docuAI

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: 0.6827
• Answer: {'precision': 0.7172949002217295, 'recall': 0.799752781211372, 'f1': 0.7562828755113968, 'number': 809}
• Header: {'precision': 0.33620689655172414, 'recall': 0.3277310924369748, 'f1': 0.33191489361702126, 'number': 119}
• Question: {'precision': 0.7637931034482759, 'recall': 0.831924882629108, 'f1': 0.7964044943820225, 'number': 1065}
• Overall Precision: 0.7218
• Overall Recall: 0.7888
• Overall F1: 0.7538
• Overall Accuracy: 0.8097

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
1.7471	1.0	10	1.5418	{'precision': 0.027965284474445518, 'recall': 0.03584672435105068, 'f1': 0.0314192849404117, 'number': 809}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 119}	{'precision': 0.26978723404255317, 'recall': 0.2976525821596244, 'f1': 0.2830357142857143, 'number': 1065}	0.1564	0.1736	0.1646	0.4109
1.3886	2.0	20	1.2047	{'precision': 0.25207100591715975, 'recall': 0.26328800988875156, 'f1': 0.2575574365175332, 'number': 809}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 119}	{'precision': 0.4723969252271139, 'recall': 0.6347417840375587, 'f1': 0.5416666666666667, 'number': 1065}	0.3906	0.4461	0.4165	0.5948
1.0458	3.0	30	0.9213	{'precision': 0.4836471754212091, 'recall': 0.6032138442521632, 'f1': 0.5368536853685368, 'number': 809}	{'precision': 0.05263157894736842, 'recall': 0.008403361344537815, 'f1': 0.014492753623188406, 'number': 119}	{'precision': 0.6089795918367347, 'recall': 0.7004694835680751, 'f1': 0.651528384279476, 'number': 1065}	0.5482	0.6197	0.5817	0.7024
0.8024	4.0	40	0.7873	{'precision': 0.5814393939393939, 'recall': 0.7589616810877626, 'f1': 0.6584450402144773, 'number': 809}	{'precision': 0.1044776119402985, 'recall': 0.058823529411764705, 'f1': 0.07526881720430108, 'number': 119}	{'precision': 0.6488427773343974, 'recall': 0.7633802816901408, 'f1': 0.7014667817083693, 'number': 1065}	0.6035	0.7195	0.6564	0.7567
0.6593	5.0	50	0.7148	{'precision': 0.6419753086419753, 'recall': 0.7713226205191595, 'f1': 0.7007299270072992, 'number': 809}	{'precision': 0.2602739726027397, 'recall': 0.15966386554621848, 'f1': 0.19791666666666666, 'number': 119}	{'precision': 0.7365217391304347, 'recall': 0.7953051643192488, 'f1': 0.7647855530474039, 'number': 1065}	0.6788	0.7476	0.7116	0.7846
0.5564	6.0	60	0.6806	{'precision': 0.6945054945054945, 'recall': 0.7812113720642769, 'f1': 0.7353112274578244, 'number': 809}	{'precision': 0.2647058823529412, 'recall': 0.226890756302521, 'f1': 0.24434389140271492, 'number': 119}	{'precision': 0.7158067158067158, 'recall': 0.8206572769953052, 'f1': 0.7646544181977254, 'number': 1065}	0.6865	0.7692	0.7255	0.7947
0.4838	7.0	70	0.6697	{'precision': 0.6844396082698585, 'recall': 0.7775030902348579, 'f1': 0.7280092592592592, 'number': 809}	{'precision': 0.2696629213483146, 'recall': 0.20168067226890757, 'f1': 0.23076923076923078, 'number': 119}	{'precision': 0.738626964433416, 'recall': 0.8384976525821596, 'f1': 0.7854001759014952, 'number': 1065}	0.6973	0.7757	0.7344	0.8004
