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: 0.6866
• Answer: {'precision': 0.7130339539978094, 'recall': 0.8046971569839307, 'f1': 0.7560975609756098, 'number': 809}
• Header: {'precision': 0.3384615384615385, 'recall': 0.3697478991596639, 'f1': 0.35341365461847385, 'number': 119}
• Question: {'precision': 0.7763975155279503, 'recall': 0.8215962441314554, 'f1': 0.7983576642335766, 'number': 1065}
• Overall Precision: 0.7235
• Overall Recall: 0.7878
• Overall F1: 0.7543
• Overall Accuracy: 0.8126

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

Training results

Training Loss	Epoch	Step	Validation Loss	Answer	Header	Question	Overall Precision	Overall Recall	Overall F1	Overall Accuracy
1.7968	1.0	10	1.5972	{'precision': 0.011235955056179775, 'recall': 0.011124845488257108, 'f1': 0.011180124223602483, 'number': 809}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 119}	{'precision': 0.1959544879898862, 'recall': 0.14553990610328638, 'f1': 0.1670258620689655, 'number': 1065}	0.1030	0.0823	0.0915	0.3535
1.4694	2.0	20	1.2467	{'precision': 0.2002053388090349, 'recall': 0.24103831891223734, 'f1': 0.21873247335950646, 'number': 809}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 119}	{'precision': 0.43186895011169024, 'recall': 0.5446009389671361, 'f1': 0.48172757475083056, 'number': 1065}	0.3345	0.3889	0.3596	0.6093
1.0892	3.0	30	0.9301	{'precision': 0.49691991786447637, 'recall': 0.5982694684796045, 'f1': 0.5429052159282108, 'number': 809}	{'precision': 0.08108108108108109, 'recall': 0.025210084033613446, 'f1': 0.038461538461538464, 'number': 119}	{'precision': 0.5869205298013245, 'recall': 0.6657276995305165, 'f1': 0.62384513858337, 'number': 1065}	0.5390	0.6001	0.5679	0.7041
0.8148	4.0	40	0.7921	{'precision': 0.5805243445692884, 'recall': 0.7663782447466008, 'f1': 0.660628662759723, 'number': 809}	{'precision': 0.2, 'recall': 0.12605042016806722, 'f1': 0.15463917525773196, 'number': 119}	{'precision': 0.6657534246575343, 'recall': 0.6845070422535211, 'f1': 0.6749999999999999, 'number': 1065}	0.6095	0.6844	0.6448	0.7498
0.6789	5.0	50	0.7126	{'precision': 0.6466942148760331, 'recall': 0.7737948084054388, 'f1': 0.7045582442318515, 'number': 809}	{'precision': 0.23809523809523808, 'recall': 0.21008403361344538, 'f1': 0.22321428571428573, 'number': 119}	{'precision': 0.6851535836177475, 'recall': 0.7539906103286385, 'f1': 0.7179257934734019, 'number': 1065}	0.6477	0.7296	0.6862	0.7822
0.5701	6.0	60	0.6734	{'precision': 0.6524390243902439, 'recall': 0.7935723114956736, 'f1': 0.7161182375906302, 'number': 809}	{'precision': 0.25, 'recall': 0.18487394957983194, 'f1': 0.21256038647342995, 'number': 119}	{'precision': 0.6886564762670957, 'recall': 0.8037558685446009, 'f1': 0.7417677642980937, 'number': 1065}	0.6566	0.7627	0.7057	0.7949
0.497	7.0	70	0.6688	{'precision': 0.6719745222929936, 'recall': 0.7824474660074165, 'f1': 0.7230154197601371, 'number': 809}	{'precision': 0.2857142857142857, 'recall': 0.2689075630252101, 'f1': 0.277056277056277, 'number': 119}	{'precision': 0.7403267411865864, 'recall': 0.8084507042253521, 'f1': 0.7728904847396768, 'number': 1065}	0.6883	0.7657	0.7249	0.7976
