whisper-base-google-fleurs-pt-br

This model is a fine-tuned version of openai/whisper-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.4987
• Wer: 22.2974
• Wer Normalized: 18.5291

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: 2.05e-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: reduce_lr_on_plateau
• lr_scheduler_warmup_steps: 120
• training_steps: 2400
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Wer Normalized
0.3716	1.01	400	0.3988	21.8039	17.8916
0.2003	2.02	800	0.4440	22.3350	18.6242
0.0571	3.02	1200	0.4960	22.5982	19.2284
0.03	4.03	1600	0.4987	22.2974	18.5291

Framework versions

• Transformers 4.36.2
• Pytorch 2.1.1
• Datasets 2.16.1
• Tokenizers 0.15.0