LLM Compass

English

简体中文

English

简体中文

Appearance

DR Tulu: Reinforcement Learning with Evolving Rubrics ​

Translation status: The English translation is pending. The section structure is synchronized with the Chinese source.

Motivation: Rewarding Open-Ended Long-Form Research ​

Method: Search-Augmented Evolving Rubrics ​

RLER Overview ​

Initialization with Search Results ​

Rubric Scoring ​

Rubric-Buffer Management ​

Training Recipe ​

Results ​

Evaluation Setup ​

Main Results ​

Cost ​

GeneticDiseasesQA Case Study ​

Short-Form QA Generalization ​

Ablations ​

Position in the Deep-Research Landscape ​