Mandi Zhao (@ZhaoMandi) / X

Mandi Zhao

169 posts

Mandi Zhao

@ZhaoMandi

PhDing @Stanford, AI & robotics. Prev @MetaAI @NVIDIAAI @berkeley_ai🐻

Joined May 2017

Pinned
Mandi Zhao
@ZhaoMandi
Jun 3, 2025
How to learn dexterous manipulation for any robot hand from a single human demonstration? Check out DexMachina, our new RL algorithm that learns long-horizon, bimanual dexterous policies for a variety of dexterous hands, articulated objects, and complex motions.
00:00
120K
Mandi Zhao
@ZhaoMandi
Dec 4, 2023
Can we talk about how inconvenient it has gotten to get a visa for Chinese students? It’s been 3 weeks since mine was “refused”. Is there any reason me, in my 2nd yr Stanford PhD, must wait for up to 180 days to be considered safe to be let in the US?
1.6M
Mandi Zhao
@ZhaoMandi
Dec 15, 2024
Thought there was a new clown in town, turns out it’s just the good old Pedro
49K
Mandi Zhao
@ZhaoMandi
Jun 14, 2024
Here’s something you didn’t know LLMs can do – reconstruct articulated objects! Introducing Real2Code – our new real2sim approach that scalably reconstructs complex, multi-part articulated objects. arxiv.org/abs/2406.08474
00:00
92K
Mandi Zhao
@ZhaoMandi
Dec 14, 2024
How hard is it to Not be racist towards exactly One nationality at a public keynote talk at a top conference? - Apparently extremely (for an MIT prof too)
Jiao Sun
@sunjiao123sun_
Dec 14, 2024
Mitigating racial bias from LLMs is a lot easier than removing it from humans! Can’t believe this happened at the best AI conference @NeurIPSConf We have ethical reviews for authors, but missed it for invited speakers? 😡
49K
Mandi Zhao
@ZhaoMandi
Jul 24, 2025
One amusing takeaway from doing RL in these massively parallelized sim environments is that reward engineering matters more so than ever. A small detail in reward function could make a huge difference: with 10k+ parallel threads to explore in, the policy Will exploit any caveat
00:00
34K
Mandi Zhao
@ZhaoMandi
Dec 4, 2023
Replying to @ZhaoMandi
You might not see many students talk about this for fear of repercussion. When your passport is taken away, all you can do is wait in silence you don’t want to mess with the roll-of-dice chance for some officer in a good mood processing your case next.
59K
Mandi Zhao
@ZhaoMandi
Dec 4, 2023
Replying to @ZhaoMandi
Things were not like this before: 7 years ago I got my 5yr F1 the literal next day (for my undergrad in Berkeley). It’s crazy how geopolitical dynamics can shift tides so quickly and affect people who don’t want anything beyond simply getting a degree and travel for conferences.
57K
Mandi Zhao
@ZhaoMandi
Feb 24, 2023
I had the pleasure to witness this project coming to shape, can confirm no one in the lab was hurt, but we were all very scared by the knife:)
AK
@_akhaliq
Feb 23, 2023
RoboNinja: Learning an Adaptive Cutting Policy for Multi-Material Objects abs: arxiv.org/abs/2302.11553 project page: roboninja.cs.columbia.edu
00:00
70K
Mandi Zhao
@ZhaoMandi
Jul 18, 2023
Excited to share RoCo: Dialectic Multi-Robot Collaboration with Large Language Models. We propose a novel approach to multi-robot collaboration that leverages LLMs for both high-level communication and low-level path planning. w/ @shreeyaajain, @SongShuran
00:00
89K
Mandi Zhao
@ZhaoMandi
Jun 16, 2025
Our lab at Stanford usually do research in AI & robotics, but very occasionally we indulge in being functional alcoholics -- Recently we hosted a lab cocktail night, and created drinks with research-related puns like 'reviewer#2' and 'make 6 figures', sharing the full recipes
27K
Mandi Zhao
@ZhaoMandi
Dec 4, 2023
Replying to @ZhaoMandi
Sure, you will get the visa eventually so no need to complain — but there’s also no need to delay hundreds of legitimate students’ lives and waste public resources when I’m pretty sure there’s a boarder crisis going on smw else.
54K
Mandi Zhao
@ZhaoMandi
Feb 27, 2023
CACTI🌵: A framework for multi-task multi-scene robot manipulation. It easily scales to many tasks, and uses text2image generative models (e.g. stable-diffusion) to augment data with visually realistic variances. w @mangahomanga @VincentMoens @SongShuran @aravindr93 @Vikashplus
00:00
55K
Mandi Zhao
@ZhaoMandi
Jan 19, 2024
We are just doing combinatorics on the hype words at this point
AK
@_akhaliq
Jan 19, 2024
ByteDance presents DiffusionGPT LLM-Driven Text-to-Image Generation System paper page: huggingface.co/papers/2401.10… propose a unified generation system DiffusionGPT, which leverages Large Language Models (LLM) to seamlessly accommodating various types of prompts input and
53K