This is the first video that comes up (1.6 million views) https://t.co/vt6GQugkHa
My favorite example of how informationally toxic YouTube's algorithm is this:
Imagine you're high school freshman and got a school assignment about the Federal Reserve.
This is the first video that comes up (1.6 million views) https://t.co/vt6GQugkHa
Once you view that, the algorithm also then suggests this on the sidebar:
"What You're Not Supposed to Know About America's Founding" (863,00 views)
https://t.co/anOLNuPJGF
Trump Tells Everyone Exactly Who Created Illuminati (4 million views)
https://t.co/2ipMBtaphA
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)
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
1/ 👋 Excited to share what we’ve been building at https://t.co/GOQJ7LjQ2t + we are going to tweetstorm our progress every week!
Week 1 highlights: getting shortlisted for YC W2019🤞, acquiring a premium domain💰, meeting Substack's @hamishmckenzie and Stripe CEO @patrickc 🤩
2/ So what is Brew?
brew / bru : / to make (beer, coffee etc.) / verb: begin to develop 🌱
A place for you to enjoy premium content while supporting your favorite creators. Sort of like a ‘Consumer-facing Patreon’ cc @jackconte
(we’re still working on the pitch)
3/ So, why be so transparent? Two words: launch strategy.
jk 😅 a) I loooove doing something consistently for a long period of time b) limited downside and infinite upside (feedback, accountability, reach).
cc @altimor, @pmarca
4/ https://t.co/GOQJ7LjQ2t domain 🍻
It started with a cold email. Guess what? He was using BuyMeACoffee on his blog, and was excited to hear about what we're building next. Within 2w, we signed the deal at @Escrowcom's SF office. You’re a pleasure to work with @MichaelCyger!
5/ @ycombinator's invite for the in-person interview arrived that evening. Quite a day!
Thanks @patio11 for the thoughtful feedback on our YC application, and @gabhubert for your directions on positioning the product — set the tone for our pitch!
Week 1 highlights: getting shortlisted for YC W2019🤞, acquiring a premium domain💰, meeting Substack's @hamishmckenzie and Stripe CEO @patrickc 🤩
2/ So what is Brew?
brew / bru : / to make (beer, coffee etc.) / verb: begin to develop 🌱
A place for you to enjoy premium content while supporting your favorite creators. Sort of like a ‘Consumer-facing Patreon’ cc @jackconte
(we’re still working on the pitch)
3/ So, why be so transparent? Two words: launch strategy.
jk 😅 a) I loooove doing something consistently for a long period of time b) limited downside and infinite upside (feedback, accountability, reach).
cc @altimor, @pmarca
4/ https://t.co/GOQJ7LjQ2t domain 🍻
It started with a cold email. Guess what? He was using BuyMeACoffee on his blog, and was excited to hear about what we're building next. Within 2w, we signed the deal at @Escrowcom's SF office. You’re a pleasure to work with @MichaelCyger!
5/ @ycombinator's invite for the in-person interview arrived that evening. Quite a day!
Thanks @patio11 for the thoughtful feedback on our YC application, and @gabhubert for your directions on positioning the product — set the tone for our pitch!