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: 0.7496
• Wer: 0.6899
• Cer: 0.1611

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: 200
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
6.2963	5.71	100	3.1890	1.0	0.9948
3.044	11.43	200	2.8236	1.0	0.9855
1.3443	17.14	300	0.8382	0.7962	0.2023
0.3201	22.86	400	0.8201	0.7369	0.1844
0.1987	28.57	500	0.7496	0.6899	0.1611

Framework versions

• Transformers 4.31.0
• Pytorch 2.0.1+cu118
• Datasets 1.18.3
• Tokenizers 0.13.3