Introduction

Awesome-Rubrics is a survey-oriented portal for organizing rubric-related research and resources.

What this site is for

This site is designed to help readers navigate the rubric literature from a research perspective:

• rubric generation and refinement
• rubric-based evaluation and judging
• rubric-guided training and reward modeling
• domain-specific versus general-purpose rubric design

Main organizing axes

The current structure follows several recurring distinctions in the literature:

• use case: evaluation, inference-time verification, and data generation
• model stage: training-time, inference-time, and evaluation-time
• generation strategy: direct generation, retrieval-augmented generation, preference-driven extraction, and refinement
• task setting: general-purpose tasks versus domain-specific tasks

Suggested next steps

You can expand the site by editing or adding Markdown files under docs/.

• Continue the reading list in docs/awesome-list.md
• Add taxonomy notes or survey summaries as separate pages
• Keep source PDFs and raw materials under doc/