kavg/LiLT-RE-ZH

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: 0.3911
• Recall: 0.6703
• F1: 0.4940
• Loss: 0.1352

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: 6
• 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
0.1469	20.83	500	0	0.1467	0	0
0.0896	41.67	1000	0.0837	0.1454	0.2946	0.0487
0.1027	62.5	1500	0.1225	0.1353	0.3333	0.0750
0.0485	83.33	2000	0.3536	0.1571	0.3364	0.3727
0.0597	104.17	2500	0.4448	0.1546	0.3535	0.5997
0.0367	125.0	3000	0.4940	0.1352	0.3911	0.6703
0.033	145.83	3500	0.4977	0.1749	0.3902	0.6870
0.0176	166.67	4000	0.5087	0.2262	0.4034	0.6883
0.0123	187.5	4500	0.5050	0.2358	0.3978	0.6915
0.0194	208.33	5000	0.5173	0.2976	0.4090	0.7037
0.0118	171.88	5500	0.4159	0.6863	0.5179	0.2836
0.0054	187.5	6000	0.4356	0.6703	0.5280	0.3100
0.01	203.12	6500	0.4229	0.6979	0.5266	0.3430
0.0062	218.75	7000	0.4272	0.7062	0.5324	0.3652
0.0051	234.38	7500	0.4306	0.6947	0.5317	0.3496
0.0048	250.0	8000	0.4400	0.6940	0.5386	0.3943
0.0087	265.62	8500	0.4290	0.6992	0.5317	0.3782
0.0077	281.25	9000	0.4394	0.7049	0.5414	0.3855
0.0014	296.88	9500	0.4363	0.7004	0.5377	0.3933
0.0035	312.5	10000	0.4350	0.6992	0.5363	0.4045

Framework versions

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