BERTInvoiceCzechV0123Test

This model is a fine-tuned version of TomasFAV/BERTInvoiceCzechV012 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0612
• Precision: 0.8944
• Recall: 0.9177
• F1: 0.9059
• Accuracy: 0.9856

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	0.1056	0.7803	0.84	0.8091	0.9716
No log	2.0	40	0.0831	0.8105	0.8901	0.8484	0.9764
No log	3.0	60	0.0704	0.8410	0.8994	0.8692	0.9804
No log	4.0	80	0.0675	0.8403	0.9095	0.8736	0.9808
No log	5.0	100	0.0632	0.8630	0.8932	0.8779	0.9821
No log	6.0	120	0.0706	0.8319	0.9111	0.8697	0.9800
No log	7.0	140	0.0611	0.8729	0.8932	0.8829	0.9834
No log	8.0	160	0.0608	0.8754	0.9056	0.8902	0.9835
No log	9.0	180	0.0595	0.8769	0.9243	0.9000	0.9848
No log	10.0	200	0.0606	0.8759	0.9153	0.8952	0.9842
No log	11.0	220	0.0610	0.8855	0.9192	0.9021	0.9850
No log	12.0	240	0.0632	0.8720	0.9258	0.8981	0.9844
No log	13.0	260	0.0608	0.8961	0.9115	0.9037	0.9853
No log	14.0	280	0.0610	0.8953	0.9165	0.9058	0.9855
No log	15.0	300	0.0615	0.8874	0.9181	0.9025	0.9853
No log	16.0	320	0.0627	0.8841	0.9216	0.9025	0.9851
No log	17.0	340	0.0625	0.8807	0.92	0.8999	0.9847
No log	18.0	360	0.0612	0.8944	0.9177	0.9059	0.9856
No log	19.0	380	0.0619	0.8893	0.92	0.9044	0.9854
No log	20.0	400	0.0618	0.8901	0.9212	0.9053	0.9856

Framework versions

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