korean_sentiment_analysis_dataset3

This model is a fine-tuned version of klue/roberta-large on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.7614
• Micro f1 score: 74.9024
• Auprc: 75.3897
• Accuracy: 0.7490

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: 1e-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
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 10.0

Training results

Training Loss	Epoch	Step	Validation Loss	Micro f1 score	Auprc	Accuracy
0.8814	1.0	5080	0.7160	74.0798	78.1155	0.7408
0.5153	2.0	10160	0.6638	76.1573	80.2492	0.7616
0.6463	3.0	15240	0.6815	76.2897	80.6829	0.7629
0.4697	4.0	20320	0.7243	76.0666	80.1682	0.7607
0.2043	5.0	25400	0.9200	75.4810	79.2632	0.7548
0.2452	6.0	30480	1.0875	74.9582	78.5166	0.7496
0.1481	7.0	35560	1.3625	74.7769	76.5613	0.7478
0.1974	8.0	40640	1.5593	75.0906	76.3100	0.7509
0.1658	9.0	45720	1.6836	74.9651	75.6953	0.7497
0.1392	10.0	50800	1.7614	74.9024	75.3897	0.7490

Framework versions

• Transformers 4.25.1
• Pytorch 1.6.0
• Datasets 2.7.1
• Tokenizers 0.13.2