README.md

metadata

tags:
  - generated_from_trainer
datasets:
  - mp-02/funsd
metrics:
  - precision
  - recall
  - f1
  - accuracy
model-index:
  - name: layoutlmv3-finetuned-funsd
    results:
      - task:
          name: Token Classification
          type: token-classification
        dataset:
          name: mp-02/funsd
          type: mp-02/funsd
        metrics:
          - name: Precision
            type: precision
            value: 0.875725338491296
          - name: Recall
            type: recall
            value: 0.9055
          - name: F1
            type: f1
            value: 0.8903638151425762
          - name: Accuracy
            type: accuracy
            value: 0.843706936150666

layoutlmv3-finetuned-funsd

This model is a fine-tuned version of layoutlmv3 on the mp-02/funsd dataset. It achieves the following results on the evaluation set:

• Loss: 0.6187
• Precision: 0.8757
• Recall: 0.9055
• F1: 0.8904
• Accuracy: 0.8437

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: 4
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• training_steps: 500

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
No log	1.32	50	0.9063	0.7006	0.757	0.7277	0.7607
No log	2.63	100	0.6387	0.7930	0.858	0.8242	0.7967
No log	3.95	150	0.5691	0.8171	0.8825	0.8486	0.8254
No log	5.26	200	0.5723	0.8315	0.881	0.8555	0.8223
No log	6.58	250	0.5897	0.8475	0.9	0.8729	0.8292
No log	7.89	300	0.6122	0.8482	0.9025	0.8745	0.8283
No log	9.21	350	0.6045	0.8505	0.899	0.8741	0.8392
No log	10.53	400	0.5662	0.8708	0.9	0.8852	0.8446
No log	11.84	450	0.5973	0.8739	0.9045	0.8889	0.8437
0.4305	13.16	500	0.6187	0.8757	0.9055	0.8904	0.8437

Framework versions

• Transformers 4.12.5
• Pytorch 1.10.0+cu111
• Datasets 2.13.2
• Tokenizers 0.10.1