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: 1.1695
• Answer: {'precision': 0.35993485342019543, 'recall': 0.546353522867738, 'f1': 0.4339715267550319, 'number': 809}
• Header: {'precision': 0.3488372093023256, 'recall': 0.25210084033613445, 'f1': 0.2926829268292683, 'number': 119}
• Question: {'precision': 0.5217762596071733, 'recall': 0.5737089201877934, 'f1': 0.5465116279069767, 'number': 1065}
• Overall Precision: 0.4358
• Overall Recall: 0.5434
• Overall F1: 0.4837
• Overall Accuracy: 0.5907

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
1.7983	1.0	10	1.5799	{'precision': 0.02035830618892508, 'recall': 0.030902348578491966, 'f1': 0.024545900834560624, 'number': 809}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 119}	{'precision': 0.19827586206896552, 'recall': 0.1295774647887324, 'f1': 0.15672913117546847, 'number': 1065}	0.0847	0.0818	0.0832	0.3629
1.5573	2.0	20	1.4500	{'precision': 0.11375661375661375, 'recall': 0.2126081582200247, 'f1': 0.14821197759586383, 'number': 809}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 119}	{'precision': 0.21101928374655649, 'recall': 0.3596244131455399, 'f1': 0.2659722222222222, 'number': 1065}	0.1668	0.2785	0.2086	0.4045
1.3717	3.0	30	1.2959	{'precision': 0.16349108789182545, 'recall': 0.3288009888751545, 'f1': 0.21839080459770116, 'number': 809}	{'precision': 0.061224489795918366, 'recall': 0.025210084033613446, 'f1': 0.03571428571428571, 'number': 119}	{'precision': 0.2739408009286129, 'recall': 0.4431924882629108, 'f1': 0.3385939741750359, 'number': 1065}	0.2180	0.3718	0.2749	0.4466
1.2568	4.0	40	1.2032	{'precision': 0.20166898470097358, 'recall': 0.3584672435105068, 'f1': 0.258121940364931, 'number': 809}	{'precision': 0.23809523809523808, 'recall': 0.16806722689075632, 'f1': 0.19704433497536947, 'number': 119}	{'precision': 0.3354214123006834, 'recall': 0.5530516431924882, 'f1': 0.4175824175824176, 'number': 1065}	0.2743	0.4511	0.3411	0.4942
1.1591	5.0	50	1.1897	{'precision': 0.23584277148567623, 'recall': 0.43757725587144625, 'f1': 0.30649350649350654, 'number': 809}	{'precision': 0.27848101265822783, 'recall': 0.18487394957983194, 'f1': 0.2222222222222222, 'number': 119}	{'precision': 0.36172765446910615, 'recall': 0.5661971830985916, 'f1': 0.44143484626647145, 'number': 1065}	0.3015	0.4912	0.3737	0.5069
1.075	6.0	60	1.1092	{'precision': 0.2832044975404076, 'recall': 0.49814585908529047, 'f1': 0.36111111111111116, 'number': 809}	{'precision': 0.34375, 'recall': 0.18487394957983194, 'f1': 0.24043715846994534, 'number': 119}	{'precision': 0.4609571788413098, 'recall': 0.5154929577464789, 'f1': 0.48670212765957444, 'number': 1065}	0.3637	0.4887	0.4170	0.5650
1.0011	7.0	70	1.0638	{'precision': 0.30843373493975906, 'recall': 0.4746600741656366, 'f1': 0.3739045764362221, 'number': 809}	{'precision': 0.3333333333333333, 'recall': 0.19327731092436976, 'f1': 0.24468085106382975, 'number': 119}	{'precision': 0.47068021892103207, 'recall': 0.5652582159624413, 'f1': 0.5136518771331057, 'number': 1065}	0.3891	0.5063	0.4400	0.5852
