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: 0.5291
• Accuracy: 0.8063

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 OptimizerNames.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
2.8542	0.0681	100	1.4936	0.3803
1.9555	0.1362	200	1.1916	0.5237
2.3132	0.2043	300	0.9826	0.6180
1.8724	0.2724	400	0.9512	0.6411
1.9331	0.3405	500	0.8138	0.6857
1.6761	0.4086	600	0.7749	0.6932
1.7902	0.4767	700	0.7694	0.7028
1.9041	0.5448	800	0.7199	0.7194
1.8659	0.6129	900	0.7010	0.7382
1.9123	0.6810	1000	0.6067	0.7753
1.2564	0.7491	1100	0.6073	0.7726
0.8368	0.8172	1200	0.6203	0.7729
1.1841	0.8853	1300	0.5312	0.7988
1.0372	0.9534	1400	0.5291	0.8063

Framework versions

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