BERTInvoiceCzechV013

This model is a fine-tuned version of TomasFAV/BERTInvoiceCzechV01 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0656
• Precision: 0.8707
• Recall: 0.8816
• F1: 0.8761
• Accuracy: 0.9835

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_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	20	0.1306	0.7273	0.7581	0.7423	0.9645
No log	2.0	40	0.1137	0.7402	0.8054	0.7714	0.9678
No log	3.0	60	0.0909	0.7828	0.8369	0.8089	0.9740
No log	4.0	80	0.0766	0.8162	0.8711	0.8428	0.9785
No log	5.0	100	0.0761	0.8141	0.8858	0.8484	0.9784
No log	6.0	120	0.0785	0.7960	0.8850	0.8382	0.9769
No log	7.0	140	0.0662	0.8520	0.8722	0.8620	0.9817
No log	8.0	160	0.0722	0.8378	0.8765	0.8567	0.9810
No log	9.0	180	0.0712	0.8250	0.8827	0.8529	0.9801
No log	10.0	200	0.0670	0.8544	0.8819	0.8680	0.9823
No log	11.0	220	0.0663	0.8518	0.8909	0.8709	0.9828
No log	12.0	240	0.0680	0.8341	0.8843	0.8584	0.9809
No log	13.0	260	0.0656	0.8704	0.8816	0.8759	0.9835
No log	14.0	280	0.0655	0.8566	0.8885	0.8723	0.9827
No log	15.0	300	0.0659	0.8466	0.8831	0.8645	0.9822
No log	16.0	320	0.0662	0.8483	0.8862	0.8669	0.9821
No log	17.0	340	0.0689	0.8402	0.8885	0.8637	0.9815
No log	18.0	360	0.0662	0.8566	0.8905	0.8732	0.9829
No log	19.0	380	0.0670	0.8519	0.8893	0.8702	0.9824
No log	20.0	400	0.0663	0.8541	0.8885	0.8710	0.9825

Framework versions

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