0.4549	8.0	80	0.6561	{'precision': 0.6881028938906752, 'recall': 0.7935723114956736, 'f1': 0.7370838117106774, 'number': 809}	{'precision': 0.25, 'recall': 0.25210084033613445, 'f1': 0.2510460251046025, 'number': 119}	{'precision': 0.7432784041630529, 'recall': 0.8046948356807512, 'f1': 0.7727682596934174, 'number': 1065}	0.6931	0.7672	0.7283	0.8045
0.4095	9.0	90	0.6514	{'precision': 0.694206008583691, 'recall': 0.799752781211372, 'f1': 0.7432510051694429, 'number': 809}	{'precision': 0.29411764705882354, 'recall': 0.29411764705882354, 'f1': 0.29411764705882354, 'number': 119}	{'precision': 0.7452830188679245, 'recall': 0.815962441314554, 'f1': 0.7790228597041686, 'number': 1065}	0.6996	0.7782	0.7368	0.8027
0.3629	10.0	100	0.6616	{'precision': 0.7035010940919038, 'recall': 0.7948084054388134, 'f1': 0.7463726059199072, 'number': 809}	{'precision': 0.29927007299270075, 'recall': 0.3445378151260504, 'f1': 0.3203125, 'number': 119}	{'precision': 0.7564216120460585, 'recall': 0.8018779342723005, 'f1': 0.7784867821330903, 'number': 1065}	0.7055	0.7717	0.7371	0.8075
0.3322	11.0	110	0.6668	{'precision': 0.7112068965517241, 'recall': 0.8158220024721878, 'f1': 0.75993091537133, 'number': 809}	{'precision': 0.336283185840708, 'recall': 0.31932773109243695, 'f1': 0.32758620689655166, 'number': 119}	{'precision': 0.783273381294964, 'recall': 0.8178403755868544, 'f1': 0.8001837390904916, 'number': 1065}	0.7288	0.7873	0.7569	0.8120
0.3188	12.0	120	0.6768	{'precision': 0.7225305216426193, 'recall': 0.8046971569839307, 'f1': 0.7614035087719299, 'number': 809}	{'precision': 0.33076923076923076, 'recall': 0.36134453781512604, 'f1': 0.34538152610441764, 'number': 119}	{'precision': 0.7759078830823738, 'recall': 0.8225352112676056, 'f1': 0.7985414767547857, 'number': 1065}	0.7269	0.7878	0.7561	0.8119
0.2936	13.0	130	0.6787	{'precision': 0.7122692725298588, 'recall': 0.8108776266996292, 'f1': 0.7583815028901735, 'number': 809}	{'precision': 0.35384615384615387, 'recall': 0.3865546218487395, 'f1': 0.3694779116465864, 'number': 119}	{'precision': 0.7807486631016043, 'recall': 0.8225352112676056, 'f1': 0.8010973936899862, 'number': 1065}	0.7262	0.7918	0.7576	0.8133
0.2894	14.0	140	0.6863	{'precision': 0.7113289760348583, 'recall': 0.8071693448702101, 'f1': 0.7562246670526924, 'number': 809}	{'precision': 0.34108527131782945, 'recall': 0.3697478991596639, 'f1': 0.35483870967741943, 'number': 119}	{'precision': 0.7852650494159928, 'recall': 0.8206572769953052, 'f1': 0.8025711662075299, 'number': 1065}	0.7273	0.7883	0.7566	0.8111
0.2813	15.0	150	0.6866	{'precision': 0.7130339539978094, 'recall': 0.8046971569839307, 'f1': 0.7560975609756098, 'number': 809}	{'precision': 0.3384615384615385, 'recall': 0.3697478991596639, 'f1': 0.35341365461847385, 'number': 119}	{'precision': 0.7763975155279503, 'recall': 0.8215962441314554, 'f1': 0.7983576642335766, 'number': 1065}	0.7235	0.7878	0.7543	0.8126

Framework versions

• Transformers 4.27.1
• Pytorch 1.12.1
• Datasets 2.6.1
• Tokenizers 0.13.2