Kemenkeu-Sentiment-Classifier

This model is a fine-tuned version of indobenchmark/indobert-base-p1 on the MoF-DAC Mini Challenge#1 dataset. It achieves the following results on the evaluation set:

• Accuracy: 0.66
• F1: 0.6368

Leaderboard score:

• Public score: 0.63733
• Private score: 0.65733

Model description & limitations

• This model can be used to classify text with four possible outputs [netral, tdk-relevan, negatif, and positif]
• only for specific cases related to the Ministry Of Finance Indonesia

How to use

You can use this model directly with a pipeline

pretrained_name = "hanifnoerr/Kemenkeu-Sentiment-Classifier"
class_model = pipeline(tokenizer=pretrained_name, model=pretrained_name)

test_data = "Mengawal APBN, Indonesia Maju"
class_model(test_data)

Training and evaluation data

The following hyperparameters were used during training:

• learning_rate: 1e-05
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
1.0131	1.0	500	0.8590	0.644	0.5964
0.7133	2.0	1000	0.8639	0.63	0.5924
0.5261	3.0	1500	0.9002	0.66	0.6368

Framework versions

• Transformers 4.27.4
• Pytorch 2.0.0+cu118
• Datasets 2.11.0
• Tokenizers 0.13.3