icdar23-entrydetector_plaintext_breaks
This model is a fine-tuned version of HueyNemud/das22-10-camembert_pretrained on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.0120
- Ebegin: {'precision': 0.9901997738409348, 'recall': 0.9879654005265137, 'f1': 0.9890813253012049, 'number': 2659}
- Eend: {'precision': 0.9916824196597354, 'recall': 0.9801943198804185, 'f1': 0.9859049050930276, 'number': 2676}
- Overall Precision: 0.9909
- Overall Recall: 0.9841
- Overall F1: 0.9875
- Overall Accuracy: 0.9977
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: 0.0001
- 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: 7500
Training results
Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1 | Accuracy |
---|---|---|---|---|---|---|---|
No log | 0.07 | 300 | 0.0318 | 0.9847 | 0.9761 | 0.9803 | 0.9966 |
0.1683 | 0.14 | 600 | 0.0164 | 0.9878 | 0.9890 | 0.9884 | 0.9978 |
0.1683 | 0.21 | 900 | 0.0146 | 0.9900 | 0.9853 | 0.9876 | 0.9976 |
0.0203 | 0.29 | 1200 | 0.0112 | 0.9862 | 0.9902 | 0.9882 | 0.9978 |
0.0123 | 0.36 | 1500 | 0.0089 | 0.9943 | 0.9878 | 0.9910 | 0.9983 |
0.0123 | 0.43 | 1800 | 0.0139 | 0.9970 | 0.9814 | 0.9891 | 0.9979 |
0.0109 | 0.5 | 2100 | 0.0101 | 0.9937 | 0.9882 | 0.9909 | 0.9982 |
0.0109 | 0.57 | 2400 | 0.0087 | 0.9949 | 0.9896 | 0.9922 | 0.9985 |
0.0092 | 0.64 | 2700 | 0.0081 | 0.9849 | 0.9919 | 0.9884 | 0.9978 |
0.0084 | 0.72 | 3000 | 0.0087 | 0.9937 | 0.9867 | 0.9902 | 0.9981 |
0.0084 | 0.79 | 3300 | 0.0089 | 0.9915 | 0.9889 | 0.9902 | 0.9981 |
0.0069 | 0.86 | 3600 | 0.0092 | 0.9899 | 0.9901 | 0.9900 | 0.9981 |
0.0069 | 0.93 | 3900 | 0.0097 | 0.9845 | 0.9915 | 0.9880 | 0.9977 |
Framework versions
- Transformers 4.26.1
- Pytorch 1.13.1+cu116
- Datasets 2.9.0
- Tokenizers 0.13.2
- Downloads last month
- 5
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.