Inkaso_beta

This model is a fine-tuned version of microsoft/layoutlm-base-uncased on the layoutlmv3 dataset. It achieves the following results on the evaluation set:

• Loss: 0.0801
• Creditor address: {'precision': 1.0, 'recall': 0.875, 'f1': 0.9333333333333333, 'number': 48}
• Creditor name: {'precision': 1.0, 'recall': 1.0, 'f1': 1.0, 'number': 34}
• Creditor proxy: {'precision': 0.8333333333333334, 'recall': 0.8108108108108109, 'f1': 0.8219178082191781, 'number': 37}
• Debtor address: {'precision': 0.9636363636363636, 'recall': 1.0, 'f1': 0.9814814814814815, 'number': 53}
• Debtor name: {'precision': 0.9428571428571428, 'recall': 1.0, 'f1': 0.9705882352941176, 'number': 33}
• Doc id: {'precision': 0.85, 'recall': 0.8947368421052632, 'f1': 0.8717948717948718, 'number': 19}
• Title: {'precision': 1.0, 'recall': 1.0, 'f1': 1.0, 'number': 34}
• Overall Precision: 0.9492
• Overall Recall: 0.9419
• Overall F1: 0.9455
• Overall Accuracy: 0.9831

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: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 10
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Creditor address	Creditor name	Creditor proxy	Debtor address	Debtor name	Doc id	Title	Overall Precision	Overall Recall	Overall F1	Overall Accuracy
0.4642	6.6667	20	0.2502	{'precision': 0.782608695652174, 'recall': 0.75, 'f1': 0.7659574468085107, 'number': 48}	{'precision': 0.9354838709677419, 'recall': 0.8529411764705882, 'f1': 0.8923076923076922, 'number': 34}	{'precision': 0.8, 'recall': 0.6486486486486487, 'f1': 0.7164179104477612, 'number': 37}	{'precision': 0.8205128205128205, 'recall': 0.6037735849056604, 'f1': 0.6956521739130435, 'number': 53}	{'precision': 0.95, 'recall': 0.5757575757575758, 'f1': 0.7169811320754716, 'number': 33}	{'precision': 1.0, 'recall': 0.2631578947368421, 'f1': 0.4166666666666667, 'number': 19}	{'precision': 0.8461538461538461, 'recall': 0.3235294117647059, 'f1': 0.46808510638297873, 'number': 34}	0.8478	0.6047	0.7059	0.9330
0.1387	13.3333	40	0.0914	{'precision': 1.0, 'recall': 0.9166666666666666, 'f1': 0.9565217391304348, 'number': 48}	{'precision': 0.9714285714285714, 'recall': 1.0, 'f1': 0.9855072463768115, 'number': 34}	{'precision': 0.7777777777777778, 'recall': 0.7567567567567568, 'f1': 0.7671232876712328, 'number': 37}	{'precision': 0.9444444444444444, 'recall': 0.9622641509433962, 'f1': 0.9532710280373832, 'number': 53}	{'precision': 0.8918918918918919, 'recall': 1.0, 'f1': 0.9428571428571428, 'number': 33}	{'precision': 0.8095238095238095, 'recall': 0.8947368421052632, 'f1': 0.8500000000000001, 'number': 19}	{'precision': 1.0, 'recall': 1.0, 'f1': 1.0, 'number': 34}	0.9234	0.9341	0.9287	0.9795
