Ibra 🐺

Software Engineer @PiedPiper

[Amazon Web Services @awscloud]

More AI-generated code doesn't make your team faster. It might actually slow you down.

[Ibra 🐺 @_0xIbra]

@DFintelligence est-ce que pour toi la bulle ia et la bulle datacenters c'est la meme chose ?

[Ibra 🐺 @_0xIbra]

@TheAhmadOsman with each new release revealing more and more.
i wonder how the overlords will restrict access to open source models for the public, it seems like the only way to get rid of that threat.

wouldn't surprise me to see these laws in eu applied in the next 2 years.

[Sandro @pupposandro]

This is still insane...

[Google Gemma @googlegemma]

Meet DiffusionGemma!

An experimental open model that explores a fast approach to text generation, released under an Apache 2.0 license.

Moving beyond sequential, token-by-token processes to generate entire blocks of text simultaneously. Here’s what’s new with DiffusionGemma: 👇

[Ibra 🐺 @_0xIbra]

@googlegemma inception inspired ?

[Ibra 🐺 @_0xIbra]

@Polymarket really loose definition of "high risk"

[Ibra 🐺 @_0xIbra]

sick of this bs, but when it comes to using our data for improving the models, then the ethics go out the window

[Ibra 🐺 @_0xIbra]

@bcherny mm-hmm, mm-hmm, yeah

[Victor M @victormustar]

open source must win

[Ibra 🐺 @_0xIbra]

@elder_plinius thank u my good sir!

[Ibra 🐺 @_0xIbra]

@testingcatalog @claudeai please -_-

[Ibra 🐺 @_0xIbra]

@robj3d3 it answered correctly not because it understands the physical impossibility of it, but because they added the right words in the dataset.

it predicted the right words. no understanding of our reality and physics.

[Ibra 🐺 @_0xIbra]

@robj3d3 you aware that, that test itself is useless right ? its proof that most people don't understand how LLMs really work.

[Ibra 🐺 @_0xIbra]

@beaverd @claudeai @felixrieseberg its a dumbed down version

[Ibra 🐺 @_0xIbra]

@noahzweben what about the uncensored op one ?

[Ibra 🐺 @_0xIbra]

is there a world far far, far away in which anthropic open sources opus ? 😆

[Claude @claudeai]

Fable 5 is state-of-the-art on nearly all tested benchmarks, with exceptional performance in software engineering, knowledge work, scientific research, and vision.

The longer and more complex the task, the larger Fable 5’s lead over our other models.

[Ibra 🐺 @_0xIbra]

@bcherny does the 2x limits in cowork also apply for fable 5 ?

[Ibra 🐺 @_0xIbra]

@ClaudeDevs im not hearing anything new here, the models are getting smarter but still the same slow, linear progress.

[Ibra 🐺 @_0xIbra]

@support_huihui @_kityru thank u my good sir