LayoutLMv3_large_2

This model is a fine-tuned version of BadreddineHug/LayoutLM_5 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.4678
• Precision: 0.7444
• Recall: 0.8462
• F1: 0.792
• Accuracy: 0.9431

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: 2
• eval_batch_size: 2
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• training_steps: 2000

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
No log	2.44	100	0.2604	0.8049	0.8462	0.8250	0.9487
No log	4.88	200	0.2887	0.6923	0.8462	0.7615	0.9294
No log	7.32	300	0.3961	0.6711	0.8547	0.7519	0.9248
No log	9.76	400	0.3117	0.7778	0.8376	0.8066	0.9465
0.1255	12.2	500	0.3344	0.7698	0.8291	0.7984	0.9419
0.1255	14.63	600	0.3892	0.7197	0.8120	0.7631	0.9339
0.1255	17.07	700	0.3865	0.7143	0.8547	0.7782	0.9419
0.1255	19.51	800	0.4737	0.6690	0.8291	0.7405	0.9226
0.1255	21.95	900	0.3876	0.7405	0.8291	0.7823	0.9442
0.0206	24.39	1000	0.3845	0.7444	0.8462	0.792	0.9465
0.0206	26.83	1100	0.4179	0.75	0.8205	0.7837	0.9442
0.0206	29.27	1200	0.3942	0.7576	0.8547	0.8032	0.9510
0.0206	31.71	1300	0.4521	0.7293	0.8291	0.776	0.9408
0.0206	34.15	1400	0.4725	0.7050	0.8376	0.7656	0.9328
0.0051	36.59	1500	0.4874	0.6849	0.8547	0.7605	0.9317
0.0051	39.02	1600	0.4366	0.7519	0.8547	0.8	0.9453
0.0051	41.46	1700	0.4978	0.6897	0.8547	0.7634	0.9317
0.0051	43.9	1800	0.4599	0.7444	0.8462	0.792	0.9431
0.0051	46.34	1900	0.4765	0.7372	0.8632	0.7953	0.9431
0.002	48.78	2000	0.4678	0.7444	0.8462	0.792	0.9431

Framework versions

• Transformers 4.29.2
• Pytorch 2.0.1+cu118
• Datasets 2.14.4
• Tokenizers 0.13.3