LiLT-SER-EN-SIN

This model is a fine-tuned version of kavg/LiLT-SER-EN on the xfun dataset. It achieves the following results on the evaluation set:

• Loss: 1.3790
• Precision: 0.7420
• Recall: 0.7759
• F1: 0.7586
• Accuracy: 0.8474

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: 5e-05
• train_batch_size: 8
• eval_batch_size: 2
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• training_steps: 10000

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
0.0089	21.74	500	0.8362	0.6606	0.7118	0.6852	0.8545
0.0013	43.48	1000	1.3605	0.7269	0.7081	0.7174	0.8230
0.0051	65.22	1500	0.9220	0.7113	0.7525	0.7313	0.8725
0.0054	86.96	2000	1.2086	0.6965	0.7291	0.7124	0.8467
0.0001	108.7	2500	1.1308	0.6843	0.7315	0.7071	0.8449
0.0001	130.43	3000	1.0934	0.7362	0.7044	0.7199	0.8606
0.0	152.17	3500	1.0390	0.7297	0.7512	0.7403	0.8590
0.0001	173.91	4000	1.1448	0.7128	0.7672	0.7390	0.8599
0.0	195.65	4500	1.1902	0.7393	0.7229	0.7310	0.8551
0.0001	217.39	5000	1.1164	0.7141	0.7783	0.7448	0.8555
0.0001	239.13	5500	1.4359	0.7197	0.7241	0.7219	0.8313
0.0	260.87	6000	1.4027	0.7256	0.7426	0.7340	0.8376
0.0	282.61	6500	1.4112	0.7085	0.7574	0.7321	0.8524
0.0	304.35	7000	1.5045	0.7627	0.7599	0.7613	0.8432
0.0	326.09	7500	1.4482	0.7390	0.7672	0.7529	0.8398
0.0	347.83	8000	1.5717	0.7155	0.7525	0.7335	0.8360
0.0	369.57	8500	1.3845	0.7348	0.7746	0.7542	0.8422
0.0	391.3	9000	1.3238	0.7283	0.7660	0.7467	0.8499
0.0	413.04	9500	1.3677	0.7321	0.7672	0.7492	0.8492
0.0001	434.78	10000	1.3790	0.7420	0.7759	0.7586	0.8474

Framework versions

• Transformers 4.39.1
• Pytorch 2.1.0+cu121
• Datasets 2.18.0
• Tokenizers 0.15.1