Zhuokai Zhao

@zhuokaiz

AI Research Scientist @Meta. Building scalable intelligence. PhD @UChicagoCS.

New York, NY

Zhuokai Zhao
@zhuokaiz
11月12日
Recently there have been lots of demos of frontier LLMs controlling robots and solving robotics tasks. Technically it makes sense, and on a high level I am a huge advocate for it. I always think the main advantage of VLAs (compared to LLMs in robots) is that they are fast and
Zhenting Qi
@ZhentingQi
11月12日
Launching the XE² Arena project: a cross-embodiment × cross-environment arena for agentic robots After @claudeai Fable 5.2’s amazing performance on RoboDojo, what's next? In XE² Arena: • Cross-embodiment: one LLM agent, on many robot bodies (YAM, xArm, Wuji Hand 2, humanoid,
Zhuokai Zhao
@zhuokaiz
11月6日
Heading out, wish I could have stayed longer. For anyone who may be interested in the talk, here are the slides. docs.google.com/presentation/d…
Zhuokai Zhao
@zhuokaiz
11月5日
+3
Very mixed feelings walking around the uiuc campus after graduating in 2016. Many things are changing so fast, but it is warm to feel the old vibe again. Thanks @YujiaZheng9 for inviting me to give a talk at the department of statistics. It could not be more special. Some
Zhuokai Zhao
@zhuokaiz
11月5日
Very mixed feelings walking around the uiuc campus after graduating in 2016. Many things are changing so fast, but it is warm to feel the old vibe again. Thanks @YujiaZheng9 for inviting me to give a talk at the department of statistics. It could not be more special. Some photos from then and now.
2
Zhuokai Zhao
@zhuokaiz
10月18日
The quantized version of Muse Spark Open runs on eight H200s, or on two 512GB Mac Studios. The last time a near-frontier model did this was DeepSeek-R1, and that was 21 months ago.
Artificial Analysis
@ArtificialAnlys
10月16日
Meta has released Muse Spark Open, an 816B (42B active) open-weights model that scores 39 on the Artificial Analysis Intelligence Index, making it the #1 US open-weights model Muse Spark Open (xhigh) scores 22 points above Muse Glimmer (high, 17), Meta's 30B dense open-weights
3
Zhuokai Zhao
@zhuokaiz
10月17日
Most multi-agent work designs the agents. However, with the current progress in foundation models, the agent is not quite the bottleneck in multi-agent systems anymore, which is why these systems often struggle to beat a single agent. They scale poorly because every agent
Zhenting Qi
@ZhentingQi
10月16日
Research index (e.g. Google Scholar) is one of the largest collaborative systems where millions of human users build on each other's work across centuries. How about we build something similar for async LLM agents? We give agents a minimal hypertextual system. They
13

Zhuokai Zhao

Zhuokai Zhaoさんのポスト