Vagan Abel

[Elon Musk @elonmusk]

Bravo @JeffBezos!

[Vagan Abel @vaganabelian]

Уверен, есть работа после которой ты не будешь чувствовать себя выжатой каждый день:)

Всякое бывает на работе, но если это постоянно, то зачем?

А насчет парня, то можно дать хотя бы роадмап того, что это не будет вечно, сейчас так, но скоро поменяется в силу X,Y,Z. Поверь, ему от этого будет гораздо легче.

Просто я был с другой стороны, и это тоже пздц

[Vagan Abel @vaganabelian]

@marclou @rauchg @andrewqu @vercel Everything is down.

Convex
Supabase
Vercel
...

[Vagan Abel @vaganabelian]

This is so cool. Will buy one day

[Massimo @Rainmaker1973]

MOTOROiD 2, Yamaha’s latest self-driving and self-balancing motorcycle.
https://t.co/uiw0RfEIUw

[Vagan Abel @vaganabelian]

@1776General_ Who even tell that ever?

[Vagan Abel @vaganabelian]

@aaron_lubeck Same shit with companies logo

[Vagan Abel @vaganabelian]

@benln https://t.co/yCXX9Li2nY

Personal AI that sounds and feels like you

[Vagan Abel @vaganabelian]

@YannuzHai local inference? It's not the problem

[Vagan Abel @vaganabelian]

This is exactly what I’m building all my time for the last 6 months - https://t.co/yCXX9Lhuyq

I hope public launch will be in a month

[Jon Hernandez @JonhernandezIA]

📁 Matthew McConaughey says he wants a private LLM, fed only with his books, notes, journals, and aspirations, so he can ask it questions and get answers based solely on that information, without any outside influence.

[Vagan Abel @vaganabelian]

@signulll Notebooklm have almost NOTHING with product he’s describing.

He’s speaking about he’s own LLM, he’s AI twin.

Not soft prompting on Gemini.

How you don’t understand that? How’s that possible

[Vagan Abel @vaganabelian]

@garrytan If you’re serious about your request

[Vagan Abel @vaganabelian]

@garrytan I’m building this for already 6 months. Working prototype and private beta launched.

Got rejected from YC batch. Revert the decision, Garry

[Vagan Abel @vaganabelian]

NotebookLM have absolutely nothing with the idea Matthew described.

He is not talking about smart prompts based on files you upload

[Steven Johnson @stevenbjohnson]

Uh, Matthew, I’d like to introduce you to a little app called @NotebookLM

[Vagan Abel @vaganabelian]

@hi_im_six @JonhernandezIA Check out https://t.co/yCXX9Lhuyq

[Vagan Abel @vaganabelian]

@sin4ch @JonhernandezIA It will work, Prefix-tuning, P-tuning, LoRA adaptation.

Take foundation model and apply LoRA adapter and you will get it

[Vagan Abel @vaganabelian]

@samar_abedrabbo @JonhernandezIA Prompt is not going change the way model thinks. It’s need refine slightly its weights. That’s where LoRA adaptation is best solution.

I’m working on project same Matthew described - https://t.co/yCXX9Lhuyq

[Vagan Abel @vaganabelian]

@kristoph @JonhernandezIA Wrong point. No matter how data you provide. You have al lot of options: P-tuning, Prefix-Tuning, LoRA Adaptation.

You’re not train model from scratch, you actually fine tune foundation model

[Jon Hernandez @JonhernandezIA]

📁 Matthew McConaughey says he wants a private LLM, fed only with his books, notes, journals, and aspirations, so he can ask it questions and get answers based solely on that information, without any outside influence.

[Vagan Abel @vaganabelian]

@nikitabier After every such post i’m losing batch of my followers 😂