polibert-malaysia-ver2

This model is a fine-tuned version of bert-base-uncased on tnwei/ms-newspapers dataset. And this model is the 2nd version of YagiASAFAS/polibert-malaysia It achieves the following results on the evaluation set:

• Loss: 0.5548
• Accuracy: 0.9459

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: 5e-05
• train_batch_size: 8
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 16
• mixed_precision_training: Native AMP

Label Mappings

• 0: Economic Concerns
• 1: Racial discrimination or polarization
• 2: Leadership weaknesses
• 3: Development and infrastructure gaps
• 4: Corruption
• 5: Political instablility
• 6: Socials and Public safety
• 7: Administration
• 8: Education
• 9: Religion issues
• 10: Environmental

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.5263	1.0	1127	0.4196	0.9024
0.3028	2.0	2254	0.3746	0.9330
0.2412	3.0	3381	0.3870	0.9326
0.1425	4.0	4508	0.3688	0.9397
0.1052	5.0	5635	0.3860	0.9454
0.0621	6.0	6762	0.4542	0.9441
0.0533	7.0	7889	0.4923	0.9392
0.0383	8.0	9016	0.4893	0.9437
0.0245	9.0	10143	0.4658	0.9445
0.0099	10.0	11270	0.5429	0.9392
0.0107	11.0	12397	0.5551	0.9450
0.0044	12.0	13524	0.5579	0.9441
0.0027	13.0	14651	0.6010	0.9419
0.0059	14.0	15778	0.5880	0.9445
0.0038	15.0	16905	0.5475	0.9459
0.0001	16.0	18032	0.5548	0.9459

Framework versions

• Transformers 4.18.0
• Pytorch 2.3.0+cu121
• Datasets 2.19.1
• Tokenizers 0.12.1