MeMo_BERT-WSD-ScandiBERT

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

• Loss: 2.0491
• F1-score: 0.4001

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	61	1.3760	0.1229
No log	2.0	122	1.4064	0.1229
No log	3.0	183	1.4753	0.1229
No log	4.0	244	1.3255	0.1229
No log	5.0	305	1.3649	0.1229
No log	6.0	366	1.3784	0.1229
No log	7.0	427	1.3312	0.2191
No log	8.0	488	1.3717	0.1229
1.2808	9.0	549	1.5063	0.1229
1.2808	10.0	610	1.2576	0.3721
1.2808	11.0	671	1.4630	0.3622
1.2808	12.0	732	1.7778	0.3258
1.2808	13.0	793	2.2233	0.3402
1.2808	14.0	854	2.1684	0.3619
1.2808	15.0	915	2.0034	0.3668
1.2808	16.0	976	1.8374	0.3818
0.9251	17.0	1037	1.9104	0.3956
0.9251	18.0	1098	2.0491	0.4001
0.9251	19.0	1159	2.1429	0.3534
0.9251	20.0	1220	2.1352	0.3648

Framework versions

• Transformers 4.38.2
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2