Skip to content

v1.0.0: DC-Jev-RL (#1 MuJoCo v5 Leaderboard SOTA Preprint & Benchmark Suite)

Choose a tag to compare

View all tags
@rainyflash rainyflash released this 17 Oct 16:56
· 78 commits to master since this release
v1.0.0
4612ebf

Official v1.0.0 release of DC-Jev-RL (Beyond Brittle Code and Noisy Tokens: System-One Decision Models as Calibrated Dynamic-Curriculum Reward Judges for Continuous Control) by Ziyang Zhang (张子阳).

  • #1 Leaderboard SOTA across 5 canonical MuJoCo v5 benchmarks (Hopper-v5, HalfCheetah-v5, Walker2d-v5, Swimmer-v5, Reacher-v5) with 2.471x Mean ONS, 2.076x IQM ONS, and 100% (5/5) win rate over Human Oracle rewards.
  • 4.0x faster single-prefill /v1/systemone trajectory evaluation on NVIDIA L4 GPU (213.2 ms vs. 852.8 ms autoregressive LLM-Judge) powered by @jaredpalmer's TypeSafe AI Jev / Kev-4B architecture.

Contributors

jaredpalmer
Assets 6
Loading