That previous tweet about the internet of the late 90s got me reminiscing down memory lane about web development in the 90s / early 2000s... (see how many of these you remember)

Front end code used to render *drastically* differently depending on browser and there used to be more even distribution of browser usage. In my first job as a web dev in 2004, we would tear our hair out over IE and Firefox rendering quirks.
https://t.co/FQFLeOvGK5 became the definitive place to look up the answer to the question “why the fuck are you doing that, internet explorer?”
Digg launched in 2004 and had the first mainstream implementation of an Ajax button. It was an epic watershed moment and front end dev became exponentially more fun and innovative from there. Starting with things like Prototype.js leading to Jquery etc.
Early 2000s, mobile phones still had slow downloads and poor rendering capability. You didn’t build “mobile first” or “responsive”, usually you would just build a shitty, basic version of your main site and shove it on a mobile sub domain to be forgotten and ill-maintained 😅
Speaking of maintenance. I worked at a web dev shop in the late 90s before subversion or git were a thing. A coworker irreversibly overwriting your code with their own was actually a thing that could happen. The horror...
In summary: things are so much better for devs now 😅

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

THIS.

Russia hasn't been a willing partner in this treaty for almost 3 decades. We should have ended the pretense long ago.

Naturally, Rand Paul is telling anyone who will listen to him that Trump is making a HUGE MISTAKE here.


Rand is just like his dad, Ron. 100% isolationist.

They've never grasped that 100% isolationist is not 'America First' when you examine it. It really means 'America Alone'.

The consistent grousing of pursuing military alliances with allies - like Trump is doing now with Saudi Arabia.

So of course Rand has also spent the last 2 days loudly calling for Trump to kill the arms deal with Saudi Arabia and end our alliance with them.

What Obama was engineering with his foreign policy was de facto isolationism: pull all the troops out of the ME, abandon the region to Iranian control as a client state of Russia.

Obama wasn't building an alliance with Iran; he was facilitating abandoning the ME to Iran.

Obama wouldn't even leave behind a token security force, so of course what happened was the rise of ISIS. He also pumped billions of dollars into the Iranian coffers, which the Mullah's used to fund destabilizing activity [wars/terrorism] & criminal enterprises all over the globe
Funny, before the election I recall lefties muttering the caravan must have been a Trump setup because it made the open borders crowd look so bad. Why would the pro-migrant crowd engineer a crisis that played into Trump's hands? THIS is why. THESE are the "optics" they wanted.


This media manipulation effort was inspired by the success of the "kids in cages" freakout, a 100% Stalinist propaganda drive that required people to forget about Obama putting migrant children in cells. It worked, so now they want pics of Trump "gassing children on the border."

There's a heavy air of Pallywood around the whole thing as well. If the Palestinians can stage huge theatrical performances of victimhood with the willing cooperation of Western media, why shouldn't the migrant caravan organizers expect the same?

It's business as usual for Anarchy, Inc. - the worldwide shredding of national sovereignty to increase the power of transnational organizations and left-wing ideology. Many in the media are true believers. Others just cannot resist the narrative of "change" and "social justice."

The product sold by Anarchy, Inc. is victimhood. It always boils down to the same formula: once the existing order can be painted as oppressors and children as their victims, chaos wins and order loses. Look at the lefties shrieking in unison about "Trump gassing children" today.