layoutlm-funsd

This model is a fine-tuned version of pabloma09/layoutlm-funsd on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.5379
• Eader: {'precision': 0.7209302325581395, 'recall': 0.543859649122807, 'f1': 0.6200000000000001, 'number': 57}
• Nswer: {'precision': 0.7183098591549296, 'recall': 0.723404255319149, 'f1': 0.7208480565371025, 'number': 141}
• Uestion: {'precision': 0.7290322580645161, 'recall': 0.7018633540372671, 'f1': 0.7151898734177216, 'number': 161}
• Overall Precision: 0.7235
• Overall Recall: 0.6852
• Overall F1: 0.7039
• Overall Accuracy: 0.9016

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
0.0751	1.0	12	0.4989	{'precision': 0.5740740740740741, 'recall': 0.543859649122807, 'f1': 0.5585585585585585, 'number': 57}	{'precision': 0.673202614379085, 'recall': 0.7304964539007093, 'f1': 0.7006802721088436, 'number': 141}	{'precision': 0.6666666666666666, 'recall': 0.6708074534161491, 'f1': 0.6687306501547988, 'number': 161}	0.6558	0.6741	0.6648	0.8675
0.0681	2.0	24	0.4233	{'precision': 0.6739130434782609, 'recall': 0.543859649122807, 'f1': 0.6019417475728156, 'number': 57}	{'precision': 0.7394366197183099, 'recall': 0.7446808510638298, 'f1': 0.7420494699646644, 'number': 141}	{'precision': 0.7044025157232704, 'recall': 0.6956521739130435, 'f1': 0.7, 'number': 161}	0.7147	0.6908	0.7025	0.9004
0.0499	3.0	36	0.4571	{'precision': 0.775, 'recall': 0.543859649122807, 'f1': 0.6391752577319588, 'number': 57}	{'precision': 0.7083333333333334, 'recall': 0.723404255319149, 'f1': 0.7157894736842105, 'number': 141}	{'precision': 0.73125, 'recall': 0.7267080745341615, 'f1': 0.7289719626168223, 'number': 161}	0.7267	0.6964	0.7112	0.8998
0.037	4.0	48	0.4636	{'precision': 0.7045454545454546, 'recall': 0.543859649122807, 'f1': 0.613861386138614, 'number': 57}	{'precision': 0.7142857142857143, 'recall': 0.7446808510638298, 'f1': 0.7291666666666666, 'number': 141}	{'precision': 0.7222222222222222, 'recall': 0.7267080745341615, 'f1': 0.7244582043343654, 'number': 161}	0.7167	0.7047	0.7107	0.9016
0.0329	5.0	60	0.5128	{'precision': 0.6530612244897959, 'recall': 0.5614035087719298, 'f1': 0.6037735849056605, 'number': 57}	{'precision': 0.697986577181208, 'recall': 0.7375886524822695, 'f1': 0.7172413793103447, 'number': 141}	{'precision': 0.6706586826347305, 'recall': 0.6956521739130435, 'f1': 0.6829268292682926, 'number': 161}	0.6795	0.6908	0.6851	0.8880
0.0263	6.0	72	0.5192	{'precision': 0.6904761904761905, 'recall': 0.5087719298245614, 'f1': 0.5858585858585859, 'number': 57}	{'precision': 0.7183098591549296, 'recall': 0.723404255319149, 'f1': 0.7208480565371025, 'number': 141}	{'precision': 0.7484276729559748, 'recall': 0.7391304347826086, 'f1': 0.7437500000000001, 'number': 161}	0.7289	0.6964	0.7123	0.8995
0.023	7.0	84	0.5452	{'precision': 0.6976744186046512, 'recall': 0.5263157894736842, 'f1': 0.6, 'number': 57}	{'precision': 0.7202797202797203, 'recall': 0.7304964539007093, 'f1': 0.7253521126760565, 'number': 141}	{'precision': 0.7, 'recall': 0.6956521739130435, 'f1': 0.6978193146417445, 'number': 161}	0.7081	0.6825	0.6950	0.8956
