xlnet-base-cased_fold_2_binary

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

• Loss: 0.4858
• F1: 0.7648

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: linear
• num_epochs: 25

Training results

Training Loss	Epoch	Step	Validation Loss	F1
No log	1.0	290	0.4361	0.7404
0.4403	2.0	580	0.5363	0.7515
0.4403	3.0	870	0.4858	0.7648
0.2505	4.0	1160	0.7127	0.7612
0.2505	5.0	1450	0.8930	0.7554
0.1425	6.0	1740	0.9897	0.7580
0.0869	7.0	2030	1.2683	0.7615
0.0869	8.0	2320	1.4988	0.7343
0.0411	9.0	2610	1.5082	0.7492
0.0411	10.0	2900	1.4974	0.7450
0.0306	11.0	3190	1.5723	0.7435
0.0306	12.0	3480	1.8446	0.7432
0.0291	13.0	3770	1.7113	0.7639
0.0207	14.0	4060	1.8073	0.7394
0.0207	15.0	4350	1.7524	0.7585
0.0171	16.0	4640	1.8751	0.7374
0.0171	17.0	4930	1.7849	0.7561
0.0084	18.0	5220	1.8618	0.7441
0.0064	19.0	5510	1.9613	0.7437
0.0064	20.0	5800	1.8898	0.7430
0.006	21.0	6090	1.9889	0.7409
0.006	22.0	6380	1.9949	0.7488
0.0049	23.0	6670	1.9453	0.7488
0.0049	24.0	6960	1.9754	0.7472
0.002	25.0	7250	1.9946	0.7504

Framework versions

• Transformers 4.21.0
• Pytorch 1.12.0+cu113
• Datasets 2.4.0
• Tokenizers 0.12.1