CORD Donut evaluation is run on validation dataset

NielsRogge / Transformers-Tutorials

This repository contains demos I made with the Transformers library by HuggingFace.

MIT License

8.45k stars 1.32k forks source link

CORD Donut evaluation is run on validation dataset #358

Open PierreMesure opened 8 months ago

PierreMesure commented 8 months ago

Hi @NielsRogge,

Thank you so much for your work!

When using your notebook for CORD Donut fine-tuning, I realised that you are evaluating the fine-tuned model on the evaluation dataset and not the test one. I'm sending a PR to fix it.