layoutlm-funsd

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

• Loss: 0.6857
• Answer: {'precision': 0.7176981541802389, 'recall': 0.8170580964153276, 'f1': 0.7641618497109827, 'number': 809}
• Header: {'precision': 0.28368794326241137, 'recall': 0.33613445378151263, 'f1': 0.3076923076923077, 'number': 119}
• Question: {'precision': 0.7773820124666073, 'recall': 0.819718309859155, 'f1': 0.7979890310786105, 'number': 1065}
• Overall Precision: 0.7204
• Overall Recall: 0.7898
• Overall F1: 0.7535
• Overall Accuracy: 0.8139

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
1.8064	1.0	10	1.6080	{'precision': 0.020618556701030927, 'recall': 0.012360939431396786, 'f1': 0.01545595054095827, 'number': 809}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 119}	{'precision': 0.2702127659574468, 'recall': 0.11924882629107982, 'f1': 0.16547231270358306, 'number': 1065}	0.1435	0.0687	0.0929	0.3378
1.4826	2.0	20	1.2520	{'precision': 0.20166320166320167, 'recall': 0.23980222496909764, 'f1': 0.21908526256352345, 'number': 809}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 119}	{'precision': 0.4309507286606523, 'recall': 0.5830985915492958, 'f1': 0.49561053471667993, 'number': 1065}	0.3392	0.4089	0.3708	0.5993
1.1438	3.0	30	0.9584	{'precision': 0.463519313304721, 'recall': 0.5339925834363412, 'f1': 0.49626651349798967, 'number': 809}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 119}	{'precision': 0.6199664429530202, 'recall': 0.6938967136150235, 'f1': 0.6548515728843598, 'number': 1065}	0.5492	0.5876	0.5678	0.6897
0.8546	4.0	40	0.7900	{'precision': 0.5885714285714285, 'recall': 0.7639060568603214, 'f1': 0.6648735879505111, 'number': 809}	{'precision': 0.06666666666666667, 'recall': 0.025210084033613446, 'f1': 0.036585365853658534, 'number': 119}	{'precision': 0.6505823627287853, 'recall': 0.7342723004694836, 'f1': 0.6898985443317159, 'number': 1065}	0.6108	0.7040	0.6541	0.7537
0.6765	5.0	50	0.7144	{'precision': 0.6514047866805411, 'recall': 0.7737948084054388, 'f1': 0.7073446327683616, 'number': 809}	{'precision': 0.09230769230769231, 'recall': 0.05042016806722689, 'f1': 0.06521739130434782, 'number': 119}	{'precision': 0.7019810508182601, 'recall': 0.7652582159624414, 'f1': 0.7322551662174304, 'number': 1065}	0.6616	0.7260	0.6923	0.7773
0.5613	6.0	60	0.6796	{'precision': 0.6635514018691588, 'recall': 0.7898640296662547, 'f1': 0.7212189616252822, 'number': 809}	{'precision': 0.15306122448979592, 'recall': 0.12605042016806722, 'f1': 0.1382488479262673, 'number': 119}	{'precision': 0.7274320771253286, 'recall': 0.7793427230046949, 'f1': 0.7524932003626473, 'number': 1065}	0.6739	0.7446	0.7075	0.7927
0.4872	7.0	70	0.6554	{'precision': 0.6592517694641051, 'recall': 0.8059332509270705, 'f1': 0.7252502780867631, 'number': 809}	{'precision': 0.22549019607843138, 'recall': 0.19327731092436976, 'f1': 0.20814479638009048, 'number': 119}	{'precision': 0.7383177570093458, 'recall': 0.815962441314554, 'f1': 0.775200713648528, 'number': 1065}	0.6808	0.7747	0.7247	0.7997
0.4334	8.0	80	0.6526	{'precision': 0.6941176470588235, 'recall': 0.8022249690976514, 'f1': 0.7442660550458714, 'number': 809}	{'precision': 0.24545454545454545, 'recall': 0.226890756302521, 'f1': 0.23580786026200873, 'number': 119}	{'precision': 0.7493627867459643, 'recall': 0.828169014084507, 'f1': 0.7867975022301517, 'number': 1065}	0.7012	0.7817	0.7393	0.8035
0.3941	9.0	90	0.6694	{'precision': 0.7048997772828508, 'recall': 0.7824474660074165, 'f1': 0.741652021089631, 'number': 809}	{'precision': 0.22099447513812154, 'recall': 0.33613445378151263, 'f1': 0.26666666666666666, 'number': 119}	{'precision': 0.7218984179850125, 'recall': 0.8140845070422535, 'f1': 0.76522506619594, 'number': 1065}	0.6754	0.7727	0.7208	0.8007
0.3556	10.0	100	0.6607	{'precision': 0.694006309148265, 'recall': 0.8158220024721878, 'f1': 0.75, 'number': 809}	{'precision': 0.25, 'recall': 0.2773109243697479, 'f1': 0.26294820717131473, 'number': 119}	{'precision': 0.7846153846153846, 'recall': 0.8140845070422535, 'f1': 0.7990783410138248, 'number': 1065}	0.7130	0.7827	0.7462	0.8068
0.3245	11.0	110	0.6728	{'precision': 0.6990595611285266, 'recall': 0.826946847960445, 'f1': 0.7576443941109853, 'number': 809}	{'precision': 0.2892561983471074, 'recall': 0.29411764705882354, 'f1': 0.2916666666666667, 'number': 119}	{'precision': 0.7817703768624014, 'recall': 0.8375586854460094, 'f1': 0.8087035358114233, 'number': 1065}	0.7192	0.8008	0.7578	0.8089
0.3113	12.0	120	0.6799	{'precision': 0.71875, 'recall': 0.796044499381953, 'f1': 0.755425219941349, 'number': 809}	{'precision': 0.25903614457831325, 'recall': 0.36134453781512604, 'f1': 0.3017543859649123, 'number': 119}	{'precision': 0.775330396475771, 'recall': 0.8262910798122066, 'f1': 0.8, 'number': 1065}	0.7132	0.7863	0.7480	0.8106
0.2921	13.0	130	0.6836	{'precision': 0.7070063694267515, 'recall': 0.823238566131026, 'f1': 0.7607081667618503, 'number': 809}	{'precision': 0.32432432432432434, 'recall': 0.3025210084033613, 'f1': 0.31304347826086953, 'number': 119}	{'precision': 0.7976513098464318, 'recall': 0.8291079812206573, 'f1': 0.8130755064456722, 'number': 1065}	0.7338	0.7953	0.7633	0.8122
0.2841	14.0	140	0.6848	{'precision': 0.7150537634408602, 'recall': 0.8220024721878862, 'f1': 0.7648073605520415, 'number': 809}	{'precision': 0.26666666666666666, 'recall': 0.33613445378151263, 'f1': 0.2973977695167286, 'number': 119}	{'precision': 0.7841726618705036, 'recall': 0.8187793427230047, 'f1': 0.8011024345429489, 'number': 1065}	0.7194	0.7913	0.7536	0.8127
0.2793	15.0	150	0.6857	{'precision': 0.7176981541802389, 'recall': 0.8170580964153276, 'f1': 0.7641618497109827, 'number': 809}	{'precision': 0.28368794326241137, 'recall': 0.33613445378151263, 'f1': 0.3076923076923077, 'number': 119}	{'precision': 0.7773820124666073, 'recall': 0.819718309859155, 'f1': 0.7979890310786105, 'number': 1065}	0.7204	0.7898	0.7535	0.8139

Framework versions

• Transformers 4.27.0.dev0
• Pytorch 1.8.0+cu101
• Tokenizers 0.13.2