roberta-base-indonesian-522M-with-sa-william-dataset-v2

This model is a fine-tuned version of cahya/roberta-base-indonesian-522M on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.1121

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: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss
0.185	1.0	299	0.1121
0.0966	2.0	598	0.1274
0.0711	3.0	897	0.1431
0.054	4.0	1196	0.1676
0.047	5.0	1495	0.1508
0.0427	6.0	1794	0.1570
0.0343	7.0	2093	0.1229
0.029	8.0	2392	0.1430
0.0316	9.0	2691	0.1748
0.0254	10.0	2990	0.1562
0.0241	11.0	3289	0.1197
0.0222	12.0	3588	0.1190
0.0244	13.0	3887	0.1471
0.0222	14.0	4186	0.1382
0.02	15.0	4485	0.1466
0.0214	16.0	4784	0.1744
0.0186	17.0	5083	0.1457
0.0196	18.0	5382	0.1515
0.0182	19.0	5681	0.1456
0.0189	20.0	5980	0.1450

Framework versions

• Transformers 4.38.2
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2