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: 0.6864
• Answer: {'precision': 0.6979722518676628, 'recall': 0.8084054388133498, 'f1': 0.7491408934707904, 'number': 809}
• Header: {'precision': 0.296875, 'recall': 0.31932773109243695, 'f1': 0.3076923076923077, 'number': 119}
• Question: {'precision': 0.7652790079716564, 'recall': 0.8112676056338028, 'f1': 0.7876025524156791, 'number': 1065}
• Overall Precision: 0.7092
• Overall Recall: 0.7807
• Overall F1: 0.7433
• Overall Accuracy: 0.8094

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: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• 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
1.7694	1.0	10	1.6060	{'precision': 0.024282560706401765, 'recall': 0.013597033374536464, 'f1': 0.01743264659270998, 'number': 809}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 119}	{'precision': 0.3227176220806794, 'recall': 0.14272300469483568, 'f1': 0.19791666666666666, 'number': 1065}	0.1764	0.0818	0.1118	0.3371
1.456	2.0	20	1.2789	{'precision': 0.21739130434782608, 'recall': 0.315203955500618, 'f1': 0.2573158425832492, 'number': 809}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 119}	{'precision': 0.37261146496815284, 'recall': 0.5492957746478874, 'f1': 0.444022770398482, 'number': 1065}	0.3062	0.4215	0.3547	0.5669
1.1265	3.0	30	0.9633	{'precision': 0.46710526315789475, 'recall': 0.6143386897404203, 'f1': 0.5306994127068873, 'number': 809}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 119}	{'precision': 0.5227606461086637, 'recall': 0.6685446009389672, 'f1': 0.5867325916769675, 'number': 1065}	0.4984	0.6066	0.5472	0.6822
0.8681	4.0	40	0.8085	{'precision': 0.5848690591658584, 'recall': 0.7453646477132262, 'f1': 0.6554347826086957, 'number': 809}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 119}	{'precision': 0.6259607173356105, 'recall': 0.6882629107981221, 'f1': 0.6556350626118067, 'number': 1065}	0.6026	0.6703	0.6347	0.7390
0.6998	5.0	50	0.7327	{'precision': 0.61875, 'recall': 0.7342398022249691, 'f1': 0.6715658564160543, 'number': 809}	{'precision': 0.14285714285714285, 'recall': 0.07563025210084033, 'f1': 0.0989010989010989, 'number': 119}	{'precision': 0.6388676358071921, 'recall': 0.784037558685446, 'f1': 0.7040472175379427, 'number': 1065}	0.6172	0.7215	0.6653	0.7721
0.5889	6.0	60	0.6932	{'precision': 0.6111111111111112, 'recall': 0.788627935723115, 'f1': 0.6886130599028603, 'number': 809}	{'precision': 0.1791044776119403, 'recall': 0.10084033613445378, 'f1': 0.12903225806451613, 'number': 119}	{'precision': 0.7079964061096137, 'recall': 0.739906103286385, 'f1': 0.7235996326905417, 'number': 1065}	0.6466	0.7215	0.6820	0.7731
0.5103	7.0	70	0.6603	{'precision': 0.6570247933884298, 'recall': 0.7861557478368356, 'f1': 0.7158131682611143, 'number': 809}	{'precision': 0.3058823529411765, 'recall': 0.2184873949579832, 'f1': 0.2549019607843137, 'number': 119}	{'precision': 0.7271937445699392, 'recall': 0.7859154929577464, 'f1': 0.7554151624548736, 'number': 1065}	0.6801	0.7521	0.7143	0.7886
0.4557	8.0	80	0.6577	{'precision': 0.649949849548646, 'recall': 0.8009888751545118, 'f1': 0.7176079734219271, 'number': 809}	{'precision': 0.2641509433962264, 'recall': 0.23529411764705882, 'f1': 0.24888888888888888, 'number': 119}	{'precision': 0.7243150684931506, 'recall': 0.7943661971830986, 'f1': 0.7577250335871025, 'number': 1065}	0.6702	0.7637	0.7139	0.7959
0.3927	9.0	90	0.6559	{'precision': 0.6729559748427673, 'recall': 0.7935723114956736, 'f1': 0.7283040272263188, 'number': 809}	{'precision': 0.29310344827586204, 'recall': 0.2857142857142857, 'f1': 0.2893617021276596, 'number': 119}	{'precision': 0.7451838879159369, 'recall': 0.7990610328638498, 'f1': 0.7711826008155868, 'number': 1065}	0.6903	0.7662	0.7263	0.8041
0.3806	10.0	100	0.6697	{'precision': 0.6778242677824268, 'recall': 0.8009888751545118, 'f1': 0.7342776203966006, 'number': 809}	{'precision': 0.2719298245614035, 'recall': 0.2605042016806723, 'f1': 0.26609442060085836, 'number': 119}	{'precision': 0.7642418930762489, 'recall': 0.8187793427230047, 'f1': 0.7905711695376247, 'number': 1065}	0.7015	0.7782	0.7379	0.8083
0.3299	11.0	110	0.6691	{'precision': 0.6905016008537886, 'recall': 0.799752781211372, 'f1': 0.7411225658648338, 'number': 809}	{'precision': 0.30578512396694213, 'recall': 0.31092436974789917, 'f1': 0.30833333333333335, 'number': 119}	{'precision': 0.7675628794449263, 'recall': 0.8309859154929577, 'f1': 0.7980162308385933, 'number': 1065}	0.7096	0.7873	0.7464	0.8102
0.3093	12.0	120	0.6782	{'precision': 0.6955128205128205, 'recall': 0.8046971569839307, 'f1': 0.746131805157593, 'number': 809}	{'precision': 0.3008130081300813, 'recall': 0.31092436974789917, 'f1': 0.3057851239669422, 'number': 119}	{'precision': 0.7582222222222222, 'recall': 0.8009389671361502, 'f1': 0.7789954337899543, 'number': 1065}	0.7056	0.7732	0.7379	0.8096
0.2923	13.0	130	0.6818	{'precision': 0.6923890063424947, 'recall': 0.8096415327564895, 'f1': 0.7464387464387465, 'number': 809}	{'precision': 0.3217391304347826, 'recall': 0.31092436974789917, 'f1': 0.3162393162393162, 'number': 119}	{'precision': 0.7752212389380531, 'recall': 0.8225352112676056, 'f1': 0.7981776765375853, 'number': 1065}	0.7157	0.7868	0.7495	0.8091
0.2694	14.0	140	0.6849	{'precision': 0.7018299246501615, 'recall': 0.8059332509270705, 'f1': 0.7502876869965477, 'number': 809}	{'precision': 0.29133858267716534, 'recall': 0.31092436974789917, 'f1': 0.3008130081300813, 'number': 119}	{'precision': 0.7716814159292036, 'recall': 0.8187793427230047, 'f1': 0.7945330296127562, 'number': 1065}	0.7141	0.7832	0.7471	0.8097
0.2738	15.0	150	0.6864	{'precision': 0.6979722518676628, 'recall': 0.8084054388133498, 'f1': 0.7491408934707904, 'number': 809}	{'precision': 0.296875, 'recall': 0.31932773109243695, 'f1': 0.3076923076923077, 'number': 119}	{'precision': 0.7652790079716564, 'recall': 0.8112676056338028, 'f1': 0.7876025524156791, 'number': 1065}	0.7092	0.7807	0.7433	0.8094

Framework versions

• Transformers 4.49.0
• Pytorch 2.6.0+cu124
• Datasets 3.4.1
• Tokenizers 0.21.1