|
--- |
|
base_model: Ransaka/sinhala-ocr-model |
|
tags: |
|
- generated_from_trainer |
|
model-index: |
|
- name: sinhala-ocr-model-v2 |
|
results: [] |
|
--- |
|
|
|
<!-- 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. --> |
|
|
|
# sinhala-ocr-model-v2 |
|
|
|
This model is a fine-tuned version of [Ransaka/sinhala-ocr-model](https://huggingface.co/Ransaka/sinhala-ocr-model) on an unknown dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 6.2306 |
|
- Cer: 0.5161 |
|
|
|
## 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: 8e-05 |
|
- train_batch_size: 4 |
|
- eval_batch_size: 4 |
|
- seed: 42 |
|
- gradient_accumulation_steps: 2 |
|
- total_train_batch_size: 8 |
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
- lr_scheduler_type: linear |
|
- training_steps: 6000 |
|
- mixed_precision_training: Native AMP |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | Cer | |
|
|:-------------:|:-----:|:----:|:---------------:|:------:| |
|
| 4.543 | 3.27 | 500 | 6.2682 | 0.7086 | |
|
| 2.6146 | 6.54 | 1000 | 5.8348 | 0.6390 | |
|
| 1.8448 | 9.8 | 1500 | 5.8076 | 0.6166 | |
|
| 1.3887 | 13.07 | 2000 | 6.0250 | 0.6072 | |
|
| 1.0271 | 16.34 | 2500 | 5.9971 | 0.5707 | |
|
| 0.8891 | 19.61 | 3000 | 5.9803 | 0.5630 | |
|
| 0.6548 | 22.88 | 3500 | 6.0045 | 0.5542 | |
|
| 0.4939 | 26.14 | 4000 | 6.0223 | 0.5354 | |
|
| 0.322 | 29.41 | 4500 | 6.1360 | 0.5233 | |
|
| 0.2459 | 32.68 | 5000 | 6.1166 | 0.5220 | |
|
| 0.123 | 35.95 | 5500 | 6.1740 | 0.5162 | |
|
| 0.1575 | 39.22 | 6000 | 6.2306 | 0.5161 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.35.2 |
|
- Pytorch 2.0.0 |
|
- Datasets 2.16.0 |
|
- Tokenizers 0.15.0 |
|
|
