wav2vec2-large-xls-r-300m-Sulaimank

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1017
• Wer: 0.1013

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: 64
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 80
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
3.0012	10.8108	400	0.3243	0.4548
0.1975	21.6216	800	0.1885	0.2591
0.0751	32.4324	1200	0.1590	0.1902
0.0398	43.2432	1600	0.1300	0.1520
0.0221	54.0541	2000	0.1195	0.1361
0.0137	64.8649	2400	0.1034	0.1112
0.0088	75.6757	2800	0.1017	0.1013

Framework versions

• Transformers 4.41.2
• Pytorch 2.1.0+cu118
• Datasets 2.20.0
• Tokenizers 0.19.1