Log inSign up
Zichen Liu profile banner
user avatar

Zichen Liu

@zzlccc
Gemini RL @GoogleDeepMind
Singapore
lkevinzc.github.io
Joined October 2021
459
Following
6,023
Followers
RepliesMedia
  • Pinned
    user avatar
    Zichen Liu
    @zzlccc
    Mar 21, 2025
    🪂Understanding R1-Zero-Like Training: A Critical Perspective * DeepSeek-V3-Base already exhibits "Aha moment" before RL-tuning?? * The ever-increasing output length in RL-tuning might be due to a BIAS in GRPO?? * Getting GRPO Done Right, we achieve a 7B AIME sota! 🧵 📜Full
    Image
  • user avatar
    Zichen Liu
    @zzlccc
    Apr 10
    rl intuition (up-scaled by the correctness of infra) is all you need when cooking with a strong base model such as gemini✨
  • user avatar
    Zichen Liu
    @zzlccc
    Mar 17
    🦎🦎 Happy to see two of our works (DrGRPO & DPPO) are highlighted here! I don’t think changing a few terms is worth a new branding, so we respectfully kept predecessors’ name while highlighting the correction/improvement on top of them. Hopefully they inspire RL algo designs.
    user avatar
    Alex Weers
    @a_weers
    Mar 15
    Finally finished! If you're interested in an overview of recent methods in reinforcement learning for reasoning LLMs, check out this blog post: aweers.de/blog/2026/rl-f… It summarizes ten methods, tries to highlight differences and trends, and has a collection of open problems
    Blog post on the current state of reinforcement learning for reasoning LLMs