zuazo/whisper-base-pt

Whisper Base Portuguese

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

• Loss: 0.6511
• Wer: 18.8085

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: 2.5e-05
• train_batch_size: 128
• eval_batch_size: 64
• 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: 20000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.0324	7.04	1000	0.4417	19.1174
0.0064	14.08	2000	0.5218	19.4279
0.0037	21.13	3000	0.5563	19.4871
0.0029	28.17	4000	0.5665	19.4279
0.0018	35.21	5000	0.6106	19.6333
0.0021	42.25	6000	0.6238	20.0588
0.0003	49.3	7000	0.6348	19.3277
0.0002	56.34	8000	0.6511	18.8085
0.0001	63.38	9000	0.6660	18.8102
0.0001	70.42	10000	0.6805	18.8595
0.0001	77.46	11000	0.6956	18.9597
0.0	84.51	12000	0.7114	19.0336
0.0	91.55	13000	0.7280	19.0583
0.0	98.59	14000	0.7444	19.1519
0.0	105.63	15000	0.7608	19.2094
0.0	112.68	16000	0.7768	19.2028
0.0	119.72	17000	0.7913	19.2735
0.0	126.76	18000	0.8042	19.2899
0.0	133.8	19000	0.8136	19.3261
0.0	140.85	20000	0.8176	19.3474

Framework versions

• Transformers 4.37.2
• Pytorch 2.0.1+cu117
• Datasets 2.14.4
• Tokenizers 0.15.1