LiLTInvoiceCzechV013

This model is a fine-tuned version of TomasFAV/LiLTInvoiceCzechV01 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0467
• Precision: 0.8824
• Recall: 0.8959
• F1: 0.8891
• Accuracy: 0.9907

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: 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_steps: 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	12	0.0851	0.7551	0.7577	0.7564	0.9790
No log	2.0	24	0.0668	0.7643	0.7526	0.7584	0.9801
No log	3.0	36	0.0664	0.8217	0.7867	0.8038	0.9833
No log	4.0	48	0.0564	0.7759	0.8567	0.8143	0.9842
No log	5.0	60	0.0501	0.8368	0.8140	0.8253	0.9866
No log	6.0	72	0.0444	0.8571	0.8601	0.8586	0.9886
No log	7.0	84	0.0435	0.8503	0.9113	0.8797	0.9896
No log	8.0	96	0.0444	0.8610	0.8771	0.8690	0.9893
No log	9.0	108	0.0431	0.8756	0.8891	0.8823	0.9904
No log	10.0	120	0.0441	0.8669	0.9113	0.8885	0.9906
No log	11.0	132	0.0450	0.8501	0.9096	0.8788	0.9897
No log	12.0	144	0.0436	0.8588	0.9027	0.8802	0.9902
No log	13.0	156	0.0434	0.8733	0.8942	0.8836	0.9905
No log	14.0	168	0.0456	0.8564	0.8959	0.8757	0.9900
No log	15.0	180	0.0451	0.8725	0.8993	0.8857	0.9907
No log	16.0	192	0.0444	0.8842	0.8857	0.8849	0.9908
No log	17.0	204	0.0451	0.8807	0.8942	0.8874	0.9908
No log	18.0	216	0.0466	0.87	0.8908	0.8803	0.9904
No log	19.0	228	0.0468	0.8807	0.8942	0.8874	0.9906
No log	20.0	240	0.0467	0.8824	0.8959	0.8891	0.9907

Framework versions

• Transformers 5.0.0
• Pytorch 2.10.0+cu128
• Datasets 4.0.0
• Tokenizers 0.22.2