | 9.08 | 400 | 0.5568 | 19.3466 |
Framework versions
- Transformers 4.26.0.dev0
- Pytorch 1.13.1+cu117
- Datasets 2.7.1.dev0
- Tokenizers 0.13.2