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: 1.1237
• Answer: {'precision': 0.38014311270125223, 'recall': 0.5253399258343634, 'f1': 0.44110015568240785, 'number': 809}
• Header: {'precision': 0.32608695652173914, 'recall': 0.25210084033613445, 'f1': 0.2843601895734597, 'number': 119}
• Question: {'precision': 0.5316760224538893, 'recall': 0.6225352112676056, 'f1': 0.5735294117647058, 'number': 1065}
• Overall Precision: 0.4550
• Overall Recall: 0.5610
• Overall F1: 0.5025
• Overall Accuracy: 0.6009

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
1.7571	1.0	10	1.5578	{'precision': 0.03216374269005848, 'recall': 0.027194066749072928, 'f1': 0.029470864032150032, 'number': 809}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 119}	{'precision': 0.24550898203592814, 'recall': 0.1539906103286385, 'f1': 0.189267166762839, 'number': 1065}	0.1376	0.0933	0.1112	0.3461
1.4754	2.0	20	1.3886	{'precision': 0.18979987088444158, 'recall': 0.36341161928306553, 'f1': 0.24936386768447838, 'number': 809}	{'precision': 0.0851063829787234, 'recall': 0.03361344537815126, 'f1': 0.048192771084337345, 'number': 119}	{'precision': 0.2655198204936425, 'recall': 0.3333333333333333, 'f1': 0.29558701082431305, 'number': 1065}	0.2226	0.3276	0.2651	0.4190
1.2882	3.0	30	1.2556	{'precision': 0.25, 'recall': 0.5030902348578492, 'f1': 0.3340172343044727, 'number': 809}	{'precision': 0.07547169811320754, 'recall': 0.03361344537815126, 'f1': 0.04651162790697674, 'number': 119}	{'precision': 0.3400431344356578, 'recall': 0.444131455399061, 'f1': 0.38517915309446255, 'number': 1065}	0.2878	0.4436	0.3491	0.4540
1.1508	4.0	40	1.1427	{'precision': 0.27153558052434457, 'recall': 0.5377008652657602, 'f1': 0.36084612194110327, 'number': 809}	{'precision': 0.23595505617977527, 'recall': 0.17647058823529413, 'f1': 0.20192307692307693, 'number': 119}	{'precision': 0.4009397024275646, 'recall': 0.4807511737089202, 'f1': 0.43723313407344155, 'number': 1065}	0.3261	0.4857	0.3902	0.5272
1.0506	5.0	50	1.1546	{'precision': 0.28481455563331, 'recall': 0.5030902348578492, 'f1': 0.36371760500446826, 'number': 809}	{'precision': 0.24719101123595505, 'recall': 0.18487394957983194, 'f1': 0.21153846153846156, 'number': 119}	{'precision': 0.4018324607329843, 'recall': 0.5765258215962441, 'f1': 0.4735827227150019, 'number': 1065}	0.3424	0.5233	0.4140	0.5441
0.9855	6.0	60	1.1005	{'precision': 0.31229012760241776, 'recall': 0.5747836835599506, 'f1': 0.4046997389033942, 'number': 809}	{'precision': 0.328125, 'recall': 0.17647058823529413, 'f1': 0.22950819672131148, 'number': 119}	{'precision': 0.47493403693931396, 'recall': 0.5070422535211268, 'f1': 0.49046321525885556, 'number': 1065}	0.3814	0.5148	0.4382	0.5656
0.9039	7.0	70	1.0551	{'precision': 0.32831608654750705, 'recall': 0.43139678615574784, 'f1': 0.37286324786324787, 'number': 809}	{'precision': 0.2743362831858407, 'recall': 0.2605042016806723, 'f1': 0.26724137931034486, 'number': 119}	{'precision': 0.4689306358381503, 'recall': 0.6093896713615023, 'f1': 0.5300122498979176, 'number': 1065}	0.4020	0.5163	0.4520	0.5981
