Uni-RLHF
Public platform page for Uni-RLHF, emphasizing the benchmark, interface, and reproducible workflow for RLHF experimentation.
Related publication: Uni-RLHF: Universal Platform and Benchmark Suite for Reinforcement Learning with Diverse Human Feedback · The Twelfth International Conference on Learning Representations
Uni-RLHF is best understood as infrastructure: a common platform for setting up, comparing, and reproducing RLHF pipelines under one interface instead of scattered experiment scripts.
This page focuses on the platform and benchmark experience: how tasks, feedback sources, and evaluation workflows are organized for actual use rather than described only at the paper level.
Use the project site if you want to see the system framing and public-facing structure first. Use the paper page if you want the research motivation behind the benchmark suite.
