xlsr53-ptbr-CORAA-2

This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53-portuguese on the None dataset. It achieves the following results on the evaluation set:

• Loss: nan
• Wer: 0.9826
• Cer: 0.9730

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-07
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 8
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
0.0	1.0	75	nan	0.9826	0.9730
0.0	2.0	150	nan	0.9826	0.9730
0.0	3.0	225	nan	0.9826	0.9730
0.0	4.0	300	nan	0.9826	0.9730
0.0	5.0	375	nan	0.9826	0.9730
0.0	6.0	450	nan	0.9826	0.9730
0.0	7.0	525	nan	0.9826	0.9730
0.0	8.0	600	nan	0.9826	0.9730
0.0	9.0	675	nan	0.9826	0.9730
0.0	10.0	750	nan	0.9826	0.9730

Framework versions

• Transformers 4.48.3
• Pytorch 2.5.1+cu124
• Datasets 3.3.2
• Tokenizers 0.21.0