0.0431	20.0	60	0.0774	{'precision': 1.0, 'recall': 1.0, 'f1': 1.0, 'number': 48}	{'precision': 1.0, 'recall': 1.0, 'f1': 1.0, 'number': 34}	{'precision': 0.8181818181818182, 'recall': 0.7297297297297297, 'f1': 0.7714285714285715, 'number': 37}	{'precision': 0.9636363636363636, 'recall': 1.0, 'f1': 0.9814814814814815, 'number': 53}	{'precision': 0.9428571428571428, 'recall': 1.0, 'f1': 0.9705882352941176, 'number': 33}	{'precision': 0.7727272727272727, 'recall': 0.8947368421052632, 'f1': 0.8292682926829269, 'number': 19}	{'precision': 1.0, 'recall': 1.0, 'f1': 1.0, 'number': 34}	0.9425	0.9535	0.9480	0.9837
0.0216	26.6667	80	0.0842	{'precision': 1.0, 'recall': 1.0, 'f1': 1.0, 'number': 48}	{'precision': 1.0, 'recall': 1.0, 'f1': 1.0, 'number': 34}	{'precision': 0.7631578947368421, 'recall': 0.7837837837837838, 'f1': 0.7733333333333334, 'number': 37}	{'precision': 0.9454545454545454, 'recall': 0.9811320754716981, 'f1': 0.9629629629629629, 'number': 53}	{'precision': 0.9166666666666666, 'recall': 1.0, 'f1': 0.9565217391304348, 'number': 33}	{'precision': 0.8095238095238095, 'recall': 0.8947368421052632, 'f1': 0.8500000000000001, 'number': 19}	{'precision': 1.0, 'recall': 1.0, 'f1': 1.0, 'number': 34}	0.9286	0.9574	0.9427	0.9825
0.0142	33.3333	100	0.0840	{'precision': 1.0, 'recall': 0.875, 'f1': 0.9333333333333333, 'number': 48}	{'precision': 1.0, 'recall': 1.0, 'f1': 1.0, 'number': 34}	{'precision': 0.8333333333333334, 'recall': 0.8108108108108109, 'f1': 0.8219178082191781, 'number': 37}	{'precision': 0.9629629629629629, 'recall': 0.9811320754716981, 'f1': 0.9719626168224299, 'number': 53}	{'precision': 0.9166666666666666, 'recall': 1.0, 'f1': 0.9565217391304348, 'number': 33}	{'precision': 0.8095238095238095, 'recall': 0.8947368421052632, 'f1': 0.8500000000000001, 'number': 19}	{'precision': 1.0, 'recall': 1.0, 'f1': 1.0, 'number': 34}	0.9416	0.9380	0.9398	0.9819
0.0105	40.0	120	0.0838	{'precision': 0.9772727272727273, 'recall': 0.8958333333333334, 'f1': 0.9347826086956522, 'number': 48}	{'precision': 1.0, 'recall': 1.0, 'f1': 1.0, 'number': 34}	{'precision': 0.8333333333333334, 'recall': 0.8108108108108109, 'f1': 0.8219178082191781, 'number': 37}	{'precision': 0.9636363636363636, 'recall': 1.0, 'f1': 0.9814814814814815, 'number': 53}	{'precision': 0.9166666666666666, 'recall': 1.0, 'f1': 0.9565217391304348, 'number': 33}	{'precision': 0.8095238095238095, 'recall': 0.8947368421052632, 'f1': 0.8500000000000001, 'number': 19}	{'precision': 1.0, 'recall': 1.0, 'f1': 1.0, 'number': 34}	0.9385	0.9457	0.9421	0.9819
0.0081	46.6667	140	0.0801	{'precision': 1.0, 'recall': 0.875, 'f1': 0.9333333333333333, 'number': 48}	{'precision': 1.0, 'recall': 1.0, 'f1': 1.0, 'number': 34}	{'precision': 0.8333333333333334, 'recall': 0.8108108108108109, 'f1': 0.8219178082191781, 'number': 37}	{'precision': 0.9636363636363636, 'recall': 1.0, 'f1': 0.9814814814814815, 'number': 53}	{'precision': 0.9428571428571428, 'recall': 1.0, 'f1': 0.9705882352941176, 'number': 33}	{'precision': 0.85, 'recall': 0.8947368421052632, 'f1': 0.8717948717948718, 'number': 19}	{'precision': 1.0, 'recall': 1.0, 'f1': 1.0, 'number': 34}	0.9492	0.9419	0.9455	0.9831

Framework versions

• Transformers 4.40.1
• Pytorch 2.3.0+cu118
• Datasets 2.19.0
• Tokenizers 0.19.1