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: 1.3429
• Answer: {'precision': 0.4732142857142857, 'recall': 0.5896168108776267, 'f1': 0.5250412768299395, 'number': 809}
• Header: {'precision': 0.3838383838383838, 'recall': 0.31932773109243695, 'f1': 0.34862385321100914, 'number': 119}
• Question: {'precision': 0.6107784431137725, 'recall': 0.6704225352112676, 'f1': 0.6392121754700089, 'number': 1065}
• Overall Precision: 0.5400
• Overall Recall: 0.6167
• Overall F1: 0.5758
• Overall Accuracy: 0.6585

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: 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: 25
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Answer	Header	Question	Overall Precision	Overall Recall	Overall F1	Overall Accuracy
1.736	1.0	10	1.4843	{'precision': 0.09492635024549918, 'recall': 0.1433868974042027, 'f1': 0.11422944362383061, 'number': 809}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 119}	{'precision': 0.20766773162939298, 'recall': 0.24413145539906103, 'f1': 0.22442813983599483, 'number': 1065}	0.1520	0.1887	0.1683	0.3955
1.3661	2.0	20	1.2557	{'precision': 0.26396807297605474, 'recall': 0.5723114956736712, 'f1': 0.3612953570035115, 'number': 809}	{'precision': 0.3384615384615385, 'recall': 0.18487394957983194, 'f1': 0.23913043478260868, 'number': 119}	{'precision': 0.3391768292682927, 'recall': 0.41784037558685444, 'f1': 0.37442153975599496, 'number': 1065}	0.2970	0.4666	0.3630	0.4549
1.182	3.0	30	1.1125	{'precision': 0.2653208363374189, 'recall': 0.45488257107540175, 'f1': 0.3351548269581056, 'number': 809}	{'precision': 0.5106382978723404, 'recall': 0.20168067226890757, 'f1': 0.2891566265060241, 'number': 119}	{'precision': 0.38895631067961167, 'recall': 0.6018779342723005, 'f1': 0.4725396240324365, 'number': 1065}	0.3352	0.5183	0.4071	0.5618
1.0359	4.0	40	1.0303	{'precision': 0.3092682926829268, 'recall': 0.39184177997527814, 'f1': 0.3456924754634678, 'number': 809}	{'precision': 0.3055555555555556, 'recall': 0.2773109243697479, 'f1': 0.2907488986784141, 'number': 119}	{'precision': 0.42207792207792205, 'recall': 0.6713615023474179, 'f1': 0.51830373323668, 'number': 1065}	0.3767	0.5344	0.4419	0.6034
0.929	5.0	50	1.1381	{'precision': 0.30710466004583653, 'recall': 0.4969097651421508, 'f1': 0.3796033994334278, 'number': 809}	{'precision': 0.35714285714285715, 'recall': 0.25210084033613445, 'f1': 0.29556650246305416, 'number': 119}	{'precision': 0.4323086984957489, 'recall': 0.6206572769953052, 'f1': 0.5096376252891287, 'number': 1065}	0.3741	0.5484	0.4448	0.5838
0.8305	6.0	60	1.1595	{'precision': 0.3615506329113924, 'recall': 0.5648949320148331, 'f1': 0.44090689821514706, 'number': 809}	{'precision': 0.37333333333333335, 'recall': 0.23529411764705882, 'f1': 0.28865979381443296, 'number': 119}	{'precision': 0.5224416517055656, 'recall': 0.5464788732394367, 'f1': 0.5341899954107389, 'number': 1065}	0.4350	0.5354	0.4800	0.5884
0.7288	7.0	70	1.0267	{'precision': 0.4050901378579003, 'recall': 0.4721878862793572, 'f1': 0.4360730593607306, 'number': 809}	{'precision': 0.308411214953271, 'recall': 0.2773109243697479, 'f1': 0.29203539823008845, 'number': 119}	{'precision': 0.48145604395604397, 'recall': 0.6582159624413145, 'f1': 0.5561285204284014, 'number': 1065}	0.4453	0.5600	0.4961	0.6406