0.841	8.0	80	1.0710	{'precision': 0.3379032258064516, 'recall': 0.5179233621755254, 'f1': 0.40897999023914106, 'number': 809}	{'precision': 0.2926829268292683, 'recall': 0.20168067226890757, 'f1': 0.23880597014925373, 'number': 119}	{'precision': 0.4723435225618632, 'recall': 0.6093896713615023, 'f1': 0.5321853218532185, 'number': 1065}	0.4050	0.5479	0.4658	0.5885
0.7758	9.0	90	1.0917	{'precision': 0.3506916192026037, 'recall': 0.5327564894932015, 'f1': 0.4229636898920511, 'number': 809}	{'precision': 0.3076923076923077, 'recall': 0.23529411764705882, 'f1': 0.26666666666666666, 'number': 119}	{'precision': 0.4916286149162861, 'recall': 0.6065727699530516, 'f1': 0.5430853299705759, 'number': 1065}	0.4195	0.5544	0.4776	0.5892
0.7737	10.0	100	1.1005	{'precision': 0.36325503355704697, 'recall': 0.5352286773794809, 'f1': 0.43278360819590206, 'number': 809}	{'precision': 0.3902439024390244, 'recall': 0.2689075630252101, 'f1': 0.31840796019900497, 'number': 119}	{'precision': 0.5075456711675933, 'recall': 0.6, 'f1': 0.5499139414802064, 'number': 1065}	0.4358	0.5539	0.4878	0.5934
0.6942	11.0	110	1.0974	{'precision': 0.3707136237256719, 'recall': 0.49443757725587145, 'f1': 0.423728813559322, 'number': 809}	{'precision': 0.34, 'recall': 0.2857142857142857, 'f1': 0.31050228310502287, 'number': 119}	{'precision': 0.5255775577557755, 'recall': 0.5981220657276995, 'f1': 0.559508124725516, 'number': 1065}	0.4479	0.5374	0.4886	0.6107
0.691	12.0	120	1.0991	{'precision': 0.381950774840474, 'recall': 0.5179233621755254, 'f1': 0.43966421825813223, 'number': 809}	{'precision': 0.36666666666666664, 'recall': 0.2773109243697479, 'f1': 0.31578947368421056, 'number': 119}	{'precision': 0.5208825847123719, 'recall': 0.6206572769953052, 'f1': 0.5664095972579263, 'number': 1065}	0.4532	0.5585	0.5003	0.6116
0.6595	13.0	130	1.1179	{'precision': 0.3776223776223776, 'recall': 0.5339925834363412, 'f1': 0.44239631336405527, 'number': 809}	{'precision': 0.3563218390804598, 'recall': 0.2605042016806723, 'f1': 0.30097087378640774, 'number': 119}	{'precision': 0.530562347188264, 'recall': 0.6112676056338028, 'f1': 0.5680628272251308, 'number': 1065}	0.4532	0.5590	0.5006	0.6010
0.6288	14.0	140	1.1441	{'precision': 0.3689075630252101, 'recall': 0.5426452410383189, 'f1': 0.4392196098049025, 'number': 809}	{'precision': 0.3595505617977528, 'recall': 0.2689075630252101, 'f1': 0.3076923076923077, 'number': 119}	{'precision': 0.54614733276884, 'recall': 0.6056338028169014, 'f1': 0.5743544078361531, 'number': 1065}	0.4537	0.5600	0.5012	0.5913
0.6245	15.0	150	1.1237	{'precision': 0.38014311270125223, 'recall': 0.5253399258343634, 'f1': 0.44110015568240785, 'number': 809}	{'precision': 0.32608695652173914, 'recall': 0.25210084033613445, 'f1': 0.2843601895734597, 'number': 119}	{'precision': 0.5316760224538893, 'recall': 0.6225352112676056, 'f1': 0.5735294117647058, 'number': 1065}	0.4550	0.5610	0.5025	0.6009

Framework versions

• Transformers 4.38.2
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2