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: 1.2144
• Answer: {'precision': 0.46146146146146144, 'recall': 0.5698393077873919, 'f1': 0.5099557522123893, 'number': 809}
• Header: {'precision': 0.4024390243902439, 'recall': 0.2773109243697479, 'f1': 0.3283582089552239, 'number': 119}
• Question: {'precision': 0.5888412017167381, 'recall': 0.644131455399061, 'f1': 0.6152466367713004, 'number': 1065}
• Overall Precision: 0.5254
• Overall Recall: 0.5921
• Overall F1: 0.5567
• Overall Accuracy: 0.6483

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: 4
• eval_batch_size: 2
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 15

Training results

Training Loss	Epoch	Step	Validation Loss	Answer	Header	Question	Overall Precision	Overall Recall	Overall F1	Overall Accuracy
1.5615	1.0	38	1.2309	{'precision': 0.23910171730515192, 'recall': 0.44746600741656367, 'f1': 0.311665949203616, 'number': 809}	{'precision': 0.2830188679245283, 'recall': 0.12605042016806722, 'f1': 0.1744186046511628, 'number': 119}	{'precision': 0.35969209237228833, 'recall': 0.48262910798122066, 'f1': 0.4121892542101042, 'number': 1065}	0.2974	0.4471	0.3572	0.4649
1.1729	2.0	76	1.0880	{'precision': 0.3109656301145663, 'recall': 0.46971569839307786, 'f1': 0.37419990152634175, 'number': 809}	{'precision': 0.423728813559322, 'recall': 0.21008403361344538, 'f1': 0.2808988764044944, 'number': 119}	{'precision': 0.507488986784141, 'recall': 0.5408450704225352, 'f1': 0.5236363636363637, 'number': 1065}	0.4060	0.4922	0.4450	0.5557
1.0126	3.0	114	1.0622	{'precision': 0.31921110299488675, 'recall': 0.5401730531520396, 'f1': 0.40128558310376494, 'number': 809}	{'precision': 0.38372093023255816, 'recall': 0.2773109243697479, 'f1': 0.32195121951219513, 'number': 119}	{'precision': 0.4930662557781202, 'recall': 0.6009389671361502, 'f1': 0.5416842996191282, 'number': 1065}	0.4032	0.5569	0.4678	0.5662
0.9042	4.0	152	1.0144	{'precision': 0.3859060402684564, 'recall': 0.5686032138442522, 'f1': 0.45977011494252873, 'number': 809}	{'precision': 0.312, 'recall': 0.3277310924369748, 'f1': 0.31967213114754095, 'number': 119}	{'precision': 0.542713567839196, 'recall': 0.6084507042253521, 'f1': 0.5737051792828685, 'number': 1065}	0.4568	0.5755	0.5093	0.6204
0.7609	5.0	190	1.0307	{'precision': 0.41357466063348414, 'recall': 0.5648949320148331, 'f1': 0.47753396029258094, 'number': 809}	{'precision': 0.3373493975903614, 'recall': 0.23529411764705882, 'f1': 0.2772277227722772, 'number': 119}	{'precision': 0.5755711775043937, 'recall': 0.6150234741784038, 'f1': 0.5946436677258284, 'number': 1065}	0.4901	0.5720	0.5279	0.6340
0.6792	6.0	228	1.0643	{'precision': 0.43541102077687444, 'recall': 0.595797280593325, 'f1': 0.5031315240083507, 'number': 809}	{'precision': 0.4142857142857143, 'recall': 0.24369747899159663, 'f1': 0.3068783068783069, 'number': 119}	{'precision': 0.5757575757575758, 'recall': 0.6065727699530516, 'f1': 0.5907636031092821, 'number': 1065}	0.5033	0.5805	0.5391	0.6180
0.6081	7.0	266	1.0222	{'precision': 0.4691780821917808, 'recall': 0.5080346106304079, 'f1': 0.4878338278931751, 'number': 809}	{'precision': 0.26618705035971224, 'recall': 0.31092436974789917, 'f1': 0.2868217054263566, 'number': 119}	{'precision': 0.5478056426332288, 'recall': 0.6563380281690141, 'f1': 0.5971806920119608, 'number': 1065}	0.5007	0.5755	0.5355	0.6424
