indobert-base-p1-multilabel-indonesian-hate-speech-new-50

This model is a fine-tuned version of indobenchmark/indobert-base-p1 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.4100
• F1: 0.7860
• Roc Auc: 0.8751
• Accuracy: 0.7225

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: 8
• eval_batch_size: 8
• seed: 42
• 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: 50

Training results

Training Loss	Epoch	Step	Validation Loss	F1	Roc Auc	Accuracy
0.2914	1.0	659	0.2262	0.7256	0.8200	0.5943
0.2012	2.0	1318	0.2007	0.7624	0.8427	0.6465
0.1543	3.0	1977	0.2230	0.7629	0.8422	0.6889
0.0761	4.0	2636	0.2460	0.7781	0.8668	0.7018
0.0508	5.0	3295	0.2869	0.7688	0.8609	0.6946
0.0463	6.0	3954	0.2881	0.7758	0.8585	0.7118
0.0244	7.0	4613	0.3081	0.7784	0.8600	0.7160
0.018	8.0	5272	0.3489	0.7799	0.8754	0.7052
0.0184	9.0	5931	0.3453	0.7830	0.8688	0.7174
0.014	10.0	6590	0.3531	0.7748	0.8732	0.6898
0.011	11.0	7249	0.3941	0.7718	0.8508	0.7166
0.012	12.0	7908	0.3781	0.7849	0.8680	0.7225
0.0081	13.0	8567	0.4381	0.7695	0.8775	0.6889
0.0077	14.0	9226	0.4100	0.7860	0.8751	0.7225
0.0079	15.0	9885	0.4354	0.7779	0.8641	0.7122
0.0083	16.0	10544	0.4287	0.7823	0.8730	0.7166
0.0072	17.0	11203	0.4874	0.7623	0.8697	0.6883
0.0073	18.0	11862	0.4516	0.7803	0.8789	0.7077
0.0104	19.0	12521	0.4467	0.7737	0.8636	0.7058

Framework versions

• Transformers 4.51.3
• Pytorch 2.7.0+cu128
• Datasets 3.6.0
• Tokenizers 0.21.1