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: 0.7672
• Answer: {'precision': 0.7376681614349776, 'recall': 0.8133498145859085, 'f1': 0.7736625514403291, 'number': 809}
• Header: {'precision': 0.37857142857142856, 'recall': 0.44537815126050423, 'f1': 0.4092664092664093, 'number': 119}
• Question: {'precision': 0.8122109158186864, 'recall': 0.8244131455399061, 'f1': 0.8182665424044735, 'number': 1065}
• Overall Precision: 0.7520
• Overall Recall: 0.7973
• Overall F1: 0.7740
• Overall Accuracy: 0.8111

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: 8
• 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
1.6896	1.0	19	1.4068	{'precision': 0.0572737686139748, 'recall': 0.06180469715698393, 'f1': 0.05945303210463733, 'number': 809}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 119}	{'precision': 0.4044834307992203, 'recall': 0.38967136150234744, 'f1': 0.3969392635102822, 'number': 1065}	0.2449	0.2333	0.2390	0.4540
1.1462	2.0	38	0.8626	{'precision': 0.5445075757575758, 'recall': 0.7107540173053152, 'f1': 0.6166219839142091, 'number': 809}	{'precision': 0.04878048780487805, 'recall': 0.01680672268907563, 'f1': 0.025, 'number': 119}	{'precision': 0.6331615120274914, 'recall': 0.692018779342723, 'f1': 0.661283086585913, 'number': 1065}	0.5812	0.6593	0.6178	0.7285
0.7687	3.0	57	0.7051	{'precision': 0.6217264791464597, 'recall': 0.792336217552534, 'f1': 0.6967391304347825, 'number': 809}	{'precision': 0.18421052631578946, 'recall': 0.11764705882352941, 'f1': 0.14358974358974358, 'number': 119}	{'precision': 0.6805084745762712, 'recall': 0.7539906103286385, 'f1': 0.7153674832962139, 'number': 1065}	0.6375	0.7316	0.6813	0.7678
0.5961	4.0	76	0.6423	{'precision': 0.6527918781725889, 'recall': 0.7948084054388134, 'f1': 0.7168338907469342, 'number': 809}	{'precision': 0.26785714285714285, 'recall': 0.25210084033613445, 'f1': 0.2597402597402597, 'number': 119}	{'precision': 0.7100949094046591, 'recall': 0.7727699530516432, 'f1': 0.7401079136690648, 'number': 1065}	0.6631	0.7506	0.7042	0.7923
0.4739	5.0	95	0.6263	{'precision': 0.7024972855591748, 'recall': 0.799752781211372, 'f1': 0.7479768786127167, 'number': 809}	{'precision': 0.33070866141732286, 'recall': 0.35294117647058826, 'f1': 0.34146341463414637, 'number': 119}	{'precision': 0.7435897435897436, 'recall': 0.8169014084507042, 'f1': 0.778523489932886, 'number': 1065}	0.7029	0.7822	0.7404	0.8059
0.3893	6.0	114	0.6456	{'precision': 0.6912681912681913, 'recall': 0.8220024721878862, 'f1': 0.7509881422924901, 'number': 809}	{'precision': 0.28169014084507044, 'recall': 0.33613445378151263, 'f1': 0.30651340996168586, 'number': 119}	{'precision': 0.768270944741533, 'recall': 0.8093896713615023, 'f1': 0.7882944673068131, 'number': 1065}	0.7040	0.7863	0.7428	0.8050
0.3267	7.0	133	0.6813	{'precision': 0.7297605473204105, 'recall': 0.7911001236093943, 'f1': 0.7591933570581256, 'number': 809}	{'precision': 0.33070866141732286, 'recall': 0.35294117647058826, 'f1': 0.34146341463414637, 'number': 119}	{'precision': 0.7810283687943262, 'recall': 0.8272300469483568, 'f1': 0.8034655722754217, 'number': 1065}	0.7331	0.7842	0.7578	0.8019
