ArOCR / README.md
gagan3012's picture
Update README.md
868c2ca
metadata
tags:
  - image-to-text
language: ar
model-index:
  - name: ArOCR
    results:
      - task:
          name: Optical Charater Recogntion
          type: image-to-text
        metrics:
          - name: Test CER
            type: cer
            value: 0.02

ArOCR

This model is a fine-tuned version of on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 0.0407
  • Cer: 0.0200

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: 5e-05
  • train_batch_size: 8
  • eval_batch_size: 8
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 5
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Cer
1.6164 0.59 1000 1.4109 0.5793
0.3434 1.18 2000 0.3876 0.2176
0.1679 1.77 3000 0.2262 0.1186
0.0816 2.37 4000 0.1274 0.0634
0.0421 2.96 5000 0.0817 0.0381
0.0067 3.55 6000 0.0520 0.0265
0.0044 4.14 7000 0.0469 0.0215
0.0027 4.73 8000 0.0407 0.0200

Framework versions

  • Transformers 4.18.0
  • Pytorch 1.9.1
  • Datasets 2.1.0
  • Tokenizers 0.11.6