layoutlmv3-finetuned-cord_500

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: 0.2339
  • Precision: 0.9509
  • Recall: 0.9573
  • F1: 0.9541
  • Accuracy: 0.9610

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: 4000

Training results

Training Loss Epoch Step Validation Loss Precision Recall F1 Accuracy
No log 2.5 250 0.9950 0.7114 0.7784 0.7434 0.7903
1.3831 5.0 500 0.5152 0.8483 0.8787 0.8632 0.8816
1.3831 7.5 750 0.3683 0.9013 0.9154 0.9083 0.9240
0.3551 10.0 1000 0.3051 0.9201 0.9304 0.9252 0.9363
0.3551 12.5 1250 0.2636 0.9375 0.9424 0.9399 0.9457
0.1562 15.0 1500 0.2498 0.9385 0.9476 0.9430 0.9508
0.1562 17.5 1750 0.2380 0.9414 0.9499 0.9456 0.9559
0.0863 20.0 2000 0.2355 0.9400 0.9491 0.9445 0.9542
0.0863 22.5 2250 0.2268 0.9451 0.9536 0.9493 0.9601
0.0512 25.0 2500 0.2277 0.9429 0.9513 0.9471 0.9588
0.0512 27.5 2750 0.2315 0.9473 0.9551 0.9512 0.9593
0.0358 30.0 3000 0.2294 0.9509 0.9573 0.9541 0.9605
0.0358 32.5 3250 0.2330 0.9458 0.9543 0.9501 0.9593
0.028 35.0 3500 0.2374 0.9487 0.9558 0.9523 0.9597
0.028 37.5 3750 0.2374 0.9501 0.9558 0.9530 0.9593
0.0244 40.0 4000 0.2339 0.9509 0.9573 0.9541 0.9610

Framework versions

  • Transformers 4.21.2
  • Pytorch 1.12.1+cu113
  • Datasets 2.4.0
  • Tokenizers 0.12.1
Downloads last month
29
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Model tree for rajistics/layoutlmv3-finetuned-cord_500

Finetunes
1 model

Evaluation results