0.0205	8.0	96	0.5398	{'precision': 0.6666666666666666, 'recall': 0.5614035087719298, 'f1': 0.6095238095238096, 'number': 57}	{'precision': 0.7083333333333334, 'recall': 0.723404255319149, 'f1': 0.7157894736842105, 'number': 141}	{'precision': 0.7151898734177216, 'recall': 0.7018633540372671, 'f1': 0.7084639498432601, 'number': 161}	0.7057	0.6880	0.6968	0.8971
0.0182	9.0	108	0.5025	{'precision': 0.62, 'recall': 0.543859649122807, 'f1': 0.5794392523364487, 'number': 57}	{'precision': 0.7482014388489209, 'recall': 0.7375886524822695, 'f1': 0.7428571428571428, 'number': 141}	{'precision': 0.7088607594936709, 'recall': 0.6956521739130435, 'f1': 0.7021943573667712, 'number': 161}	0.7118	0.6880	0.6997	0.9046
0.0175	10.0	120	0.5017	{'precision': 0.6888888888888889, 'recall': 0.543859649122807, 'f1': 0.6078431372549019, 'number': 57}	{'precision': 0.7183098591549296, 'recall': 0.723404255319149, 'f1': 0.7208480565371025, 'number': 141}	{'precision': 0.7133757961783439, 'recall': 0.6956521739130435, 'f1': 0.7044025157232704, 'number': 161}	0.7122	0.6825	0.6970	0.9031
0.0157	11.0	132	0.5034	{'precision': 0.7272727272727273, 'recall': 0.5614035087719298, 'f1': 0.6336633663366337, 'number': 57}	{'precision': 0.7357142857142858, 'recall': 0.7304964539007093, 'f1': 0.7330960854092528, 'number': 141}	{'precision': 0.7243589743589743, 'recall': 0.7018633540372671, 'f1': 0.7129337539432177, 'number': 161}	0.7294	0.6908	0.7096	0.9037
0.0151	12.0	144	0.5181	{'precision': 0.7209302325581395, 'recall': 0.543859649122807, 'f1': 0.6200000000000001, 'number': 57}	{'precision': 0.7183098591549296, 'recall': 0.723404255319149, 'f1': 0.7208480565371025, 'number': 141}	{'precision': 0.7290322580645161, 'recall': 0.7018633540372671, 'f1': 0.7151898734177216, 'number': 161}	0.7235	0.6852	0.7039	0.9040
0.0122	13.0	156	0.5368	{'precision': 0.7209302325581395, 'recall': 0.543859649122807, 'f1': 0.6200000000000001, 'number': 57}	{'precision': 0.7394366197183099, 'recall': 0.7446808510638298, 'f1': 0.7420494699646644, 'number': 141}	{'precision': 0.7261146496815286, 'recall': 0.7080745341614907, 'f1': 0.7169811320754716, 'number': 161}	0.7310	0.6964	0.7133	0.9019
0.0114	14.0	168	0.5372	{'precision': 0.7272727272727273, 'recall': 0.5614035087719298, 'f1': 0.6336633663366337, 'number': 57}	{'precision': 0.7272727272727273, 'recall': 0.7375886524822695, 'f1': 0.7323943661971831, 'number': 141}	{'precision': 0.7197452229299363, 'recall': 0.7018633540372671, 'f1': 0.7106918238993711, 'number': 161}	0.7238	0.6936	0.7084	0.9022
0.0126	15.0	180	0.5379	{'precision': 0.7209302325581395, 'recall': 0.543859649122807, 'f1': 0.6200000000000001, 'number': 57}	{'precision': 0.7183098591549296, 'recall': 0.723404255319149, 'f1': 0.7208480565371025, 'number': 141}	{'precision': 0.7290322580645161, 'recall': 0.7018633540372671, 'f1': 0.7151898734177216, 'number': 161}	0.7235	0.6852	0.7039	0.9016

Framework versions

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