w2v-bert-2.0-ptbr2

This model is a fine-tuned version of facebook/w2v-bert-2.0 on the None dataset. It achieves the following results on the evaluation set:

• Loss: inf
• Wer: 0.7379

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: 0.0003
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 16
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 20
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
24.0954	3.64	25	inf	1.0049
9.1013	7.16	50	inf	0.9126
4.5420	10.8	75	inf	0.7524
2.5565	14.32	100	inf	0.7718
1.6594	17.96	125	inf	0.7961
0.9546	21.48	150	inf	0.7573
0.4813	25.0	175	inf	0.7379
0.1315	28.64	200	inf	0.7330

Framework versions

• Transformers 5.0.0
• Pytorch 2.10.0+cu128
• Datasets 3.6.0
• Tokenizers 0.22.2