zuazo/whisper-tiny-eu-train

Whisper Tiny Basque

This model is a fine-tuned version of openai/whisper-tiny on the mozilla-foundation/common_voice_13_0 eu dataset. It achieves the following results on the evaluation set:

• Loss: 0.6509
• Wer: 34.0869

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: 3.75e-05
• train_batch_size: 256
• eval_batch_size: 128
• 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
0.3011	23.26	1000	0.6017	41.4677
0.0904	46.51	2000	0.5919	35.6284
0.0408	69.77	3000	0.6267	34.7095
0.0265	93.02	4000	0.6420	34.3932
0.0212	116.28	5000	0.6509	34.0869

Framework versions

• Transformers 4.37.2
• Pytorch 2.2.0+cu121
• Datasets 2.16.1
• Tokenizers 0.15.1