Ge Li Board Memberships:
Scripps Research Institute
Wuxi Biologics Cayman Inc.

"WuXi Biologics (Cayman) Inc. (藥明生物技術有限公司*), an exempted company incorporated in the Cayman Islands with limited liability on February 27, 2014"

https://t.co/jgZ7ZdmSKs
Gerald Chan is also on the Scripps Research Institute Board of Directors - @WhereIsYanLing

https://t.co/zoUeDYXX6m
Gerald Chan is co-founder of Morningside and brother of Ronnie Chan. @WhereIsYanLing
https://t.co/b1aE3KBhPV
Gerald Chan, who is on the Scripps Research Institute Board of Directors, is a trustee of Fudan University in Shanghai.

https://t.co/NNNrxmoWpt

https://t.co/bjJpc6jYF4

More from All

How can we use language supervision to learn better visual representations for robotics?

Introducing Voltron: Language-Driven Representation Learning for Robotics!

Paper: https://t.co/gIsRPtSjKz
Models: https://t.co/NOB3cpATYG
Evaluation: https://t.co/aOzQu95J8z

🧵👇(1 / 12)


Videos of humans performing everyday tasks (Something-Something-v2, Ego4D) offer a rich and diverse resource for learning representations for robotic manipulation.

Yet, an underused part of these datasets are the rich, natural language annotations accompanying each video. (2/12)

The Voltron framework offers a simple way to use language supervision to shape representation learning, building off of prior work in representations for robotics like MVP (
https://t.co/Pb0mk9hb4i) and R3M (https://t.co/o2Fkc3fP0e).

The secret is *balance* (3/12)

Starting with a masked autoencoder over frames from these video clips, make a choice:

1) Condition on language and improve our ability to reconstruct the scene.

2) Generate language given the visual representation and improve our ability to describe what's happening. (4/12)

By trading off *conditioning* and *generation* we show that we can learn 1) better representations than prior methods, and 2) explicitly shape the balance of low and high-level features captured.

Why is the ability to shape this balance important? (5/12)

You May Also Like