layoutlm-with-funsd

This model is a fine-tuned version of pabloma09/layoutlm-with-funsd on the funsd dataset. It achieves the following results on the evaluation set:

• Loss: 0.6344
• Eader: {'precision': 0.4888888888888889, 'recall': 0.38596491228070173, 'f1': 0.4313725490196078, 'number': 57}
• Nswer: {'precision': 0.577922077922078, 'recall': 0.6312056737588653, 'f1': 0.6033898305084746, 'number': 141}
• Uestion: {'precision': 0.5172413793103449, 'recall': 0.5590062111801242, 'f1': 0.537313432835821, 'number': 161}
• Overall Precision: 0.5389
• Overall Recall: 0.5599
• Overall F1: 0.5492
• Overall Accuracy: 0.8364

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 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	Eader	Nswer	Uestion	Overall Precision	Overall Recall	Overall F1	Overall Accuracy
0.3894	1.0	9	0.5238	{'precision': 0.34782608695652173, 'recall': 0.2807017543859649, 'f1': 0.3106796116504854, 'number': 57}	{'precision': 0.515527950310559, 'recall': 0.5886524822695035, 'f1': 0.5496688741721855, 'number': 141}	{'precision': 0.4010989010989011, 'recall': 0.453416149068323, 'f1': 0.4256559766763849, 'number': 161}	0.4422	0.4791	0.4599	0.8174
0.3489	2.0	18	0.5037	{'precision': 0.2978723404255319, 'recall': 0.24561403508771928, 'f1': 0.2692307692307692, 'number': 57}	{'precision': 0.5125, 'recall': 0.5815602836879432, 'f1': 0.5448504983388704, 'number': 141}	{'precision': 0.4, 'recall': 0.4472049689440994, 'f1': 0.42228739002932547, 'number': 161}	0.4341	0.4680	0.4504	0.8270
0.2657	3.0	27	0.5258	{'precision': 0.3333333333333333, 'recall': 0.2807017543859649, 'f1': 0.3047619047619048, 'number': 57}	{'precision': 0.5123456790123457, 'recall': 0.5886524822695035, 'f1': 0.5478547854785478, 'number': 141}	{'precision': 0.3901098901098901, 'recall': 0.4409937888198758, 'f1': 0.4139941690962099, 'number': 161}	0.4337	0.4735	0.4527	0.8261
0.1907	4.0	36	0.5390	{'precision': 0.38461538461538464, 'recall': 0.2631578947368421, 'f1': 0.3125, 'number': 57}	{'precision': 0.5827814569536424, 'recall': 0.624113475177305, 'f1': 0.6027397260273973, 'number': 141}	{'precision': 0.47878787878787876, 'recall': 0.4906832298136646, 'f1': 0.48466257668711654, 'number': 161}	0.5127	0.5070	0.5098	0.8286
0.175	5.0	45	0.5489	{'precision': 0.42105263157894735, 'recall': 0.2807017543859649, 'f1': 0.3368421052631579, 'number': 57}	{'precision': 0.5246913580246914, 'recall': 0.6028368794326241, 'f1': 0.561056105610561, 'number': 141}	{'precision': 0.449438202247191, 'recall': 0.4968944099378882, 'f1': 0.471976401179941, 'number': 161}	0.4788	0.5042	0.4912	0.8361
0.1685	6.0	54	0.5678	{'precision': 0.4, 'recall': 0.2807017543859649, 'f1': 0.32989690721649484, 'number': 57}	{'precision': 0.5769230769230769, 'recall': 0.6382978723404256, 'f1': 0.6060606060606061, 'number': 141}	{'precision': 0.45901639344262296, 'recall': 0.5217391304347826, 'f1': 0.4883720930232558, 'number': 161}	0.5013	0.5292	0.5149	0.8370
0.1156	7.0	63	0.5749	{'precision': 0.4864864864864865, 'recall': 0.3157894736842105, 'f1': 0.3829787234042553, 'number': 57}	{'precision': 0.50920245398773, 'recall': 0.5886524822695035, 'f1': 0.5460526315789473, 'number': 141}	{'precision': 0.43575418994413406, 'recall': 0.484472049689441, 'f1': 0.45882352941176474, 'number': 161}	0.4723	0.4986	0.4851	0.8409
