roberta-large-finetuned-augmentation-LUNAR-TAPT-DAIR

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: 0.3531
• F1: 0.8753
• Roc Auc: 0.9173
• Accuracy: 0.7721

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
0.2394	1.0	627	0.2378	0.7928	0.8559	0.6579
0.2141	2.0	1254	0.2185	0.8216	0.8747	0.6850
0.1336	3.0	1881	0.2149	0.8388	0.8913	0.7150
0.1087	4.0	2508	0.2141	0.8479	0.8978	0.7357
0.0701	5.0	3135	0.2390	0.8499	0.8992	0.7361
0.065	6.0	3762	0.2600	0.8483	0.9022	0.7377
0.0518	7.0	4389	0.2756	0.8593	0.9038	0.7425
0.0307	8.0	5016	0.2917	0.8598	0.9050	0.7497
0.0227	9.0	5643	0.3293	0.8559	0.9062	0.7433
0.0131	10.0	6270	0.3357	0.8598	0.9005	0.7577
0.0068	11.0	6897	0.3465	0.8654	0.9109	0.7569
0.0014	12.0	7524	0.3500	0.8711	0.9116	0.7641
0.0009	13.0	8151	0.3531	0.8753	0.9173	0.7721
0.0013	14.0	8778	0.3676	0.8718	0.9188	0.7681
0.0004	15.0	9405	0.3715	0.8748	0.9165	0.7725
0.0012	16.0	10032	0.3709	0.8745	0.9172	0.7737
0.0008	17.0	10659	0.3775	0.8734	0.9182	0.7717

Framework versions

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