whisper-tiny-french-HanNeurAI

This model is a fine-tuned version of openai/whisper-tiny on the Common Voice 11.0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.6998
• Wer: 38.8453

This fine-tuning model is part of my school project. With limitation of my compute, I scaled down the dataset

Additional information and demo code can be found in this github: HanCreation/Whisper-Tiny-German

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: 4000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.6833	0.16	1000	0.8090	43.6285
0.6272	0.32	2000	0.7441	41.3900
0.5671	0.48	3000	0.7124	40.0427
0.5593	0.64	4000	0.6998	38.8453

Framework versions

• Transformers 4.40.2
• Pytorch 2.3.0
• Datasets 2.19.1
• Tokenizers 0.19.1