0.6547	8.0	80	1.0727	{'precision': 0.41427247451343835, 'recall': 0.5525339925834364, 'f1': 0.47351694915254233, 'number': 809}	{'precision': 0.36046511627906974, 'recall': 0.2605042016806723, 'f1': 0.30243902439024395, 'number': 119}	{'precision': 0.49452154857560265, 'recall': 0.6356807511737089, 'f1': 0.5562859490550535, 'number': 1065}	0.4558	0.5795	0.5103	0.6323
0.6	9.0	90	1.0490	{'precision': 0.4189723320158103, 'recall': 0.5241038318912238, 'f1': 0.4656781987918726, 'number': 809}	{'precision': 0.2972972972972973, 'recall': 0.2773109243697479, 'f1': 0.28695652173913044, 'number': 119}	{'precision': 0.5518341307814992, 'recall': 0.6497652582159624, 'f1': 0.5968089693833549, 'number': 1065}	0.4834	0.5765	0.5259	0.6329
0.5657	10.0	100	1.1953	{'precision': 0.40772200772200773, 'recall': 0.6526576019777504, 'f1': 0.5019011406844107, 'number': 809}	{'precision': 0.41333333333333333, 'recall': 0.2605042016806723, 'f1': 0.3195876288659794, 'number': 119}	{'precision': 0.5609540636042403, 'recall': 0.596244131455399, 'f1': 0.5780609922621757, 'number': 1065}	0.4772	0.5991	0.5313	0.6268
0.4991	11.0	110	1.1014	{'precision': 0.4277056277056277, 'recall': 0.6106304079110012, 'f1': 0.5030549898167006, 'number': 809}	{'precision': 0.3763440860215054, 'recall': 0.29411764705882354, 'f1': 0.33018867924528306, 'number': 119}	{'precision': 0.5501730103806228, 'recall': 0.5971830985915493, 'f1': 0.5727149932462855, 'number': 1065}	0.4846	0.5845	0.5299	0.6306
0.4602	12.0	120	1.1289	{'precision': 0.45584988962472406, 'recall': 0.5105067985166872, 'f1': 0.4816326530612245, 'number': 809}	{'precision': 0.2846153846153846, 'recall': 0.31092436974789917, 'f1': 0.29718875502008035, 'number': 119}	{'precision': 0.5492857142857143, 'recall': 0.7220657276995305, 'f1': 0.6239350912778904, 'number': 1065}	0.5004	0.6116	0.5505	0.6382
0.4175	13.0	130	1.2651	{'precision': 0.467502850627138, 'recall': 0.5067985166872683, 'f1': 0.4863582443653618, 'number': 809}	{'precision': 0.3114754098360656, 'recall': 0.31932773109243695, 'f1': 0.3153526970954357, 'number': 119}	{'precision': 0.5882352941176471, 'recall': 0.6948356807511737, 'f1': 0.6371071889797676, 'number': 1065}	0.5264	0.5961	0.5591	0.6272
0.3663	14.0	140	1.2097	{'precision': 0.4597918637653737, 'recall': 0.6007416563658838, 'f1': 0.5209003215434083, 'number': 809}	{'precision': 0.2962962962962963, 'recall': 0.2689075630252101, 'f1': 0.28193832599118945, 'number': 119}	{'precision': 0.5774533657745337, 'recall': 0.6685446009389672, 'f1': 0.6196692776327242, 'number': 1065}	0.5129	0.6172	0.5602	0.6399
0.3358	15.0	150	1.2039	{'precision': 0.4482758620689655, 'recall': 0.5945611866501854, 'f1': 0.5111583421891605, 'number': 809}	{'precision': 0.3522727272727273, 'recall': 0.2605042016806723, 'f1': 0.29951690821256044, 'number': 119}	{'precision': 0.5680131904369332, 'recall': 0.6469483568075117, 'f1': 0.6049165935030728, 'number': 1065}	0.5059	0.6026	0.5500	0.6425
0.3061	16.0	160	1.2335	{'precision': 0.46646942800788954, 'recall': 0.584672435105068, 'f1': 0.5189248491497532, 'number': 809}	{'precision': 0.3780487804878049, 'recall': 0.2605042016806723, 'f1': 0.30845771144278605, 'number': 119}	{'precision': 0.586352148272957, 'recall': 0.6535211267605634, 'f1': 0.6181172291296625, 'number': 1065}	0.5256	0.6021	0.5613	0.6572
