<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
CUP
This model is a fine-tuned version of microsoft/layoutlmv3-base on the sroie dataset. It achieves the following results on the evaluation set:
- Loss: 0.0166
- Precision: 0.9674
- Recall: 0.9952
- F1: 0.9811
- Accuracy: 0.9978
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
- training_steps: 1000
Training results
| Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1 | Accuracy |
|---|---|---|---|---|---|---|---|
| No log | 1.47 | 100 | 0.0258 | 0.8851 | 0.9952 | 0.9369 | 0.9905 |
| No log | 2.94 | 200 | 0.0183 | 0.92 | 0.9904 | 0.9539 | 0.9941 |
| No log | 4.41 | 300 | 0.0143 | 0.9674 | 0.9952 | 0.9811 | 0.9978 |
| No log | 5.88 | 400 | 0.0166 | 0.9498 | 0.9952 | 0.9720 | 0.9963 |
| 0.0211 | 7.35 | 500 | 0.0179 | 0.9412 | 0.9952 | 0.9674 | 0.9956 |
| 0.0211 | 8.82 | 600 | 0.0161 | 0.9674 | 0.9952 | 0.9811 | 0.9978 |
| 0.0211 | 10.29 | 700 | 0.0168 | 0.9674 | 0.9952 | 0.9811 | 0.9978 |
| 0.0211 | 11.76 | 800 | 0.0168 | 0.9674 | 0.9952 | 0.9811 | 0.9978 |
| 0.0211 | 13.24 | 900 | 0.0170 | 0.9674 | 0.9952 | 0.9811 | 0.9978 |
| 0.0003 | 14.71 | 1000 | 0.0166 | 0.9674 | 0.9952 | 0.9811 | 0.9978 |
Framework versions
- Transformers 4.26.0.dev0
- Pytorch 1.13.0+cu116
- Datasets 2.2.2
- Tokenizers 0.13.2
