TrOCR_0216 / README.md
yoon1000's picture
End of training
412eb0d verified
metadata
base_model: microsoft/trocr-base-stage1
tags:
  - generated_from_trainer
model-index:
  - name: TrOCR_0216
    results: []

TrOCR_0216

This model is a fine-tuned version of microsoft/trocr-base-stage1 on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 1.8095
  • Cer: 0.0608

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: 10
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Cer
0.626 0.5 500 0.9106 0.1848
0.5456 1.0 1000 0.7220 0.1927
0.4588 1.5 1500 0.7064 0.1240
0.6771 2.0 2000 0.7207 0.2169
0.3778 2.5 2500 0.6689 0.1283
0.4543 3.0 3000 0.6833 0.3052
0.4428 3.5 3500 0.6604 0.0893
0.4899 4.0 4000 0.7024 0.0692
0.2548 4.5 4500 1.0137 0.1599
0.0851 5.0 5000 1.8095 0.0608

Framework versions

  • Transformers 4.35.2
  • Pytorch 2.1.1+cu121
  • Datasets 2.13.0
  • Tokenizers 0.15.0