kavg/LiLT-RE-IT

checkpoints

This model is a fine-tuned version of nielsr/lilt-xlm-roberta-base on the xfun dataset. It achieves the following results on the evaluation set:

• Precision: 0.4301
• Recall: 0.7110
• F1: 0.5360
• Loss: 0.1529

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
0.1937	14.71	500	0	0.3651	0	0
0.1605	29.41	1000	0.2033	0.3287	0.4497	0.1314
0.1037	44.12	1500	0.4002	0.3697	0.4064	0.3941
0.0952	58.82	2000	0.4622	0.4930	0.3722	0.6097
0.0503	73.53	2500	0.4886	0.6168	0.3840	0.6716
0.0657	88.24	3000	0.4949	0.3243	0.3857	0.6901
0.0262	102.94	3500	0.5050	0.2840	0.4005	0.6832
0.0238	117.65	4000	0.5205	0.4294	0.4156	0.6963
0.0258	132.35	4500	0.5198	0.0871	0.4183	0.6862
0.0136	147.06	5000	0.5216	0.1642	0.4143	0.7040
0.0259	161.76	5500	0.5270	0.3042	0.4223	0.7009
0.0107	176.47	6000	0.5261	0.2665	0.4208	0.7017
0.0074	191.18	6500	0.5345	0.2884	0.4258	0.7179
0.0105	205.88	7000	0.5429	0.2051	0.4414	0.7048
0.0079	220.59	7500	0.4348	0.7063	0.5383	0.3553
0.0075	235.29	8000	0.4301	0.7110	0.5360	0.1529
0.0026	250.0	8500	0.4263	0.7264	0.5373	0.4010
0.001	264.71	9000	0.4352	0.7133	0.5406	0.2411
0.003	279.41	9500	0.4298	0.7141	0.5366	0.2696
0.0118	294.12	10000	0.4284	0.7172	0.5364	0.2205

Framework versions

• Transformers 4.35.2
• Pytorch 2.1.0+cu121
• Datasets 2.16.1
• Tokenizers 0.15.1