layoutlm-Synthetic-only

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: 0.9766
• Eader: {'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 57}
• Nswer: {'precision': 0.07159353348729793, 'recall': 0.2198581560283688, 'f1': 0.10801393728222997, 'number': 141}
• Uestion: {'precision': 0.1038135593220339, 'recall': 0.30434782608695654, 'f1': 0.15481832543443919, 'number': 161}
• Overall Precision: 0.0880
• Overall Recall: 0.2228
• Overall F1: 0.1262
• Overall Accuracy: 0.6103

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: 9
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Eader	Nswer	Uestion	Overall Precision	Overall Recall	Overall F1	Overall Accuracy
1.3476	1.0	4	1.3017	{'precision': 0.01, 'recall': 0.05263157894736842, 'f1': 0.016806722689075633, 'number': 57}	{'precision': 0.012711864406779662, 'recall': 0.0425531914893617, 'f1': 0.019575856443719414, 'number': 141}	{'precision': 0.015772870662460567, 'recall': 0.062111801242236024, 'f1': 0.025157232704402514, 'number': 161}	0.0135	0.0529	0.0215	0.3592
1.0607	2.0	8	1.2217	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 57}	{'precision': 0.015384615384615385, 'recall': 0.02127659574468085, 'f1': 0.017857142857142856, 'number': 141}	{'precision': 0.010050251256281407, 'recall': 0.012422360248447204, 'f1': 0.011111111111111113, 'number': 161}	0.0127	0.0139	0.0133	0.3607
0.8532	3.0	12	1.1632	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 57}	{'precision': 0.034375, 'recall': 0.07801418439716312, 'f1': 0.047722342733188726, 'number': 141}	{'precision': 0.021671826625386997, 'recall': 0.043478260869565216, 'f1': 0.02892561983471074, 'number': 161}	0.0280	0.0501	0.0359	0.3963
0.7208	4.0	16	1.1060	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 57}	{'precision': 0.02895752895752896, 'recall': 0.10638297872340426, 'f1': 0.04552352048558422, 'number': 141}	{'precision': 0.0380952380952381, 'recall': 0.12422360248447205, 'f1': 0.05830903790087465, 'number': 161}	0.0336	0.0975	0.0499	0.4848
0.6082	5.0	20	1.0625	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 57}	{'precision': 0.040229885057471264, 'recall': 0.14893617021276595, 'f1': 0.06334841628959276, 'number': 141}	{'precision': 0.06554307116104868, 'recall': 0.21739130434782608, 'f1': 0.10071942446043164, 'number': 161}	0.0530	0.1560	0.0792	0.5349
0.4981	6.0	24	1.0294	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 57}	{'precision': 0.04573804573804574, 'recall': 0.15602836879432624, 'f1': 0.0707395498392283, 'number': 141}	{'precision': 0.08695652173913043, 'recall': 0.2732919254658385, 'f1': 0.13193403298350825, 'number': 161}	0.0667	0.1838	0.0979	0.5663
0.416	7.0	28	1.0031	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 57}	{'precision': 0.05908096280087528, 'recall': 0.19148936170212766, 'f1': 0.09030100334448161, 'number': 141}	{'precision': 0.09475806451612903, 'recall': 0.2919254658385093, 'f1': 0.1430745814307458, 'number': 161}	0.0774	0.2061	0.1125	0.5868
0.3618	8.0	32	0.9854	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 57}	{'precision': 0.06919642857142858, 'recall': 0.2198581560283688, 'f1': 0.10526315789473685, 'number': 141}	{'precision': 0.10103092783505155, 'recall': 0.30434782608695654, 'f1': 0.15170278637770898, 'number': 161}	0.0855	0.2228	0.1236	0.6034
0.3256	9.0	36	0.9766	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 57}	{'precision': 0.07159353348729793, 'recall': 0.2198581560283688, 'f1': 0.10801393728222997, 'number': 141}	{'precision': 0.1038135593220339, 'recall': 0.30434782608695654, 'f1': 0.15481832543443919, 'number': 161}	0.0880	0.2228	0.1262	0.6103

Framework versions

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