0.2731	8.0	152	0.6628	{'precision': 0.7153846153846154, 'recall': 0.8046971569839307, 'f1': 0.7574171029668412, 'number': 809}	{'precision': 0.3697478991596639, 'recall': 0.3697478991596639, 'f1': 0.3697478991596639, 'number': 119}	{'precision': 0.7877145438121048, 'recall': 0.8187793427230047, 'f1': 0.8029465930018417, 'number': 1065}	0.7336	0.7863	0.7590	0.8137
0.2425	9.0	171	0.6992	{'precision': 0.7209302325581395, 'recall': 0.8046971569839307, 'f1': 0.7605140186915887, 'number': 809}	{'precision': 0.3851851851851852, 'recall': 0.4369747899159664, 'f1': 0.4094488188976378, 'number': 119}	{'precision': 0.801980198019802, 'recall': 0.8366197183098592, 'f1': 0.8189338235294118, 'number': 1065}	0.7417	0.7998	0.7697	0.8104
0.2145	10.0	190	0.7271	{'precision': 0.7373167981961668, 'recall': 0.8084054388133498, 'f1': 0.7712264150943396, 'number': 809}	{'precision': 0.36075949367088606, 'recall': 0.4789915966386555, 'f1': 0.41155234657039713, 'number': 119}	{'precision': 0.8250950570342205, 'recall': 0.8150234741784037, 'f1': 0.8200283419933868, 'number': 1065}	0.7530	0.7923	0.7721	0.8047
0.1882	11.0	209	0.7348	{'precision': 0.7400681044267877, 'recall': 0.8059332509270705, 'f1': 0.7715976331360946, 'number': 809}	{'precision': 0.375, 'recall': 0.453781512605042, 'f1': 0.4106463878326997, 'number': 119}	{'precision': 0.8254716981132075, 'recall': 0.8215962441314554, 'f1': 0.8235294117647057, 'number': 1065}	0.7583	0.7933	0.7754	0.8103
0.1668	12.0	228	0.7541	{'precision': 0.7360178970917226, 'recall': 0.8133498145859085, 'f1': 0.7727539635936582, 'number': 809}	{'precision': 0.3984375, 'recall': 0.42857142857142855, 'f1': 0.41295546558704455, 'number': 119}	{'precision': 0.8007246376811594, 'recall': 0.8300469483568075, 'f1': 0.8151221761180267, 'number': 1065}	0.7493	0.7993	0.7735	0.8097
0.1595	13.0	247	0.7616	{'precision': 0.7370786516853932, 'recall': 0.8108776266996292, 'f1': 0.7722189523248971, 'number': 809}	{'precision': 0.38345864661654133, 'recall': 0.42857142857142855, 'f1': 0.4047619047619047, 'number': 119}	{'precision': 0.8153988868274582, 'recall': 0.8253521126760563, 'f1': 0.8203453103126458, 'number': 1065}	0.7549	0.7958	0.7748	0.8122
0.1451	14.0	266	0.7638	{'precision': 0.7361894024802705, 'recall': 0.8071693448702101, 'f1': 0.7700471698113207, 'number': 809}	{'precision': 0.375886524822695, 'recall': 0.44537815126050423, 'f1': 0.40769230769230774, 'number': 119}	{'precision': 0.8150557620817844, 'recall': 0.8234741784037559, 'f1': 0.8192433442316676, 'number': 1065}	0.7524	0.7943	0.7728	0.8110
0.1449	15.0	285	0.7672	{'precision': 0.7376681614349776, 'recall': 0.8133498145859085, 'f1': 0.7736625514403291, 'number': 809}	{'precision': 0.37857142857142856, 'recall': 0.44537815126050423, 'f1': 0.4092664092664093, 'number': 119}	{'precision': 0.8122109158186864, 'recall': 0.8244131455399061, 'f1': 0.8182665424044735, 'number': 1065}	0.7520	0.7973	0.7740	0.8111

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.1+cu121
• Datasets 2.21.0
• Tokenizers 0.19.1