Arjun Virk

I build generalist policies and share my learnings here | ML @bracketbot | Researcher @UCLA's BAIR Lab | Software Engineering @UWaterloo

[Arjun Virk @virkvarjun]

Day 9/30: Learning RL in Robotics

Today, I'm proud to release my new research: FAACT (Failure-Aware Action Chunking Transformer).

Through lots of testing and design, I was able to create an embedding space for my model to understand failure states and exploit ACT to self-correct its motion.

Today, I demoed this at @newsystems_ to @join_ef and @KrishivThakuria, @thenadsusanto

More to come in the next few days at @socraticainfo...

[Arjun Virk @virkvarjun]

@PeterDiamandis 👀

[Arjun Virk @virkvarjun]

@BrookeaJoseph @jack_ceroni How can I join?

[Arjun Virk @virkvarjun]

When you touch grass in SF

@FarzaTV

[Arjun Virk @virkvarjun]

@lucashjin @bracketbot @sincethestudy ur the goat fr man

[Arjun Virk @virkvarjun]

Life Update: I've moved to SF to build the future of robotics learning @bracketbot with @sincethestudy.

My research focuses on unlocking continual learning for robotics policies.

More soon.

[Arjun Virk @virkvarjun]

@RamV1520 @bracketbot @sincethestudy w's

[Arjun Virk @virkvarjun]

@SaivivekVe43174 @bracketbot @sincethestudy ur the real inspo man

[Arjun Virk @virkvarjun]

@patjaival @bracketbot @sincethestudy 🫡

[Arjun Virk @virkvarjun]

@NaomiWalch @bracketbot @sincethestudy it's gonna be sick

[Arjun Virk @virkvarjun]

@TheMingjie @bracketbot @sincethestudy 💪

[Arjun Virk @virkvarjun]

@pranavcmarthi @bracketbot @sincethestudy thanks! good luck at corgi :)

[Arjun Virk @virkvarjun]

@aayanr07 @bracketbot @sincethestudy nah ur the goat man

[Arjun Virk @virkvarjun]

@vkleban @bracketbot @sincethestudy Let's chat

[Arjun Virk @virkvarjun]

@joshuajerinn @bracketbot @sincethestudy dm me

[Arjun Virk @virkvarjun]

@fdotinc SF

[Prabhdeep @PrabhdeepS_]

Every ANN method either gets slower as your corpus grows or starts retrieving the wrong things.

We built one that does neither.

50K vectors → 10 µs / query
1M vectors → 10 µs / query
5M vectors → 10 µs / query

Introducing MeshRAG: retrieve anything with the same speed, same accuracy, no matter how much data you have.

Here is how we did it 👇

[Arjun Virk @virkvarjun]

@_tenZdhon_ 🔥

[Thinking Machines @thinkymachines]

People talk, listen, watch, think, and collaborate at the same time, in real time. We've designed an AI that works with people the same way.

We share our approach, early results, and a quick look at our model in action.

https://t.co/AFJZ5kH7Ku

[Arjun Virk @virkvarjun]

@tejasthind4 let's grab coffee?

[Arjun Virk @virkvarjun]

@HyperLogistix @Andercot @bscholl @EdwardMehr @jordannoone @jimbelosic @aphysicist Signed up thanks