joshycodes's profile picture. I engineer fun things. Prev: @uchicago

josh :)

@joshycodes

I engineer fun things. Prev: @uchicago

i am just addicted to data lol


These thumbnails seem so similar, yet have incredibly different results. I believe it's because of the contrast of the white and the black, but it just goes to show that making a video is only half of what's needed to successfully making videos lol

joshycodes's tweet image. These thumbnails seem so similar, yet have incredibly different results.

I believe it's because of the contrast of the white and the black, but it just goes to show that making a video is only half of what's needed to successfully making videos lol

pov: your neo pilot

joshycodes's tweet image. pov: your neo pilot

You can train household AI in a lab. MIT is currently working on generating millions of virtual houses to deploy these models into for training. I’m sure this will be WAY cheaper, efficient, and less invasive than just deploying teleoperators into people’s home for $20,000…

The teleoperator model IS the entire strategy. NEO uses Redwood AI, 1X's Generalist AI model, trained on real-world data. It arrives with basic autonomy and grows in capability over time. You can't train household AI in a lab.



“your car’s extended warranty has expired”

joshycodes's tweet image. “your car’s extended warranty has expired”

Only 1/5th what Nvidia is worth? Weak

JUST IN: OpenAI prepares for IPO at $1 trillion valuation, Reuters reports.



Loading...

Something went wrong.


Something went wrong.