0.2758	17.0	170	1.2667	{'precision': 0.47320525783619816, 'recall': 0.5784919653893696, 'f1': 0.5205784204671858, 'number': 809}	{'precision': 0.35135135135135137, 'recall': 0.3277310924369748, 'f1': 0.3391304347826087, 'number': 119}	{'precision': 0.6026431718061674, 'recall': 0.6422535211267606, 'f1': 0.6218181818181818, 'number': 1065}	0.5329	0.5976	0.5634	0.6511
0.2599	18.0	180	1.2470	{'precision': 0.467280163599182, 'recall': 0.5648949320148331, 'f1': 0.5114717403469503, 'number': 809}	{'precision': 0.38144329896907214, 'recall': 0.31092436974789917, 'f1': 0.34259259259259256, 'number': 119}	{'precision': 0.5965770171149144, 'recall': 0.6873239436619718, 'f1': 0.6387434554973822, 'number': 1065}	0.5326	0.6152	0.5709	0.6569
0.2519	19.0	190	1.3156	{'precision': 0.48720472440944884, 'recall': 0.6118665018541409, 'f1': 0.5424657534246575, 'number': 809}	{'precision': 0.37755102040816324, 'recall': 0.31092436974789917, 'f1': 0.3410138248847926, 'number': 119}	{'precision': 0.5979557069846678, 'recall': 0.6591549295774648, 'f1': 0.6270656543099598, 'number': 1065}	0.5393	0.6192	0.5765	0.6572
0.2372	20.0	200	1.2986	{'precision': 0.4742967992240543, 'recall': 0.6044499381953028, 'f1': 0.5315217391304348, 'number': 809}	{'precision': 0.3333333333333333, 'recall': 0.3277310924369748, 'f1': 0.3305084745762712, 'number': 119}	{'precision': 0.6078098471986417, 'recall': 0.672300469483568, 'f1': 0.6384306732055283, 'number': 1065}	0.5348	0.6242	0.5761	0.6582
0.2123	21.0	210	1.3440	{'precision': 0.4794238683127572, 'recall': 0.5760197775030902, 'f1': 0.523301516002246, 'number': 809}	{'precision': 0.4117647058823529, 'recall': 0.35294117647058826, 'f1': 0.3800904977375566, 'number': 119}	{'precision': 0.6082830025884383, 'recall': 0.6619718309859155, 'f1': 0.6339928057553956, 'number': 1065}	0.5432	0.6086	0.5741	0.6528
0.219	22.0	220	1.3150	{'precision': 0.48422090729783035, 'recall': 0.6069221260815822, 'f1': 0.5386725178277565, 'number': 809}	{'precision': 0.37383177570093457, 'recall': 0.33613445378151263, 'f1': 0.3539823008849558, 'number': 119}	{'precision': 0.5941666666666666, 'recall': 0.6694835680751173, 'f1': 0.6295805739514349, 'number': 1065}	0.5360	0.6242	0.5767	0.6548
0.2011	23.0	230	1.3252	{'precision': 0.474559686888454, 'recall': 0.5995055624227441, 'f1': 0.5297651556526488, 'number': 809}	{'precision': 0.37, 'recall': 0.31092436974789917, 'f1': 0.3378995433789954, 'number': 119}	{'precision': 0.5970915312232677, 'recall': 0.6553990610328638, 'f1': 0.6248880931065354, 'number': 1065}	0.5325	0.6121	0.5696	0.6469
0.1942	24.0	240	1.3343	{'precision': 0.4917864476386037, 'recall': 0.5920889987639061, 'f1': 0.5372966909702749, 'number': 809}	{'precision': 0.3925233644859813, 'recall': 0.35294117647058826, 'f1': 0.37168141592920356, 'number': 119}	{'precision': 0.5986733001658375, 'recall': 0.6779342723004694, 'f1': 0.6358432408630559, 'number': 1065}	0.5435	0.6237	0.5808	0.6583
0.1963	25.0	250	1.3429	{'precision': 0.4732142857142857, 'recall': 0.5896168108776267, 'f1': 0.5250412768299395, 'number': 809}	{'precision': 0.3838383838383838, 'recall': 0.31932773109243695, 'f1': 0.34862385321100914, 'number': 119}	{'precision': 0.6107784431137725, 'recall': 0.6704225352112676, 'f1': 0.6392121754700089, 'number': 1065}	0.5400	0.6167	0.5758	0.6585

Framework versions

• Transformers 4.39.0
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2