<!-- 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. -->
REPROCESO
This model is a fine-tuned version of microsoft/layoutlmv3-large on the sroie dataset. It achieves the following results on the evaluation set:
- Loss: 0.0069
- Precision: 0.9667
- Recall: 0.9667
- F1: 0.9667
- Accuracy: 0.9991
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 | 7.14 | 100 | 0.0048 | 0.9667 | 0.9667 | 0.9667 | 0.9991 |
| No log | 14.29 | 200 | 0.0054 | 0.9667 | 0.9667 | 0.9667 | 0.9991 |
| No log | 21.43 | 300 | 0.0058 | 0.9667 | 0.9667 | 0.9667 | 0.9991 |
| No log | 28.57 | 400 | 0.0061 | 0.9667 | 0.9667 | 0.9667 | 0.9991 |
| 0.0095 | 35.71 | 500 | 0.0063 | 0.9667 | 0.9667 | 0.9667 | 0.9991 |
| 0.0095 | 42.86 | 600 | 0.0065 | 0.9667 | 0.9667 | 0.9667 | 0.9991 |
| 0.0095 | 50.0 | 700 | 0.0067 | 0.9667 | 0.9667 | 0.9667 | 0.9991 |
| 0.0095 | 57.14 | 800 | 0.0068 | 0.9667 | 0.9667 | 0.9667 | 0.9991 |
| 0.0095 | 64.29 | 900 | 0.0069 | 0.9667 | 0.9667 | 0.9667 | 0.9991 |
| 0.0 | 71.43 | 1000 | 0.0069 | 0.9667 | 0.9667 | 0.9667 | 0.9991 |
Framework versions
- Transformers 4.25.0.dev0
- Pytorch 1.12.1+cu113
- Datasets 2.2.2
- Tokenizers 0.13.2
