Kratik

@kayleecodez Imo it's more important to impress yourself first

@moonboyknm not me. im telling y'all to do ts

stay hydrated. be wet. build some cool shit. woo your hbs/hgs.

@anirudhbv_ce @agihouse_org @Waymo @SebastianThrun @parsaxa @blelbach Common W for you atp 🐐

Today, I demo'd turboquant-gpu at @agihouse_org in Hillsborough. The founder of @Waymo asked me about the next steps and where I'll take it. I said I'll show him in two weeks and took his email. Stay tuned. Super super excited for what's coming. ✌️

@KyrieBlunders Oh man I feel you 😭 I've been learning raw cuda as of late

how i feel rn

Ok I have some catching up to do here

@KyrieBlunders Ooh ok gotcha!

@moonboyknm since I kinda worked on cutlass, so it was very intuitive when i moved to cutedsl

Finally finished writing this blog!! tried to connect the “what is this?” to “oh this is how it actually works” gap if CuTeDSL ever felt like abstract math cosplay, this might help would love to hear your thought regarding this!

@KyrieBlunders @PrimeIntellect The best 😂

the culture at @PrimeIntellect seems a little too good to be true lol majority of the employees are chronically online ngl

@KyrieBlunders What did you like about CuteDSL?

@moonboyknm both has its pros and cons

@pixperk Bro set reminders or smth

gotta build something that reminds me to read my frens' x articles. but I don't have no energy in me for that.

@JustPratiyush Wow 🤯

X is a weird place. My tweets only get 20–50 views on average, but I’m at least getting likes from legends 🥰

@4ster_light @AlexCaswen Real there's sm reasons for us to go poor then 😭

@AlexCaswen I just know I'm gonna be poor when I get a salary because I'll spend whatever is left to spare on physical books

Reading is Fundamental

@anirudhbv_ce @silicognition @ranaharshraj7 The goat has spoken

@silicognition @ranaharshraj7 I would suggest you go over PyTorch and numpy basics first. That way, JAX syntax looks familiar and easier to grasp. Gl!

Started learning JAX; Day 1 > understood the differences between JAX and Pytorch > how it uses tracers first to compile binaries using XLA > finally got around pure functions > when Jax is eager vs when its not I will try to go a bit more deeper to understand XLA sorcery. Then move towards re writing my decoder stack in JAX. Here are some of my notes.

@VishnuSatish_ Damn that's pretty cool

I built and trained a ~6M parameter GPT-2 entirely from scratch in C++, and it actually generates English text with mostly correct grammar! No PyTorch and no external dependencies. Just pure C++ 20. More info, GitHub link, and screenshots below!

Funny I'm using a statistical inference engine to study statistical inference

@offsitedark Nice reference

how offsite make money need money soon. question?

@eigentia Ig there's been many studies for that too but I fear the old correlation causation issue might make these a bit less useful

@moonboyknm Now that you say it, I am too. Yk there should be studies done on unsuccessful people as much as they are done on successful ones. Ah what a curious topic.

"owning a billion dollar tech company that runs without me while i'm on a mountain" is statistically the most common dream among people who have never built anything.

@anirudhbv_ce @Shopify @fnthawar Peakkkkk

I presented TurboQuant-GPU at @Shopify Toronto today :) Had a blast talking about KV cache compression and discussing future implementations for agents 🔥 Thank you @fnthawar for inviting me and hosting the talk today 🙏 pip install turboquant-gpu ✌️ ( 1.4k downloads!! )

@eigentia But now I'm wondering what could the real most common dream be 😭

@moonboyknm oh wait it was just humour

@almanfarooqui_1 Wow bro congrats ig

Thrilled to announce that I got my first bee sting today🥳