danskbert-FGN

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

• Loss: 0.6479
• F1-score: 0.8731

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	0.4105	0.8402
No log	2.0	240	0.4530	0.8506
No log	3.0	360	0.6479	0.8731
No log	4.0	480	1.1574	0.8042
0.3733	5.0	600	0.7922	0.8489
0.3733	6.0	720	0.8570	0.8598
0.3733	7.0	840	0.8900	0.8508
0.3733	8.0	960	0.9645	0.8634
0.0904	9.0	1080	1.1572	0.8294
0.0904	10.0	1200	1.0706	0.8549
0.0904	11.0	1320	1.0462	0.8675
0.0904	12.0	1440	1.0676	0.8675
0.034	13.0	1560	1.2254	0.8526
0.034	14.0	1680	1.2085	0.8636
0.034	15.0	1800	1.2398	0.8524
0.034	16.0	1920	1.2987	0.8325
0.01	17.0	2040	1.2628	0.8416
0.01	18.0	2160	1.2489	0.8526
0.01	19.0	2280	1.2504	0.8526
0.01	20.0	2400	1.2521	0.8526

Framework versions

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