kavg/LiLT-RE-EN-SIN

checkpoints

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

• Precision: 0.4702
• Recall: 0.5581
• F1: 0.5104
• Loss: 0.5301

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: 1e-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
• lr_scheduler_warmup_ratio: 0.1
• training_steps: 10000

Training results

Training Loss	Epoch	Step	F1	Validation Loss	Precision	Recall
0.0661	41.67	500	0.3769	0.2542	0.3360	0.4293
0.0547	83.33	1000	0.4404	0.2624	0.3805	0.5227
0.0671	125.0	1500	0.4580	0.2623	0.4031	0.5303
0.0284	166.67	2000	0.4764	0.3800	0.4142	0.5606
0.0177	208.33	2500	0.4883	0.3349	0.4371	0.5530
0.0164	250.0	3000	0.4926	0.3123	0.4491	0.5455
0.0081	291.67	3500	0.4966	0.3830	0.4458	0.5606
0.0067	333.33	4000	0.4916	0.3459	0.4424	0.5530
0.0048	375.0	4500	0.4989	0.4200	0.4527	0.5556
0.0112	416.67	5000	0.5158	0.4377	0.4672	0.5758
0.0052	458.33	5500	0.5085	0.4983	0.4619	0.5657
0.0023	500.0	6000	0.5086	0.5621	0.4654	0.5606
0.0022	541.67	6500	0.5074	0.5063	0.4635	0.5606
0.0083	583.33	7000	0.5109	0.5471	0.4693	0.5606
0.0023	625.0	7500	0.5028	0.5268	0.4559	0.5606
0.0027	666.67	8000	0.5098	0.5385	0.4674	0.5606
0.0078	708.33	8500	0.4581	0.5657	0.5062	0.5135
0.0015	750.0	9000	0.4702	0.5581	0.5104	0.5301
0.0043	791.67	9500	0.4595	0.5581	0.5040	0.5684
0.0007	833.33	10000	0.4587	0.5606	0.5045	0.5711

Framework versions

• Transformers 4.38.2
• Pytorch 2.1.0+cu121
• Datasets 2.18.0
• Tokenizers 0.15.1