kavg/LiLT-RE-PT

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: 0.2199
• Recall: 0.5308
• F1: 0.3109
• Loss: 0.1355

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	Precision	Recall	F1	Validation Loss
0.1811	16.67	500	0	0	0	0.2498
0.1861	33.33	1000	0.4207	0.0457	0.0825	0.2486
0.0373	50.0	1500	0.4845	0.0517	0.0934	0.2116
0.0856	66.67	2000	0.3184	0.1504	0.2043	0.1696
0.0867	83.33	2500	0.2201	0.3472	0.2694	0.1691
0.0832	100.0	3000	0.2369	0.3685	0.2884	0.1687
0.0756	116.67	3500	0.2152	0.4301	0.2869	0.1561
0.0454	133.33	4000	0.2075	0.4818	0.2900	0.1513
0.0629	150.0	4500	0.2202	0.5282	0.3108	0.1748
0.0503	166.67	5000	0.2058	0.5573	0.3006	0.1832
0.05	183.33	5500	0.2263	0.5381	0.3186	0.1863
0.0365	200.0	6000	0.2233	0.5712	0.3211	0.1524
0.0366	216.67	6500	0.2219	0.5699	0.3195	0.1254
0.0352	233.33	7000	0.2199	0.5308	0.3109	0.1355
0.0524	250.0	7500	0.2184	0.5500	0.3126	0.1456
0.0375	266.67	8000	0.2158	0.5401	0.3084	0.1549
0.0324	283.33	8500	0.2068	0.5507	0.3007	0.1840
0.0168	300.0	9000	0.2155	0.5520	0.3100	0.1686
0.0306	316.67	9500	0.2160	0.5666	0.3128	0.1558
0.0229	333.33	10000	0.2152	0.5712	0.3126	0.1633

Framework versions

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