breco/whisper-tiny-few-reports

Whisper Tiny Few Reports - vfranchis

This model is a fine-tuned version of openai/whisper-tiny on the Few reports 1.0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.9024
• Wer: 99.3499

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: 4
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 10
• training_steps: 100
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
3.9187	3.0769	10	3.0330	102.4919
2.7035	6.1538	20	2.1277	126.2189
2.0516	9.2308	30	1.6559	82.9902
1.5833	12.3077	40	1.3689	83.5320
1.301	15.3846	50	1.1881	98.0498
1.1102	18.4615	60	1.0681	111.9177
1.0133	21.5385	70	0.9908	106.1755
0.8926	24.6154	80	0.9402	100.3250
0.8362	27.6923	90	0.9127	99.2416
0.845	30.7692	100	0.9024	99.3499

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.1+cu121
• Datasets 2.21.0
• Tokenizers 0.19.1