my-lilt-en-funsd

This model is a fine-tuned version of SCUT-DLVCLab/lilt-roberta-en-base on the funsd-layoutlmv3 dataset. It achieves the following results on the evaluation set:

• Loss: 1.7942
• Answer: {'precision': 0.8597914252607184, 'recall': 0.9082007343941249, 'f1': 0.8833333333333333, 'number': 817}
• Header: {'precision': 0.6666666666666666, 'recall': 0.5714285714285714, 'f1': 0.6153846153846153, 'number': 119}
• Question: {'precision': 0.9046746104491292, 'recall': 0.9164345403899722, 'f1': 0.9105166051660516, 'number': 1077}
• Overall Precision: 0.8740
• Overall Recall: 0.8927
• Overall F1: 0.8833
• Overall Accuracy: 0.8042

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: 5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• training_steps: 2500
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Answer	Header	Question	Overall Precision	Overall Recall	Overall F1	Overall Accuracy
0.1935	26.32	500	1.2125	{'precision': 0.8702830188679245, 'recall': 0.9033047735618115, 'f1': 0.8864864864864864, 'number': 817}	{'precision': 0.6296296296296297, 'recall': 0.5714285714285714, 'f1': 0.5991189427312775, 'number': 119}	{'precision': 0.8748921484037964, 'recall': 0.9415041782729805, 'f1': 0.9069767441860466, 'number': 1077}	0.8605	0.9041	0.8818	0.8024
0.0063	52.63	1000	1.4406	{'precision': 0.8732394366197183, 'recall': 0.9106487148102815, 'f1': 0.8915518274415818, 'number': 817}	{'precision': 0.632183908045977, 'recall': 0.46218487394957986, 'f1': 0.5339805825242718, 'number': 119}	{'precision': 0.8827708703374778, 'recall': 0.9229340761374187, 'f1': 0.902405810258738, 'number': 1077}	0.8683	0.8907	0.8794	0.8175
0.002	78.95	1500	1.6624	{'precision': 0.861904761904762, 'recall': 0.8861689106487148, 'f1': 0.8738684369342186, 'number': 817}	{'precision': 0.6363636363636364, 'recall': 0.5294117647058824, 'f1': 0.5779816513761468, 'number': 119}	{'precision': 0.8920863309352518, 'recall': 0.9210770659238626, 'f1': 0.9063499314755596, 'number': 1077}	0.8674	0.8838	0.8755	0.7998
0.0006	105.26	2000	1.7942	{'precision': 0.8597914252607184, 'recall': 0.9082007343941249, 'f1': 0.8833333333333333, 'number': 817}	{'precision': 0.6666666666666666, 'recall': 0.5714285714285714, 'f1': 0.6153846153846153, 'number': 119}	{'precision': 0.9046746104491292, 'recall': 0.9164345403899722, 'f1': 0.9105166051660516, 'number': 1077}	0.8740	0.8927	0.8833	0.8042
0.0002	131.58	2500	1.8161	{'precision': 0.8591385331781141, 'recall': 0.9033047735618115, 'f1': 0.8806682577565632, 'number': 817}	{'precision': 0.6346153846153846, 'recall': 0.5546218487394958, 'f1': 0.5919282511210763, 'number': 119}	{'precision': 0.9047619047619048, 'recall': 0.9173630454967502, 'f1': 0.9110189027201475, 'number': 1077}	0.8720	0.8902	0.8810	0.8021

Framework versions

• Transformers 4.26.0
• Pytorch 1.13.1+cu116
• Datasets 2.9.0
• Tokenizers 0.13.2