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: 0.7076
• Answer: {'precision': 0.7112831858407079, 'recall': 0.7948084054388134, 'f1': 0.7507297139521306, 'number': 809}
• Header: {'precision': 0.2932330827067669, 'recall': 0.3277310924369748, 'f1': 0.30952380952380953, 'number': 119}
• Question: {'precision': 0.768303186907838, 'recall': 0.8375586854460094, 'f1': 0.8014375561545374, 'number': 1065}
• Overall Precision: 0.7161
• Overall Recall: 0.7898
• Overall F1: 0.7511
• Overall Accuracy: 0.8076

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
1.8402	1.0	10	1.5987	{'precision': 0.021546261089987327, 'recall': 0.021013597033374538, 'f1': 0.02127659574468085, 'number': 809}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 119}	{'precision': 0.18303571428571427, 'recall': 0.11549295774647887, 'f1': 0.14162348877374784, 'number': 1065}	0.0958	0.0702	0.0811	0.3696
1.4639	2.0	20	1.2465	{'precision': 0.11695906432748537, 'recall': 0.09888751545117429, 'f1': 0.10716677829872738, 'number': 809}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 119}	{'precision': 0.4292786421499293, 'recall': 0.5699530516431925, 'f1': 0.48971359419120614, 'number': 1065}	0.3275	0.3447	0.3359	0.5870
1.1099	3.0	30	0.9764	{'precision': 0.4803921568627451, 'recall': 0.5451174289245982, 'f1': 0.5107122177185871, 'number': 809}	{'precision': 0.1282051282051282, 'recall': 0.04201680672268908, 'f1': 0.06329113924050632, 'number': 119}	{'precision': 0.5776255707762558, 'recall': 0.7126760563380282, 'f1': 0.6380832282471627, 'number': 1065}	0.5306	0.6046	0.5652	0.7145
0.8446	4.0	40	0.7999	{'precision': 0.6241900647948164, 'recall': 0.7144622991347342, 'f1': 0.6662824207492796, 'number': 809}	{'precision': 0.23880597014925373, 'recall': 0.13445378151260504, 'f1': 0.17204301075268816, 'number': 119}	{'precision': 0.629976580796253, 'recall': 0.7577464788732394, 'f1': 0.6879795396419438, 'number': 1065}	0.6161	0.7030	0.6567	0.7649
0.6679	5.0	50	0.7382	{'precision': 0.6525515743756786, 'recall': 0.7428924598269468, 'f1': 0.6947976878612716, 'number': 809}	{'precision': 0.2891566265060241, 'recall': 0.20168067226890757, 'f1': 0.23762376237623764, 'number': 119}	{'precision': 0.6767441860465117, 'recall': 0.819718309859155, 'f1': 0.7414012738853504, 'number': 1065}	0.6530	0.7516	0.6989	0.7784
0.5711	6.0	60	0.7029	{'precision': 0.6705882352941176, 'recall': 0.7750309023485785, 'f1': 0.7190366972477065, 'number': 809}	{'precision': 0.3150684931506849, 'recall': 0.19327731092436976, 'f1': 0.23958333333333334, 'number': 119}	{'precision': 0.7184873949579832, 'recall': 0.8028169014084507, 'f1': 0.7583148558758316, 'number': 1065}	0.6847	0.7551	0.7182	0.7926
0.5019	7.0	70	0.6781	{'precision': 0.6785714285714286, 'recall': 0.7750309023485785, 'f1': 0.7236006924408541, 'number': 809}	{'precision': 0.3263157894736842, 'recall': 0.2605042016806723, 'f1': 0.2897196261682243, 'number': 119}	{'precision': 0.7317686504610227, 'recall': 0.819718309859155, 'f1': 0.7732506643046944, 'number': 1065}	0.6921	0.7682	0.7282	0.7984
