layoutlmv3-finetuned-cord_100

This model is a fine-tuned version of microsoft/layoutlmv3-base on the cord-layoutlmv3 dataset. It achieves the following results on the evaluation set:

• Loss: 0.7215
• Precision: 0.8720
• Recall: 0.8870
• F1: 0.8794
• Accuracy: 0.8790

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: 5
• eval_batch_size: 5
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• training_steps: 2500

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
No log	12.5	250	1.0892	0.7345	0.7867	0.7597	0.7806
1.3039	25.0	500	0.7150	0.8054	0.8428	0.8237	0.8281
1.3039	37.5	750	0.6320	0.8335	0.8615	0.8473	0.8540
0.2171	50.0	1000	0.6427	0.8651	0.8832	0.8741	0.8722
0.2171	62.5	1250	0.6640	0.8672	0.8847	0.8759	0.8765
0.0654	75.0	1500	0.6758	0.8650	0.8825	0.8737	0.8731
0.0654	87.5	1750	0.7028	0.8684	0.8840	0.8761	0.8765
0.0338	100.0	2000	0.7252	0.8710	0.8847	0.8778	0.8769
0.0338	112.5	2250	0.7227	0.8710	0.8847	0.8778	0.8778
0.0257	125.0	2500	0.7215	0.8720	0.8870	0.8794	0.8790

Framework versions

• Transformers 4.21.2
• Pytorch 1.12.1+cu113
• Datasets 2.4.0
• Tokenizers 0.12.1