roberta-unfair

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

• Loss: 0.1969
• Accuracy: 0.9679
• F1 Macro: 0.9620
• Precision Macro: 0.9559
• Recall Macro: 0.9688
• Recall Unfair: 0.9710

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: 2e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 20
• 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: 3

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1 Macro	Precision Macro	Recall Macro	Recall Unfair
0.1406	0.4237	200	0.2180	0.9390	0.9284	0.9197	0.9389	0.9384
0.11	0.8475	400	0.1185	0.9690	0.9632	0.9582	0.9685	0.9674
0.044	1.2712	600	0.1787	0.9615	0.9544	0.9484	0.9611	0.9601
0.0487	1.6949	800	0.1957	0.9583	0.9507	0.9437	0.9588	0.9601
0.0222	2.1186	1000	0.1170	0.9754	0.9707	0.9674	0.9741	0.9710
0.0139	2.5424	1200	0.2559	0.9594	0.9522	0.9434	0.9627	0.9710
0.0165	2.9661	1400	0.1969	0.9679	0.9620	0.9559	0.9688	0.9710

Framework versions

• Transformers 4.55.1
• Pytorch 2.6.0+cu124
• Datasets 4.0.0
• Tokenizers 0.21.4