kobert-patent-baseline

This model is a fine-tuned version of monologg/kobert on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0005
• Weighted F1: 0.8105
• Micro F1: 0.8111
• Macro F1: 0.7857
• P@1: 0.8940
• P@3: 0.9640
• P@5: 0.9788

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
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED 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: 0.1
• num_epochs: 12
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Macro F1	Micro F1	P@1	P@3	P@5	Weighted F1
0.0010	1.0	25237	0.0009	0.5947	0.6484	0.7155	0.8598	0.9090	0.6307
0.0006	2.0	50474	0.0006	0.7071	0.7419	0.8183	0.9262	0.9580	0.7371
0.0006	3.0	75711	0.0005	0.7373	0.7686	0.8476	0.9416	0.9696	0.7665
0.0005	4.0	100948	0.0005	0.7851	0.7862	0.7578	0.8660	0.9523	0.9756
0.0003	5.0	126185	0.0005	0.7889	0.7890	0.7624	0.8725	0.9559	0.9766
0.0004	6.0	151422	0.0005	0.7881	0.7896	0.7626	0.8721	0.9528	0.9736
0.0003	7.0	176659	0.0004	0.8046	0.8053	0.7793	0.8849	0.9626	0.9790
0.0003	8.0	201896	0.0004	0.8078	0.8086	0.7846	0.8890	0.9635	0.9805
0.0002	9.0	227133	0.0005	0.8080	0.8085	0.7836	0.8922	0.9634	0.9804
0.0002	10.0	252370	0.0005	0.8106	0.8117	0.7873	0.8929	0.9631	0.9797
0.0001	11.0	277607	0.0005	0.8134	0.8139	0.7882	0.8945	0.9642	0.9803
0.0001	12.0	302844	0.0005	0.8105	0.8111	0.7857	0.8940	0.9640	0.9788

Framework versions

• Transformers 5.12.1
• Pytorch 2.11.0+cu128
• Datasets 4.0.0
• Tokenizers 0.22.2