Visualize in Weights & Biases

trocr-base-handwritten-iam-Final

This model is a fine-tuned version of microsoft/trocr-base-handwritten on Teklia/IAM-line dataset. It achieves the following results on the evaluation set:

  • Loss: 1.2551
  • Cer: 0.4517
  • Wer: 0.5937

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: 4
  • eval_batch_size: 4
  • seed: 42
  • gradient_accumulation_steps: 2
  • total_train_batch_size: 8
  • optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
  • lr_scheduler_type: linear
  • num_epochs: 3
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Cer Wer
1.995 0.6169 500 2.5889 0.5503 0.7422
1.0942 1.2332 1000 1.7821 0.5084 0.7079
0.7976 1.8501 1500 1.4544 0.4628 0.6171
0.3857 2.4664 2000 1.2551 0.4517 0.5937

Framework versions

  • Transformers 4.55.0
  • Pytorch 2.6.0+cu124
  • Datasets 4.0.0
  • Tokenizers 0.21.4
Downloads last month
12
Safetensors
Model size
334M params
Tensor type
F32
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for zippbler/trocr-base-handwritten-iam-Final

Finetuned
(17)
this model

Dataset used to train zippbler/trocr-base-handwritten-iam-Final