roberta-large-finetuned-augmentation-LUNAR

This model is a fine-tuned version of FacebookAI/roberta-large on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.6061
• F1: 0.7909
• Roc Auc: 0.8390
• Accuracy: 0.5680

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: 2e-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: cosine
• lr_scheduler_warmup_steps: 100
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	F1	Roc Auc	Accuracy
0.4606	1.0	179	0.3928	0.5956	0.7155	0.4320
0.3171	2.0	358	0.3380	0.7156	0.7768	0.4727
0.2294	3.0	537	0.3398	0.7321	0.7927	0.5077
0.1528	4.0	716	0.3813	0.7577	0.8113	0.5175
0.0887	5.0	895	0.4250	0.7669	0.8306	0.5175
0.0583	6.0	1074	0.4355	0.7686	0.8278	0.5273
0.0448	7.0	1253	0.5045	0.7498	0.8029	0.5316
0.0298	8.0	1432	0.4862	0.7809	0.8321	0.5554
0.0227	9.0	1611	0.5282	0.7793	0.8248	0.5484
0.0111	10.0	1790	0.5567	0.7787	0.8340	0.5428
0.0082	11.0	1969	0.5762	0.7845	0.8408	0.5498
0.0055	12.0	2148	0.5771	0.7796	0.8325	0.5582
0.0032	13.0	2327	0.5884	0.7865	0.8336	0.5610
0.003	14.0	2506	0.6064	0.7901	0.8380	0.5568
0.0024	15.0	2685	0.6061	0.7909	0.8390	0.5680
0.002	16.0	2864	0.6041	0.7878	0.8399	0.5736
0.0016	17.0	3043	0.6129	0.7848	0.8346	0.5596
0.0014	18.0	3222	0.6129	0.7860	0.8366	0.5694
0.0038	19.0	3401	0.6143	0.7893	0.8400	0.5722

Framework versions

• Transformers 4.45.1
• Pytorch 2.4.0
• Datasets 3.0.1
• Tokenizers 0.20.0