korean_kws2

This model is a fine-tuned version of Kkonjeong/wav2vec2-base-korean on the None dataset. It achieves the following results on the evaluation set:

• Loss: nan
• Accuracy: 0.1053

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: 3e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Use adamw_torch_fused with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	1.0	2	1.9242	0.3158
No log	2.0	4	1.9025	0.3684
No log	3.0	6	nan	0.1053
No log	4.0	8	nan	0.1053
No log	5.0	10	nan	0.1053
No log	6.0	12	nan	0.1053
No log	7.0	14	nan	0.1053
No log	8.0	16	nan	0.1053
No log	9.0	18	nan	0.1053
No log	10.0	20	nan	0.1053
No log	11.0	22	nan	0.1053
No log	12.0	24	nan	0.1053
No log	13.0	26	nan	0.1053
No log	14.0	28	nan	0.1053
No log	15.0	30	nan	0.1053
No log	16.0	32	nan	0.1053
No log	17.0	34	nan	0.1053
No log	18.0	36	nan	0.1053
No log	19.0	38	nan	0.1053
No log	20.0	40	nan	0.1053
No log	21.0	42	nan	0.1053
No log	22.0	44	nan	0.1053
No log	23.0	46	nan	0.1053
No log	24.0	48	nan	0.1053
No log	25.0	50	nan	0.1053
No log	26.0	52	nan	0.1053
No log	27.0	54	nan	0.1053
No log	28.0	56	nan	0.1053
No log	29.0	58	nan	0.1053
No log	30.0	60	nan	0.1053

Framework versions

• Transformers 4.57.6
• Pytorch 2.10.0
• Datasets 3.0.0
• Tokenizers 0.22.2