wav2vec2-large-xlrs-korean-v4

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

• Loss: 0.1717
• Wer: 0.3764

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: 32
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 256
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 1000
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
2.1045	5.7471	500	0.7609	0.8794
0.3215	11.4943	1000	0.2375	0.4843
0.1928	17.2414	1500	0.1717	0.3764

Framework versions

• Transformers 4.41.0
• Pytorch 2.3.0+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1