0.9357	8.0	80	1.1937	{'precision': 0.3132867132867133, 'recall': 0.553770086526576, 'f1': 0.4001786511835641, 'number': 809}	{'precision': 0.2987012987012987, 'recall': 0.19327731092436976, 'f1': 0.23469387755102045, 'number': 119}	{'precision': 0.48117839607201307, 'recall': 0.5521126760563381, 'f1': 0.5142107564494972, 'number': 1065}	0.3881	0.5314	0.4485	0.5559
0.8751	9.0	90	1.0553	{'precision': 0.3413173652694611, 'recall': 0.4932014833127318, 'f1': 0.40343781597573314, 'number': 809}	{'precision': 0.29577464788732394, 'recall': 0.17647058823529413, 'f1': 0.2210526315789474, 'number': 119}	{'precision': 0.47987851176917234, 'recall': 0.5934272300469483, 'f1': 0.5306465155331653, 'number': 1065}	0.4114	0.5278	0.4624	0.5986
0.8585	10.0	100	1.1844	{'precision': 0.33745454545454545, 'recall': 0.5735475896168108, 'f1': 0.4249084249084249, 'number': 809}	{'precision': 0.37333333333333335, 'recall': 0.23529411764705882, 'f1': 0.28865979381443296, 'number': 119}	{'precision': 0.5013333333333333, 'recall': 0.5295774647887324, 'f1': 0.5150684931506849, 'number': 1065}	0.4101	0.5299	0.4623	0.5706
0.7894	11.0	110	1.1752	{'precision': 0.3395311236863379, 'recall': 0.519159456118665, 'f1': 0.41055718475073316, 'number': 809}	{'precision': 0.4, 'recall': 0.23529411764705882, 'f1': 0.29629629629629634, 'number': 119}	{'precision': 0.5020885547201337, 'recall': 0.564319248826291, 'f1': 0.5313881520778072, 'number': 1065}	0.4189	0.5263	0.4665	0.5781
0.7809	12.0	120	1.2025	{'precision': 0.34820031298904536, 'recall': 0.5500618046971569, 'f1': 0.4264494489698131, 'number': 809}	{'precision': 0.37037037037037035, 'recall': 0.25210084033613445, 'f1': 0.3, 'number': 119}	{'precision': 0.5093062605752962, 'recall': 0.5652582159624413, 'f1': 0.5358255451713396, 'number': 1065}	0.4238	0.5404	0.4751	0.5800
0.7452	13.0	130	1.1974	{'precision': 0.3550436854646545, 'recall': 0.5525339925834364, 'f1': 0.4323017408123791, 'number': 809}	{'precision': 0.358974358974359, 'recall': 0.23529411764705882, 'f1': 0.28426395939086296, 'number': 119}	{'precision': 0.5247787610619469, 'recall': 0.5568075117370892, 'f1': 0.5403189066059226, 'number': 1065}	0.4329	0.5359	0.4789	0.5830
0.7108	14.0	140	1.1267	{'precision': 0.373015873015873, 'recall': 0.522867737948084, 'f1': 0.43540916109109623, 'number': 809}	{'precision': 0.3448275862068966, 'recall': 0.25210084033613445, 'f1': 0.2912621359223301, 'number': 119}	{'precision': 0.5193548387096775, 'recall': 0.6046948356807512, 'f1': 0.5587852494577007, 'number': 1065}	0.4458	0.5504	0.4926	0.6044
0.702	15.0	150	1.1695	{'precision': 0.35993485342019543, 'recall': 0.546353522867738, 'f1': 0.4339715267550319, 'number': 809}	{'precision': 0.3488372093023256, 'recall': 0.25210084033613445, 'f1': 0.2926829268292683, 'number': 119}	{'precision': 0.5217762596071733, 'recall': 0.5737089201877934, 'f1': 0.5465116279069767, 'number': 1065}	0.4358	0.5434	0.4837	0.5907

Framework versions

• Transformers 4.38.2
• Pytorch 2.1.0+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2