wav2vec2-large-xls-r-300m-tr-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: 0.3889
• Wer: 0.3006

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
3.8274	3.67	400	0.6752	0.6946
0.4002	7.34	800	0.4440	0.5183
0.1961	11.01	1200	0.4133	0.4052
0.1285	14.68	1600	0.4249	0.3737
0.0966	18.35	2000	0.4019	0.3606
0.0789	22.02	2400	0.4019	0.3316
0.0599	25.69	2800	0.3996	0.3078
0.047	29.36	3200	0.3889	0.3006

Framework versions

• Transformers 4.33.3
• Pytorch 2.0.1+cu118
• Datasets 2.14.5
• Tokenizers 0.13.3