zuazo/whisper-medium-eu-from-es

Whisper Medium Basque

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

• Loss: 0.4263
• Wer: 14.6867

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_steps: 500
• training_steps: 10000

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.0225	4.02	1000	0.3040	17.9209
0.0033	9.01	2000	0.3343	16.1833
0.0019	14.01	3000	0.3607	15.8127
0.0015	19.01	4000	0.3632	15.6203
0.0013	24.01	5000	0.3704	15.7520
0.0001	29.0	6000	0.3977	14.7940
0.0001	33.02	7000	0.4104	14.7191
0.0001	38.02	8000	0.4198	14.6928
0.0001	43.01	9000	0.4263	14.6867
0.0	48.01	10000	0.4293	14.7130

Framework versions

• Transformers 4.33.0.dev0
• Pytorch 2.0.1+cu117
• Datasets 2.14.4
• Tokenizers 0.13.3