layoutlm-funsd

This model is a fine-tuned version of microsoft/layoutlm-base-uncased on the funsd dataset. It achieves the following results on the evaluation set:

• Loss: 0.7047
• Answer: {'precision': 0.7122222222222222, 'recall': 0.792336217552534, 'f1': 0.7501462843768285, 'number': 809}
• Header: {'precision': 0.325, 'recall': 0.3277310924369748, 'f1': 0.3263598326359833, 'number': 119}
• Question: {'precision': 0.7794508414526129, 'recall': 0.8262910798122066, 'f1': 0.8021877848678213, 'number': 1065}
• Overall Precision: 0.7259
• Overall Recall: 0.7827
• Overall F1: 0.7533
• Overall Accuracy: 0.8068

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: 3e-05
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 15
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Answer	Header	Question	Overall Precision	Overall Recall	Overall F1	Overall Accuracy
1.8367	1.0	10	1.6199	{'precision': 0.017991004497751123, 'recall': 0.014833127317676144, 'f1': 0.016260162601626018, 'number': 809}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 119}	{'precision': 0.21987951807228914, 'recall': 0.13708920187793427, 'f1': 0.16888374783111626, 'number': 1065}	0.1187	0.0793	0.0951	0.3426
1.4807	2.0	20	1.2596	{'precision': 0.18181818181818182, 'recall': 0.20519159456118666, 'f1': 0.1927990708478513, 'number': 809}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 119}	{'precision': 0.46142208774583965, 'recall': 0.5727699530516432, 'f1': 0.5111018014243821, 'number': 1065}	0.3472	0.3894	0.3671	0.5904
1.1183	3.0	30	0.9406	{'precision': 0.4608967674661105, 'recall': 0.546353522867738, 'f1': 0.5, 'number': 809}	{'precision': 0.03225806451612903, 'recall': 0.008403361344537815, 'f1': 0.013333333333333332, 'number': 119}	{'precision': 0.56973293768546, 'recall': 0.7211267605633803, 'f1': 0.6365520099461252, 'number': 1065}	0.5180	0.6076	0.5592	0.7099
0.8592	4.0	40	0.7881	{'precision': 0.5798237022526934, 'recall': 0.7317676143386898, 'f1': 0.6469945355191257, 'number': 809}	{'precision': 0.10714285714285714, 'recall': 0.05042016806722689, 'f1': 0.06857142857142856, 'number': 119}	{'precision': 0.6663865546218487, 'recall': 0.7446009389671362, 'f1': 0.7033259423503326, 'number': 1065}	0.6136	0.6979	0.6531	0.7604
0.6864	5.0	50	0.7305	{'precision': 0.6115261472785486, 'recall': 0.7082818294190358, 'f1': 0.6563573883161512, 'number': 809}	{'precision': 0.23529411764705882, 'recall': 0.16806722689075632, 'f1': 0.19607843137254902, 'number': 119}	{'precision': 0.6808510638297872, 'recall': 0.8112676056338028, 'f1': 0.7403598971722366, 'number': 1065}	0.6360	0.7311	0.6802	0.7749
0.584	6.0	60	0.6955	{'precision': 0.6358024691358025, 'recall': 0.7639060568603214, 'f1': 0.6939921392476137, 'number': 809}	{'precision': 0.28205128205128205, 'recall': 0.18487394957983194, 'f1': 0.2233502538071066, 'number': 119}	{'precision': 0.7248495270851246, 'recall': 0.7915492957746478, 'f1': 0.7567324955116697, 'number': 1065}	0.6701	0.7441	0.7052	0.7810
0.5083	7.0	70	0.6726	{'precision': 0.6795698924731183, 'recall': 0.7812113720642769, 'f1': 0.7268545140885566, 'number': 809}	{'precision': 0.27, 'recall': 0.226890756302521, 'f1': 0.24657534246575347, 'number': 119}	{'precision': 0.7439236111111112, 'recall': 0.8046948356807512, 'f1': 0.7731168245376635, 'number': 1065}	0.6948	0.7607	0.7262	0.7933