0.4342	8.0	80	0.6709	{'precision': 0.7062780269058296, 'recall': 0.7787391841779975, 'f1': 0.7407407407407407, 'number': 809}	{'precision': 0.30097087378640774, 'recall': 0.2605042016806723, 'f1': 0.27927927927927926, 'number': 119}	{'precision': 0.7415359207266722, 'recall': 0.8431924882629108, 'f1': 0.7891036906854131, 'number': 1065}	0.7067	0.7822	0.7426	0.8020
0.386	9.0	90	0.6592	{'precision': 0.7107258938244854, 'recall': 0.8108776266996292, 'f1': 0.7575057736720554, 'number': 809}	{'precision': 0.2764227642276423, 'recall': 0.2857142857142857, 'f1': 0.2809917355371901, 'number': 119}	{'precision': 0.75, 'recall': 0.8253521126760563, 'f1': 0.7858739383102369, 'number': 1065}	0.7074	0.7873	0.7452	0.8106
0.3572	10.0	100	0.6611	{'precision': 0.7080213903743315, 'recall': 0.8182941903584673, 'f1': 0.7591743119266056, 'number': 809}	{'precision': 0.29906542056074764, 'recall': 0.2689075630252101, 'f1': 0.28318584070796454, 'number': 119}	{'precision': 0.7532133676092545, 'recall': 0.8253521126760563, 'f1': 0.7876344086021505, 'number': 1065}	0.7121	0.7893	0.7487	0.8102
0.3264	11.0	110	0.6828	{'precision': 0.7325056433408578, 'recall': 0.8022249690976514, 'f1': 0.7657817109144542, 'number': 809}	{'precision': 0.3, 'recall': 0.3025210084033613, 'f1': 0.301255230125523, 'number': 119}	{'precision': 0.7525773195876289, 'recall': 0.8225352112676056, 'f1': 0.7860026917900403, 'number': 1065}	0.7194	0.7832	0.7499	0.8055
0.3132	12.0	120	0.6722	{'precision': 0.7123893805309734, 'recall': 0.796044499381953, 'f1': 0.7518972562755399, 'number': 809}	{'precision': 0.3391304347826087, 'recall': 0.3277310924369748, 'f1': 0.3333333333333333, 'number': 119}	{'precision': 0.7548605240912933, 'recall': 0.8384976525821596, 'f1': 0.7944839857651246, 'number': 1065}	0.7157	0.7908	0.7514	0.8082
0.293	13.0	130	0.6817	{'precision': 0.7109634551495017, 'recall': 0.7935723114956736, 'f1': 0.75, 'number': 809}	{'precision': 0.3277310924369748, 'recall': 0.3277310924369748, 'f1': 0.3277310924369748, 'number': 119}	{'precision': 0.7680776014109347, 'recall': 0.8178403755868544, 'f1': 0.7921782628467485, 'number': 1065}	0.7199	0.7787	0.7481	0.8078
0.282	14.0	140	0.6845	{'precision': 0.712707182320442, 'recall': 0.7972805933250927, 'f1': 0.7526254375729288, 'number': 809}	{'precision': 0.3333333333333333, 'recall': 0.3277310924369748, 'f1': 0.3305084745762712, 'number': 119}	{'precision': 0.768624014022787, 'recall': 0.8234741784037559, 'f1': 0.7951042611060744, 'number': 1065}	0.7217	0.7832	0.7512	0.8094
0.2728	15.0	150	0.6827	{'precision': 0.7172949002217295, 'recall': 0.799752781211372, 'f1': 0.7562828755113968, 'number': 809}	{'precision': 0.33620689655172414, 'recall': 0.3277310924369748, 'f1': 0.33191489361702126, 'number': 119}	{'precision': 0.7637931034482759, 'recall': 0.831924882629108, 'f1': 0.7964044943820225, 'number': 1065}	0.7218	0.7888	0.7538	0.8097

Framework versions

• Transformers 4.33.2
• Pytorch 2.0.1+cu118
• Datasets 2.14.5
• Tokenizers 0.13.3