wav2vec2-xlsr-korean-dialect-recognition

This model is a fine-tuned version of fleek/wav2vec-large-xlsr-korean on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0752
• Accuracy: 0.9783

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: 0.0001
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 8
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 1
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.8264	0.0794	500	0.3492	0.8641
0.674	0.1588	1000	0.2810	0.8985
0.3338	0.2382	1500	0.2596	0.9269
0.3121	0.3176	2000	0.2037	0.9403
0.2074	0.3970	2500	0.1472	0.9494
0.4901	0.4764	3000	0.1448	0.9582
0.2544	0.5558	3500	0.1676	0.9535
0.2138	0.6352	4000	0.1057	0.9684
0.1705	0.7146	4500	0.1463	0.9551
0.4207	0.7940	5000	0.0907	0.9722
0.0229	0.8734	5500	0.0887	0.9738
0.203	0.9528	6000	0.0752	0.9783

Framework versions

• Transformers 4.47.1
• Pytorch 2.5.1+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0