de Rodrigo PRO

de-Rodrigo

AI & ML interests

Synthetic Datasets, Multimodal LLMs, Computer Vision

Organizations

The Hidden Gallery's profile picture CICLAB Comillas ICAI's profile picture

Posts 1

view post
Post
994
A few weeks ago, we uploaded the MERIT Dataset ๐ŸŽ’๐Ÿ“ƒ๐Ÿ† into Hugging Face ๐Ÿค—!

Now, we are excited to share the Merit Dataset paper via arXiv! ๐Ÿ“ƒ๐Ÿ’ซ
The MERIT Dataset: Modelling and Efficiently Rendering Interpretable Transcripts (2409.00447)

The MERIT Dataset is a fully synthetic, labeled dataset created for training and benchmarking LLMs on Visually Rich Document Understanding tasks. It is also designed to help detect biases and improve interpretability in LLMs, where we are actively working. ๐Ÿ”ง๐Ÿ”จ

MERIT contains synthetically rendered students' transcripts of records from different schools in English and Spanish. We plan to expand the dataset into different contexts (synth medical/insurance documents, synth IDS, etc.) Want to collaborate? Do you have any feedback? ๐Ÿง

Resources:

- Dataset: de-Rodrigo/merit
- Code and generation pipeline: https://github.com/nachoDRT/MERIT-Dataset

PD: We are grateful to Hugging Face ๐Ÿค— for providing the fantastic tools and resources we find in the platform and, more specifically, to @nielsr for sharing the fine-tuning/inference scripts we have used in our benchmark.