bdc2024-indobart-gpt-aug

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

• Loss: 0.4480
• Accuracy: 0.9273
• Balanced Accuracy: 0.8560
• Precision: 0.9296
• Recall: 0.9273
• F1: 0.9205

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: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 6

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Balanced Accuracy	Precision	Recall	F1
No log	1.0	483	0.7053	0.7820	0.5122	0.7407	0.7820	0.7499
0.8051	2.0	966	0.5075	0.8757	0.6954	0.8779	0.8757	0.8622
0.4597	3.0	1449	0.4041	0.9197	0.8361	0.9198	0.9197	0.9122
0.2475	4.0	1932	0.4224	0.9254	0.8626	0.9255	0.9254	0.9202
0.1303	5.0	2415	0.4438	0.9273	0.8559	0.9295	0.9273	0.9214
0.0771	6.0	2898	0.4480	0.9273	0.8560	0.9296	0.9273	0.9205

Framework versions

• Transformers 4.33.1
• Pytorch 2.3.0+cu121
• Datasets 2.19.2
• Tokenizers 0.13.3