xlnet-large-cased-finetuned-augmentation-LUNAR-TAPT

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

• Loss: 0.4904
• F1: 0.8291
• Roc Auc: 0.8646
• Accuracy: 0.6215

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
0.3732	1.0	318	0.3677	0.6417	0.7249	0.4227
0.291	2.0	636	0.2986	0.7666	0.8238	0.5426
0.2296	3.0	954	0.2937	0.7774	0.8291	0.5552
0.1332	4.0	1272	0.3269	0.7980	0.8559	0.5797
0.0964	5.0	1590	0.3768	0.7977	0.8473	0.5505
0.0618	6.0	1908	0.4196	0.7833	0.8416	0.5552
0.0356	7.0	2226	0.4305	0.8041	0.8509	0.5726
0.0214	8.0	2544	0.4510	0.8112	0.8482	0.5883
0.0196	9.0	2862	0.4708	0.8118	0.8582	0.5970
0.0111	10.0	3180	0.4950	0.8174	0.8590	0.5994
0.0124	11.0	3498	0.5083	0.8094	0.8572	0.5852
0.0079	12.0	3816	0.4904	0.8291	0.8646	0.6215
0.0062	13.0	4134	0.5218	0.8155	0.8578	0.5954
0.001	14.0	4452	0.5225	0.8194	0.8636	0.6073
0.0024	15.0	4770	0.5248	0.8244	0.8646	0.6088
0.0012	16.0	5088	0.5259	0.8235	0.8652	0.6073

Framework versions

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