for_test13

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

• Loss: 9.0800
• Per: 0.8663
• Learning Rate: 0.0000

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: 32
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• 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
• lr_scheduler_warmup_steps: 20
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Per	Rate
9.8433	1.0417	50	10.8927	1.9730	0.0001
6.2994	2.0833	100	10.1801	1.4889	0.0001
5.6979	3.125	150	9.8748	1.1627	0.0001
5.5696	4.1667	200	9.6279	0.9856	0.0001
5.5354	5.2083	250	9.4447	0.9282	0.0001
5.3749	6.25	300	9.3013	0.8952	4e-05
5.6517	7.2917	350	9.1784	0.8771	0.0000
5.1293	8.3333	400	9.0618	0.8661	0.0000
5.5912	9.375	450	9.0800	0.8663	0.0000

Framework versions

• Transformers 4.46.2
• Pytorch 2.5.1+cu121
• Datasets 3.1.0
• Tokenizers 0.20.3