wav2vec2-large-xls-r-300m-turkish-colab

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the common_voice dataset. It achieves the following results on the evaluation set:

• Loss: 0.3767
• Wer: 0.3320

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: 16
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
3.7794	3.67	400	0.6654	0.7095
0.3983	7.34	800	0.4403	0.4597
0.1914	11.01	1200	0.4143	0.4442
0.1289	14.68	1600	0.4276	0.3993
0.0966	18.35	2000	0.3938	0.3750
0.0775	22.02	2400	0.3959	0.3604
0.0597	25.69	2800	0.3758	0.3443
0.0471	29.36	3200	0.3767	0.3320

Framework versions

• Transformers 4.32.0
• Pytorch 2.0.1+cu118
• Datasets 1.18.3
• Tokenizers 0.13.3