0.4481	8.0	80	0.6777	{'precision': 0.678609062170706, 'recall': 0.796044499381953, 'f1': 0.732650739476678, 'number': 809}	{'precision': 0.29523809523809524, 'recall': 0.2605042016806723, 'f1': 0.2767857142857143, 'number': 119}	{'precision': 0.7263763352506163, 'recall': 0.8300469483568075, 'f1': 0.7747589833479404, 'number': 1065}	0.6865	0.7822	0.7312	0.8020
0.3933	9.0	90	0.6705	{'precision': 0.7040704070407041, 'recall': 0.7911001236093943, 'f1': 0.7450523864959255, 'number': 809}	{'precision': 0.30275229357798167, 'recall': 0.2773109243697479, 'f1': 0.28947368421052627, 'number': 119}	{'precision': 0.7557643040136636, 'recall': 0.8309859154929577, 'f1': 0.7915921288014313, 'number': 1065}	0.7117	0.7817	0.7451	0.8054
0.387	10.0	100	0.6786	{'precision': 0.6973969631236443, 'recall': 0.7948084054388134, 'f1': 0.7429231658001155, 'number': 809}	{'precision': 0.26717557251908397, 'recall': 0.29411764705882354, 'f1': 0.28, 'number': 119}	{'precision': 0.7676240208877284, 'recall': 0.828169014084507, 'f1': 0.7967479674796749, 'number': 1065}	0.7084	0.7827	0.7437	0.8098
0.3298	11.0	110	0.6932	{'precision': 0.6968716289104638, 'recall': 0.7985166872682324, 'f1': 0.7442396313364056, 'number': 809}	{'precision': 0.2824427480916031, 'recall': 0.31092436974789917, 'f1': 0.29600000000000004, 'number': 119}	{'precision': 0.7559121621621622, 'recall': 0.8403755868544601, 'f1': 0.7959092930191196, 'number': 1065}	0.7038	0.7918	0.7452	0.8026
0.3049	12.0	120	0.6984	{'precision': 0.7032115171650055, 'recall': 0.7849196538936959, 'f1': 0.7418224299065419, 'number': 809}	{'precision': 0.30833333333333335, 'recall': 0.31092436974789917, 'f1': 0.3096234309623431, 'number': 119}	{'precision': 0.7710320901994796, 'recall': 0.8347417840375587, 'f1': 0.8016230838593327, 'number': 1065}	0.7174	0.7832	0.7489	0.8083
0.293	13.0	130	0.7063	{'precision': 0.706140350877193, 'recall': 0.796044499381953, 'f1': 0.7484020918070889, 'number': 809}	{'precision': 0.2932330827067669, 'recall': 0.3277310924369748, 'f1': 0.30952380952380953, 'number': 119}	{'precision': 0.7664359861591695, 'recall': 0.831924882629108, 'f1': 0.7978388113462405, 'number': 1065}	0.7129	0.7873	0.7482	0.8041
0.2732	14.0	140	0.7057	{'precision': 0.7161862527716186, 'recall': 0.7985166872682324, 'f1': 0.7551139684395091, 'number': 809}	{'precision': 0.2932330827067669, 'recall': 0.3277310924369748, 'f1': 0.30952380952380953, 'number': 119}	{'precision': 0.7714285714285715, 'recall': 0.8366197183098592, 'f1': 0.8027027027027026, 'number': 1065}	0.7196	0.7908	0.7535	0.8080
0.2762	15.0	150	0.7076	{'precision': 0.7112831858407079, 'recall': 0.7948084054388134, 'f1': 0.7507297139521306, 'number': 809}	{'precision': 0.2932330827067669, 'recall': 0.3277310924369748, 'f1': 0.30952380952380953, 'number': 119}	{'precision': 0.768303186907838, 'recall': 0.8375586854460094, 'f1': 0.8014375561545374, 'number': 1065}	0.7161	0.7898	0.7511	0.8076

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.0+cu121
• Datasets 3.0.0
• Tokenizers 0.19.1