wav2vec2-arabic-colab

This model is a fine-tuned version of facebook/wav2vec2-xls-r-1b on the audiofolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.7757
• Wer: 36.8162
• Cer: 11.3997

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 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: 100
• training_steps: 1000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
1.7976	5.2685	200	0.5703	48.0926	14.8747
0.1992	10.5369	400	0.6327	43.5101	13.4424
0.095	15.8054	600	0.7045	39.7553	12.3694
0.0598	21.0537	800	0.7791	39.0475	11.9655
0.0442	26.3221	1000	0.7757	36.8162	11.3997

Framework versions

• Transformers 4.51.1
• Pytorch 2.6.0+cu124
• Datasets 3.5.0
• Tokenizers 0.21.1