0.5218	8.0	304	1.0641	{'precision': 0.42940038684719534, 'recall': 0.5488257107540173, 'f1': 0.481823114487249, 'number': 809}	{'precision': 0.3409090909090909, 'recall': 0.25210084033613445, 'f1': 0.2898550724637681, 'number': 119}	{'precision': 0.5291512915129152, 'recall': 0.6732394366197183, 'f1': 0.5925619834710744, 'number': 1065}	0.4808	0.5976	0.5329	0.6167
0.468	9.0	342	1.1145	{'precision': 0.4584942084942085, 'recall': 0.5871446229913473, 'f1': 0.5149051490514905, 'number': 809}	{'precision': 0.3924050632911392, 'recall': 0.2605042016806723, 'f1': 0.31313131313131315, 'number': 119}	{'precision': 0.5921501706484642, 'recall': 0.6516431924882629, 'f1': 0.6204738489047832, 'number': 1065}	0.5247	0.6021	0.5607	0.6527
0.4159	10.0	380	1.1606	{'precision': 0.4683281412253375, 'recall': 0.5574783683559951, 'f1': 0.5090293453724605, 'number': 809}	{'precision': 0.367816091954023, 'recall': 0.2689075630252101, 'f1': 0.31067961165048547, 'number': 119}	{'precision': 0.5958369470945359, 'recall': 0.6450704225352113, 'f1': 0.6194770063119928, 'number': 1065}	0.5311	0.5871	0.5577	0.6521
0.3764	11.0	418	1.2370	{'precision': 0.4515828677839851, 'recall': 0.5995055624227441, 'f1': 0.5151354221986192, 'number': 809}	{'precision': 0.3888888888888889, 'recall': 0.29411764705882354, 'f1': 0.3349282296650718, 'number': 119}	{'precision': 0.6041083099906629, 'recall': 0.6075117370892019, 'f1': 0.6058052434456929, 'number': 1065}	0.5221	0.5855	0.5520	0.6248
0.3393	12.0	456	1.2263	{'precision': 0.46161515453639085, 'recall': 0.5723114956736712, 'f1': 0.5110375275938189, 'number': 809}	{'precision': 0.35135135135135137, 'recall': 0.2184873949579832, 'f1': 0.26943005181347146, 'number': 119}	{'precision': 0.5891132572431957, 'recall': 0.6300469483568075, 'f1': 0.6088929219600726, 'number': 1065}	0.5235	0.5820	0.5512	0.6369
0.3253	13.0	494	1.2059	{'precision': 0.4658590308370044, 'recall': 0.522867737948084, 'f1': 0.49271986022131625, 'number': 809}	{'precision': 0.3402061855670103, 'recall': 0.2773109243697479, 'f1': 0.3055555555555556, 'number': 119}	{'precision': 0.5623028391167192, 'recall': 0.6694835680751173, 'f1': 0.6112301757393913, 'number': 1065}	0.5143	0.5866	0.5481	0.6376
0.2996	14.0	532	1.2311	{'precision': 0.46296296296296297, 'recall': 0.5871446229913473, 'f1': 0.5177111716621253, 'number': 809}	{'precision': 0.3263157894736842, 'recall': 0.2605042016806723, 'f1': 0.2897196261682243, 'number': 119}	{'precision': 0.5991189427312775, 'recall': 0.6384976525821596, 'f1': 0.6181818181818182, 'number': 1065}	0.5257	0.5951	0.5582	0.6350
0.2892	15.0	570	1.2144	{'precision': 0.46146146146146144, 'recall': 0.5698393077873919, 'f1': 0.5099557522123893, 'number': 809}	{'precision': 0.4024390243902439, 'recall': 0.2773109243697479, 'f1': 0.3283582089552239, 'number': 119}	{'precision': 0.5888412017167381, 'recall': 0.644131455399061, 'f1': 0.6152466367713004, 'number': 1065}	0.5254	0.5921	0.5567	0.6483

Framework versions

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