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

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: 0.3389
• F1: 0.7981
• Roc Auc: 0.8621
• Accuracy: 0.6672

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
0.5069	1.0	627	0.5087	0.1392	0.5511	0.1141
0.4852	2.0	1254	0.5375	0.1380	0.5503	0.1141
0.483	3.0	1881	0.4748	0.2573	0.5929	0.2402
0.4274	4.0	2508	0.4217	0.3887	0.6509	0.4134
0.4097	5.0	3135	0.3956	0.4075	0.6626	0.4505
0.3566	6.0	3762	0.3691	0.4944	0.7070	0.4916
0.344	7.0	4389	0.3530	0.5637	0.7385	0.5243
0.3145	8.0	5016	0.3265	0.6867	0.7866	0.5874
0.2944	9.0	5643	0.3415	0.6197	0.7661	0.5607
0.2168	10.0	6270	0.3160	0.7367	0.8176	0.6373
0.1664	11.0	6897	0.3014	0.7569	0.8345	0.6333
0.1604	12.0	7524	0.3070	0.7606	0.8411	0.6453
0.1616	13.0	8151	0.3060	0.7700	0.8411	0.6592
0.1155	14.0	8778	0.3160	0.7831	0.8532	0.6536
0.1226	15.0	9405	0.3307	0.7886	0.8556	0.6600
0.0968	16.0	10032	0.3346	0.7919	0.8594	0.6604
0.0939	17.0	10659	0.3389	0.7981	0.8621	0.6672
0.0632	18.0	11286	0.3417	0.7970	0.8634	0.6648
0.0816	19.0	11913	0.3438	0.7970	0.8625	0.6676
0.0759	20.0	12540	0.3446	0.7969	0.8626	0.6676

Framework versions

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