metadata
language:
- sn
license: apache-2.0
library_name: peft
tags:
- whisper-event and peft-lora
- generated_from_trainer
base_model: openai/whisper-tiny
datasets:
- google/fleurs
metrics:
- wer
model-index:
- name: Whisper Tiny Sn - Bright Chirindo
results:
- task:
type: automatic-speech-recognition
name: Automatic Speech Recognition
dataset:
name: Google Fleurs
type: google/fleurs
config: sn_zw
split: None
args: sn_zw
metrics:
- type: wer
value: 95.27619047619048
name: Wer
Whisper Tiny Sn - Bright Chirindo
This model is a fine-tuned version of openai/whisper-tiny on the Google Fleurs dataset. It achieves the following results on the evaluation set:
- Loss: 1.9909
- Wer: 95.2762
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: 8
- 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: 5000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 2.087 | 3.0164 | 1000 | 2.4026 | 109.4095 |
| 1.8305 | 6.0328 | 2000 | 2.1613 | 101.2419 |
| 1.7145 | 9.0492 | 3000 | 2.0536 | 99.8705 |
| 1.6314 | 13.0044 | 4000 | 2.0050 | 99.0095 |
| 1.665 | 16.0208 | 5000 | 1.9909 | 95.2762 |
Framework versions
- PEFT 0.10.1.dev0
- Transformers 4.40.2
- Pytorch 2.2.1+cu121
- Datasets 2.19.2.dev0
- Tokenizers 0.19.1