Layoutlmv3InvoiceCzech

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

• Loss: 0.0649
• Precision: 0.8992
• Recall: 0.9155
• F1: 0.9072
• Accuracy: 0.9862

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: 1e-05
• train_batch_size: 16
• eval_batch_size: 2
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
No log	1.0	43	1.5391	0.0	0.0	0.0	0.8323
No log	2.0	86	1.0087	0.0	0.0	0.0	0.8323
No log	3.0	129	0.6795	0.0739	0.0338	0.0464	0.8375
No log	4.0	172	0.5549	0.2447	0.2802	0.2613	0.8690
No log	5.0	215	0.4249	0.4758	0.5821	0.5236	0.9183
No log	6.0	258	0.3194	0.5484	0.6836	0.6086	0.9374
No log	7.0	301	0.2431	0.6235	0.7379	0.6759	0.9502
No log	8.0	344	0.1812	0.7525	0.8116	0.7809	0.9673
No log	9.0	387	0.1548	0.7991	0.8357	0.8170	0.9705
No log	10.0	430	0.1248	0.8182	0.8478	0.8327	0.9758
No log	11.0	473	0.1104	0.8451	0.8696	0.8571	0.9786
0.64	12.0	516	0.0965	0.8506	0.8732	0.8617	0.9804
0.64	13.0	559	0.0909	0.8688	0.8877	0.8781	0.9820
0.64	14.0	602	0.0819	0.8826	0.8986	0.8905	0.9837
0.64	15.0	645	0.0779	0.8897	0.9058	0.8977	0.9850
0.64	16.0	688	0.0734	0.8942	0.9082	0.9011	0.9852
0.64	17.0	731	0.0700	0.8955	0.9106	0.9030	0.9856
0.64	18.0	774	0.0666	0.8988	0.9118	0.9053	0.9861
0.64	19.0	817	0.0655	0.8992	0.9155	0.9072	0.9862
0.64	20.0	860	0.0649	0.8992	0.9155	0.9072	0.9862

Framework versions

• Transformers 4.57.6
• Pytorch 2.9.0+cu126
• Datasets 4.0.0
• Tokenizers 0.22.2