wav2vec-xlsr-53-turkish-v3

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

• Loss: inf
• Wer: 0.2599
• Cer: 0.0658

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: 3e-05
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• 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: 750
• num_epochs: 12
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
3.038	1.3123	1000	inf	1.0	1.0
0.7162	2.6247	2000	inf	0.4285	0.1028
0.5657	3.9370	3000	inf	0.3114	0.0762
0.5062	5.2493	4000	inf	0.2852	0.0699
0.5259	6.5617	5000	inf	0.2752	0.0688
0.4808	7.8740	6000	inf	0.2638	0.0664
0.4881	9.1864	7000	inf	0.2616	0.0664
0.5237	10.4987	8000	inf	0.2639	0.0666
0.5179	11.8110	9000	inf	0.2599	0.0658

Framework versions

• Transformers 4.56.1
• Pytorch 2.8.0+cu128
• Datasets 3.6.0
• Tokenizers 0.22.0