Layoutv3test

This model is a fine-tuned version of microsoft/layoutlmv3-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.9405
• F1: 0.7563
• Recall: 0.6959
• Precision: 0.8281
• Pred Bestellnummer: 146
• Percentage Pred Act Bestellnummer: 1.0210
• Pred Kundennr.: 57
• Percentage Pred Act Kundennr.: 1.1875
• Pred Bezug 1: 35
• Percentage Pred Act Bezug 1: 2.5
• Pred Modell 1: 114
• Percentage Pred Act Modell 1: 1.1515
• Pred Menge1: 74
• Percentage Pred Act Menge1: 3.5238
• Pred Möbelhaus: 93
• Percentage Pred Act Möbelhaus: 1.0220
• Pred Termin kundenwunsch - kw: 30
• Percentage Pred Act Termin kundenwunsch - kw: 0.9375
• Pred Kommission: 60
• Percentage Pred Act Kommission: 1.0345
• Pred Holz 1: 14
• Percentage Pred Act Holz 1: 0.7368
• Pred Modell 2: 57
• Percentage Pred Act Modell 2: 0.9194
• Pred Zusatz 1: 11
• Percentage Pred Act Zusatz 1: 0.7857
• Pred Holz 2: 39
• Percentage Pred Act Holz 2: 1.8571
• Pred Modell 3: 72
• Percentage Pred Act Modell 3: 1.0909
• Pred Var-ausf 1: 6
• Percentage Pred Act Var-ausf 1: 0.75
• Pred Menge3: 1
• Percentage Pred Act Menge3: 0.0455
• Act Bestellnummer: 143
• Act Kundennr.: 48
• Act Bezug 1: 14
• Act Modell 1: 99
• Act Menge1: 21
• Act Menge4: 10
• Act Möbelhaus: 91
• Act Bezug 2: 13
• Act Zusatz 2: 1
• Act Termin kundenwunsch - kw: 32
• Act Kommission: 58
• Act Holz 1: 19
• Act Menge3: 22
• Act Modell 2: 62
• Act Modell 3: 66
• Act Modell 4: 6
• Act Bezug 4: 7
• Act Zusatz 3: 1
• Act Holz 2: 21
• Act Menge2: 18
• Act Bezug 3: 4
• Act Var-ausf 1: 8
• Act Holz 3: 5
• Act Zusatz 1: 14
• Act Var-ausf. 2: 7
• Act Var-ausf. 3: 4
• Act Pv 3: 1
• Act Holz 4: 1
• Act Var-ausf. 5: 1
• Act Modell 5: 5
• Act La-anschrift: 6
• Act Menge5: 1

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 OptimizerNames.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: 5
• mixed_precision_training: Native AMP

Training results

Framework versions

• Transformers 4.52.4
• Pytorch 2.7.0+cu126
• Datasets 3.6.0
• Tokenizers 0.21.1