IFBench

IFBench: Dataset for evaluating instruction-following reward models This repository contains the data of the paper "Agentic Reward Modeling: Integrating Human Preferences with Verifiable Correctness Signals for Reliable Reward Systems" Paper: https://arxiv.org/abs/2502.19328 GitHub: https://github.com/THU-KEG/Agentic-R

Papers using IFBench (2)

Scaling the Horizon, Not the Parameters: Reaching Trillion-Parameter Performance with a 35B Agent2026Mach-Mind-4-Flash Technical Report2026