LiLTInvoiceCzechV03

This model is a fine-tuned version of TomasFAV/LiLTInvoiceCzechV0 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0473
• Precision: 0.8752
• Recall: 0.8976
• F1: 0.8863
• Accuracy: 0.9899

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	0.0974	0.7395	0.7218	0.7306	0.9775
No log	2.0	24	0.0776	0.6830	0.7867	0.7312	0.9764
No log	3.0	36	0.0662	0.7488	0.7884	0.7681	0.9809
No log	4.0	48	0.0576	0.7648	0.8823	0.8193	0.9836
No log	5.0	60	0.0498	0.825	0.8447	0.8347	0.9862
No log	6.0	72	0.0495	0.8102	0.8379	0.8238	0.9861
No log	7.0	84	0.0500	0.8078	0.9181	0.8594	0.9871
No log	8.0	96	0.0454	0.8629	0.8805	0.8716	0.9890
No log	9.0	108	0.0444	0.8479	0.8942	0.8704	0.9892
No log	10.0	120	0.0467	0.8344	0.9113	0.8711	0.9887
No log	11.0	132	0.0457	0.8509	0.8959	0.8728	0.9892
No log	12.0	144	0.0450	0.8553	0.8976	0.8759	0.9893
No log	13.0	156	0.0463	0.8719	0.8942	0.8829	0.9897
No log	14.0	168	0.0474	0.8555	0.8993	0.8769	0.9894
No log	15.0	180	0.0468	0.8765	0.8959	0.8861	0.9897
No log	16.0	192	0.0473	0.8752	0.8976	0.8863	0.9899
No log	17.0	204	0.0467	0.8731	0.8925	0.8827	0.9896
No log	18.0	216	0.0473	0.8709	0.8976	0.8840	0.9897
No log	19.0	228	0.0474	0.8746	0.8925	0.8834	0.9897
No log	20.0	240	0.0473	0.8763	0.8942	0.8851	0.9897

Framework versions

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