bert-base-uncased-multilabel-indonesian-hate-speech-new

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

• Loss: 0.3500
• F1: 0.7366
• Roc Auc: 0.8382
• Accuracy: 0.6638

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: 15

Training results

Training Loss	Epoch	Step	Validation Loss	F1	Roc Auc	Accuracy
0.3625	1.0	659	0.2901	0.6377	0.7723	0.5154
0.2824	2.0	1318	0.2594	0.6698	0.7828	0.5623
0.2386	3.0	1977	0.2546	0.6983	0.8090	0.5957
0.1788	4.0	2636	0.2620	0.7112	0.8102	0.6443
0.1428	5.0	3295	0.2599	0.7223	0.8187	0.6465
0.1324	6.0	3954	0.2812	0.7172	0.8204	0.6384
0.0897	7.0	4613	0.3083	0.7215	0.8208	0.6610
0.0699	8.0	5272	0.3303	0.7140	0.8138	0.6558
0.0653	9.0	5931	0.3354	0.7289	0.8242	0.6566
0.044	10.0	6590	0.3500	0.7366	0.8382	0.6638
0.035	11.0	7249	0.3609	0.7289	0.8292	0.6688
0.0321	12.0	7908	0.3724	0.7324	0.8388	0.6589
0.0228	13.0	8567	0.3798	0.7361	0.8375	0.6680
0.0212	14.0	9226	0.3874	0.7301	0.8344	0.6653
0.018	15.0	9885	0.3888	0.7297	0.8318	0.6695

Framework versions

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