roberta-large-finetuned-augmentation-LUNAR-TAPT-MICRO

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: 0.4895
• F1: 0.8563
• Roc Auc: 0.8926
• Accuracy: 0.6522

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
0.3483	1.0	317	0.3076	0.7777	0.8325	0.5118
0.2331	2.0	634	0.2906	0.8011	0.8453	0.5513
0.1736	3.0	951	0.2906	0.8187	0.8659	0.5662
0.1174	4.0	1268	0.2952	0.8286	0.8695	0.5962
0.0857	5.0	1585	0.3265	0.8326	0.8755	0.6104
0.0574	6.0	1902	0.3470	0.8295	0.8692	0.6065
0.0455	7.0	2219	0.3953	0.8354	0.8764	0.6065
0.033	8.0	2536	0.4079	0.8328	0.8733	0.6151
0.0119	9.0	2853	0.4188	0.8468	0.8859	0.6285
0.0173	10.0	3170	0.4492	0.8476	0.8913	0.6246
0.0034	11.0	3487	0.4630	0.8488	0.8916	0.6230
0.0035	12.0	3804	0.4759	0.8531	0.8939	0.6341
0.0046	13.0	4121	0.4858	0.8487	0.8874	0.6293
0.0076	14.0	4438	0.4798	0.8542	0.8926	0.6427
0.0036	15.0	4755	0.4899	0.8512	0.8888	0.6356
0.0008	16.0	5072	0.4882	0.8543	0.8925	0.6443
0.0016	17.0	5389	0.4895	0.8563	0.8926	0.6522
0.0008	18.0	5706	0.4894	0.8561	0.8934	0.6498
0.0006	19.0	6023	0.4905	0.8550	0.8930	0.6475
0.0014	20.0	6340	0.4903	0.8555	0.8933	0.6483

Framework versions

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