KoModernBERT-large-mlm-v3

This model is a fine-tuned version of CocoRoF/KoModernBERT-large-mlm-v2 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 2.3605

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: 1e-06
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• gradient_accumulation_steps: 32
• total_train_batch_size: 1024
• total_eval_batch_size: 32
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.98) and epsilon=1e-07 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 1.0

Training results

Training Loss	Epoch	Step	Validation Loss
97.07	0.0599	500	3.0658
97.6329	0.1199	1000	3.0890
94.2036	0.1798	1500	2.9848
93.88	0.2397	2000	2.8945
91.2063	0.2997	2500	2.8105
88.1768	0.3596	3000	2.7460
86.8633	0.4196	3500	2.6891
84.4198	0.4795	4000	2.6377
81.6633	0.5394	4500	2.5888
81.2498	0.5994	5000	2.5427
79.082	0.6593	5500	2.5001
78.3493	0.7192	6000	2.4693
77.7304	0.7792	6500	2.4339
77.1332	0.8391	7000	2.4003
75.6454	0.8990	7500	2.3795
74.7334	0.9590	8000	2.3605

Framework versions

• Transformers 4.49.0
• Pytorch 2.5.1+cu124
• Datasets 3.3.2
• Tokenizers 0.21.1