Fahad

[Fahad @fahadjibla]

@code_star Parallel people

[Fahad @fahadjibla]

@rohanpaul_ai Seems like the opposite of Marc Andreessen thesis

[Noam Brown @polynoamial]

When we announced @OpenAI o1 some researchers from other labs told me we made a strategic mistake and should have kept it secret so we could accelerate ourselves and pull farther ahead of the competition. Studies like these make me confident we made the right choice.

[Fahad @fahadjibla]

@antirez @ivanfioravanti I do. Europe has the capital they just need the will

[Fahad @fahadjibla]

@antirez @ivanfioravanti You should build a model rather than trying to run Chinese models.

I’m sure you will do a much better job than Mistral or Deepseek and you probably can easily get funding given your background.

Europe can rally around one of its tech legends.

[Fahad @fahadjibla]

@mattparlmer Or a ton of false positives

[Fahad @fahadjibla]

@rodydavis @antigravity Add Fable

[Fahad @fahadjibla]

@Yuchenj_UW They built models from the start like the diff model that applies what the llm spits out

[Fahad @fahadjibla]

@zxjroma @antirez I don’t deny they made valuable contributions to ml, but building a sota model takes more than just being smart.

[Fahad @fahadjibla]

@latentumx @antirez Any open source lab in the US will not do this because the legal threat is very real within the US

[Fahad @fahadjibla]

@antirez Thanks for the thread. Missed it

[Fahad @fahadjibla]

@antirez Really? Any references?

[Fahad @fahadjibla]

@zoink What makes you think Fable is 10T parameters

[Fahad @fahadjibla]

@thsottiaux @MehakdeepK81 I love codex but let’s be honest 5.5 is nothing like Fable.. I was a Codex user most of the time with Claude being used for some ui work, but after Fable I was almost exclusively using Claude.

Fable is a real threat to codex.

[Fahad @fahadjibla]

@basedjensen He turned meta in a data sweatshop lol

[Fahad @fahadjibla]

@hallerite Then how does deepmind operate

[Ido Aizenbud @IdoAizenbud]

What can a neuron compute?

Real biological neurons are complex, but how capable are they?

Using a new method, we found that a single cortical neuron can classify cats vs dogs, recognize spoken words, and solve 10-bit parity, all tasks thought to require entire networks. (1/15)

[Fahad @fahadjibla]

@DavidSHolz Don’t tempt them man

[Fahad @fahadjibla]

@timzaman How much did the wafer cost