0.4552	8.0	80	0.6811	{'precision': 0.6750788643533123, 'recall': 0.7935723114956736, 'f1': 0.7295454545454545, 'number': 809}	{'precision': 0.23214285714285715, 'recall': 0.2184873949579832, 'f1': 0.22510822510822512, 'number': 119}	{'precision': 0.7482817869415808, 'recall': 0.8178403755868544, 'f1': 0.781516375056079, 'number': 1065}	0.6911	0.7722	0.7294	0.7959
0.4053	9.0	90	0.6773	{'precision': 0.7058177826564215, 'recall': 0.7948084054388134, 'f1': 0.7476744186046511, 'number': 809}	{'precision': 0.27450980392156865, 'recall': 0.23529411764705882, 'f1': 0.2533936651583711, 'number': 119}	{'precision': 0.7614840989399293, 'recall': 0.8093896713615023, 'f1': 0.7847064178425124, 'number': 1065}	0.7147	0.7692	0.7409	0.7999
0.3938	10.0	100	0.6783	{'precision': 0.6976998904709748, 'recall': 0.7873918417799752, 'f1': 0.7398373983739838, 'number': 809}	{'precision': 0.2894736842105263, 'recall': 0.2773109243697479, 'f1': 0.2832618025751073, 'number': 119}	{'precision': 0.7643979057591623, 'recall': 0.8225352112676056, 'f1': 0.7924016282225238, 'number': 1065}	0.7115	0.7757	0.7422	0.8007
0.3377	11.0	110	0.6881	{'precision': 0.7136465324384788, 'recall': 0.788627935723115, 'f1': 0.7492660011743981, 'number': 809}	{'precision': 0.3103448275862069, 'recall': 0.3025210084033613, 'f1': 0.30638297872340425, 'number': 119}	{'precision': 0.7664359861591695, 'recall': 0.831924882629108, 'f1': 0.7978388113462405, 'number': 1065}	0.7202	0.7827	0.7502	0.8033
0.3211	12.0	120	0.6958	{'precision': 0.7075575027382256, 'recall': 0.7985166872682324, 'f1': 0.7502903600464577, 'number': 809}	{'precision': 0.31092436974789917, 'recall': 0.31092436974789917, 'f1': 0.31092436974789917, 'number': 119}	{'precision': 0.7762923351158645, 'recall': 0.8178403755868544, 'f1': 0.79652491998171, 'number': 1065}	0.7214	0.7797	0.7495	0.8048
0.3036	13.0	130	0.7008	{'precision': 0.7138121546961326, 'recall': 0.7985166872682324, 'f1': 0.7537922987164527, 'number': 809}	{'precision': 0.32727272727272727, 'recall': 0.3025210084033613, 'f1': 0.314410480349345, 'number': 119}	{'precision': 0.7775800711743772, 'recall': 0.8206572769953052, 'f1': 0.7985381452718137, 'number': 1065}	0.7274	0.7807	0.7531	0.8049
0.2798	14.0	140	0.7025	{'precision': 0.7131696428571429, 'recall': 0.7898640296662547, 'f1': 0.7495601173020529, 'number': 809}	{'precision': 0.3274336283185841, 'recall': 0.31092436974789917, 'f1': 0.3189655172413793, 'number': 119}	{'precision': 0.7727272727272727, 'recall': 0.8300469483568075, 'f1': 0.8003621548211861, 'number': 1065}	0.7246	0.7827	0.7525	0.8066
0.279	15.0	150	0.7047	{'precision': 0.7122222222222222, 'recall': 0.792336217552534, 'f1': 0.7501462843768285, 'number': 809}	{'precision': 0.325, 'recall': 0.3277310924369748, 'f1': 0.3263598326359833, 'number': 119}	{'precision': 0.7794508414526129, 'recall': 0.8262910798122066, 'f1': 0.8021877848678213, 'number': 1065}	0.7259	0.7827	0.7533	0.8068

Framework versions

• Transformers 4.44.2
• Pytorch 2.5.0+cu121
• Datasets 3.1.0
• Tokenizers 0.19.1