kavg/LiLT-RE-DE

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: 0.3054
• Recall: 0.6032
• F1: 0.4055
• Loss: 0.2164

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
0.1914	20.83	500	0	0	0	0.2039
0.1638	41.67	1000	0.4688	0.0252	0.0478	0.2196
0.0928	62.5	1500	0.3790	0.1669	0.2318	0.2127
0.0948	83.33	2000	0.3125	0.4245	0.3600	0.2987
0.0796	104.17	2500	0.3102	0.5587	0.3989	0.3636
0.0469	125.0	3000	0.3204	0.5134	0.3946	0.3587
0.0471	145.83	3500	0.3303	0.5243	0.4053	0.2792
0.0486	166.67	4000	0.2967	0.5973	0.3964	0.2973
0.0381	187.5	4500	0.3066	0.6007	0.4060	0.3003
0.0392	208.33	5000	0.3054	0.6032	0.4055	0.2164
0.0268	229.17	5500	0.3052	0.6158	0.4081	0.3159
0.029	250.0	6000	0.2850	0.6292	0.3923	0.3108
0.0217	270.83	6500	0.2964	0.6141	0.3998	0.3130
0.0241	291.67	7000	0.3012	0.6216	0.4058	0.3197
0.038	312.5	7500	0.3051	0.6216	0.4093	0.2627
0.0374	333.33	8000	0.2914	0.6359	0.3997	0.3388
0.0194	354.17	8500	0.2975	0.6275	0.4037	0.3155
0.0189	375.0	9000	0.3037	0.625	0.4088	0.2911
0.0147	395.83	9500	0.2993	0.6242	0.4046	0.3417
0.0328	416.67	10000	0.3012	0.6242	0.4063	0.3210

Framework versions

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