0.1019	8.0	72	0.5907	{'precision': 0.43137254901960786, 'recall': 0.38596491228070173, 'f1': 0.40740740740740744, 'number': 57}	{'precision': 0.5408805031446541, 'recall': 0.6099290780141844, 'f1': 0.5733333333333333, 'number': 141}	{'precision': 0.5113636363636364, 'recall': 0.5590062111801242, 'f1': 0.5341246290801187, 'number': 161}	0.5130	0.5515	0.5315	0.8337
0.0885	9.0	81	0.5899	{'precision': 0.5, 'recall': 0.43859649122807015, 'f1': 0.46728971962616817, 'number': 57}	{'precision': 0.55, 'recall': 0.624113475177305, 'f1': 0.584717607973422, 'number': 141}	{'precision': 0.5084745762711864, 'recall': 0.5590062111801242, 'f1': 0.5325443786982249, 'number': 161}	0.5245	0.5655	0.5442	0.8400
0.0852	10.0	90	0.6170	{'precision': 0.45454545454545453, 'recall': 0.3508771929824561, 'f1': 0.396039603960396, 'number': 57}	{'precision': 0.564935064935065, 'recall': 0.6170212765957447, 'f1': 0.5898305084745763, 'number': 141}	{'precision': 0.5027932960893855, 'recall': 0.5590062111801242, 'f1': 0.5294117647058824, 'number': 161}	0.5225	0.5487	0.5353	0.8364
0.0854	11.0	99	0.6107	{'precision': 0.5111111111111111, 'recall': 0.40350877192982454, 'f1': 0.45098039215686275, 'number': 57}	{'precision': 0.5506329113924051, 'recall': 0.6170212765957447, 'f1': 0.5819397993311038, 'number': 141}	{'precision': 0.5113636363636364, 'recall': 0.5590062111801242, 'f1': 0.5341246290801187, 'number': 161}	0.5277	0.5571	0.5420	0.8358
0.0665	12.0	108	0.6090	{'precision': 0.5111111111111111, 'recall': 0.40350877192982454, 'f1': 0.45098039215686275, 'number': 57}	{'precision': 0.5365853658536586, 'recall': 0.624113475177305, 'f1': 0.5770491803278689, 'number': 141}	{'precision': 0.4946236559139785, 'recall': 0.5714285714285714, 'f1': 0.5302593659942363, 'number': 161}	0.5139	0.5655	0.5385	0.8464
0.0632	13.0	117	0.6200	{'precision': 0.44680851063829785, 'recall': 0.3684210526315789, 'f1': 0.40384615384615385, 'number': 57}	{'precision': 0.5370370370370371, 'recall': 0.6170212765957447, 'f1': 0.5742574257425743, 'number': 141}	{'precision': 0.4945054945054945, 'recall': 0.5590062111801242, 'f1': 0.5247813411078717, 'number': 161}	0.5064	0.5515	0.528	0.8412
0.0758	14.0	126	0.6326	{'precision': 0.5, 'recall': 0.38596491228070173, 'f1': 0.43564356435643564, 'number': 57}	{'precision': 0.5705128205128205, 'recall': 0.6312056737588653, 'f1': 0.5993265993265993, 'number': 141}	{'precision': 0.5142857142857142, 'recall': 0.5590062111801242, 'f1': 0.5357142857142856, 'number': 161}	0.536	0.5599	0.5477	0.8382
0.0573	15.0	135	0.6344	{'precision': 0.4888888888888889, 'recall': 0.38596491228070173, 'f1': 0.4313725490196078, 'number': 57}	{'precision': 0.577922077922078, 'recall': 0.6312056737588653, 'f1': 0.6033898305084746, 'number': 141}	{'precision': 0.5172413793103449, 'recall': 0.5590062111801242, 'f1': 0.537313432835821, 'number': 161}	0.5389	0.5599	0.5492	0.8364

Framework versions

• Transformers 4.49.0
• Pytorch 2.6.0+cu124
• Datasets 3.3.2
• Tokenizers 0.21.0