bert-base-indonesian-1.5G-multilabel-indonesian-hate-speech-10

This model is a fine-tuned version of cahya/bert-base-indonesian-1.5G on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.4548
• F1: 0.7953
• Roc Auc: 0.8769
• Accuracy: 0.7246

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: 20

Training results

Training Loss	Epoch	Step	Validation Loss	F1	Roc Auc	Accuracy
0.2343	1.0	1148	0.2380	0.7112	0.8008	0.6228
0.1706	2.0	2296	0.2073	0.7634	0.8462	0.6561
0.1192	3.0	3444	0.2316	0.7759	0.8570	0.6819
0.0816	4.0	4592	0.2625	0.7702	0.8501	0.6844
0.0528	5.0	5740	0.2877	0.7826	0.8628	0.7037
0.0341	6.0	6888	0.3313	0.7713	0.8496	0.7027
0.0273	7.0	8036	0.3514	0.7791	0.8580	0.7082
0.0207	8.0	9184	0.3604	0.7765	0.8602	0.6973
0.0123	9.0	10332	0.3955	0.7811	0.8670	0.7032
0.0116	10.0	11480	0.3994	0.7895	0.8705	0.7166
0.009	11.0	12628	0.4016	0.7856	0.8642	0.7097
0.0049	12.0	13776	0.4325	0.7852	0.8729	0.7032
0.0059	13.0	14924	0.4336	0.7856	0.8699	0.7052
0.005	14.0	16072	0.4466	0.7851	0.8685	0.7176
0.0023	15.0	17220	0.4508	0.7817	0.8668	0.7047
0.0038	16.0	18368	0.4354	0.7899	0.8679	0.7201
0.004	17.0	19516	0.4529	0.7877	0.8644	0.7246
0.0017	18.0	20664	0.4548	0.7953	0.8769	0.7246
0.0019	19.0	21812	0.4532	0.7905	0.8649	0.7270
0.0011	20.0	22960	0.4557	0.7883	0.8648	0.7270

Framework versions

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