wav2vec2-large-xls-r-300m-gn

This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53-spanish on the common_voice_13_0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.3713
• Wer: 0.3431

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: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.7177	3.62	400	0.3649	0.5816
0.2738	7.24	800	0.4029	0.5024
0.1768	10.86	1200	0.3779	0.4285
0.1128	14.48	1600	0.3929	0.4205
0.0842	18.1	2000	0.3683	0.3916
0.0616	21.72	2400	0.3943	0.3675
0.0461	25.34	2800	0.4127	0.3571
0.0368	28.96	3200	0.3713	0.3431

Framework versions

• Transformers 4.35.0.dev0
• Pytorch 2.1.0+cu118
• Datasets 2.14.6
• Tokenizers 0.14.1