nb-bert-FGN

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

• Loss: 0.8455
• F1-score: 0.8640

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: 8
• 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	F1-score
No log	1.0	120	0.4694	0.8310
No log	2.0	240	0.4810	0.8225
No log	3.0	360	0.3942	0.8528
No log	4.0	480	0.7082	0.7709
0.4938	5.0	600	0.7041	0.8333
0.4938	6.0	720	0.6616	0.8528
0.4938	7.0	840	0.9447	0.8226
0.4938	8.0	960	0.8971	0.8464
0.2424	9.0	1080	0.9245	0.8348
0.2424	10.0	1200	0.8455	0.8640
0.2424	11.0	1320	0.8109	0.8571
0.2424	12.0	1440	1.0194	0.8566
0.1235	13.0	1560	0.9609	0.8533
0.1235	14.0	1680	1.0777	0.8435
0.1235	15.0	1800	1.1128	0.8450
0.1235	16.0	1920	1.0391	0.8582
0.0621	17.0	2040	1.1569	0.8507
0.0621	18.0	2160	1.1449	0.8492
0.0621	19.0	2280	1.1715	0.8492
0.0621	20.0	2400	1.1702	0.8564

Framework versions

• Transformers 4.41.1
• Pytorch 2.3.0+cu121
• Datasets 2.19.2
• Tokenizers 0.19.1