Get live statistics and analysis of Yacine Mahdid's profile on X / Twitter

an ai product???? no no no no product why would release a product? if you show a product people will ask about benchmarks and it will never be enough frontier labs that were the 100X becomes the 2X saas dog but if you have no product you can say you building superintelligence a potential pure play

for those struggling with deep learning math I resurface this tutorial I made last year in 27min I show my process by working through QHAdam optimizer which is this alphabet soup of letters

I got you guys I got you

my last brain cell at 1:00 AM trying to resolve the CUDA errors

you think agi will end up being some overengineered mess like this and we will have to specialize in like synthetic ribosome engineering or cyber epithelial processing?

man, scientists working on optimizing matrix multiplications have oppenheimer level of aura - use a RL agent to spit out heckload of bilinear products - slap two MILP to combine and filter those - iterate on top of a Large Neighborhood Search flow until it’s fast fast what the

me supporting the wife in whatever interpersonal issue there is with 0 context because I love that woman

how to read research paper should be formally thought this stuff ain’t hard

in light of claude hitting 1M context here is a 27min video showing how bonker insane it is in it I review most strategies from interpolation to chunking to linearization to hybridization to making gpu go brrbrr to just pray for Nikolai Savinov to make it happen please god

dwarkesh is experiencing first hand the PhD candidate experience of arguing with your PI on the theoretical implication of your thesis

I’m about to DM primeagean for an interview, if he runs a background check with any of you this is an alt account thanks

ok chat what do I do

you will never believe what optimizers these bad boys are using

see? same pattern : - super duper expert know what the heck he’s doing - cut the project into micro functional sub tasks - feed to the llm and iterate on the context - brush up the output and ship it

I’m always surprised when a brand I respect is down to do a sponsorship because the technical tutorials I make are kind of niche sometime haha

someone asked for a tutorial on how to read deep learning paper code base and here it is it’s 45min long because I do a complete deep dive on the Segment Anything Model (SAM) result, architecture and codebase

ok chat what do I do

you think agi will end up being some overengineered mess like this and we will have to specialize in like synthetic ribosome engineering or cyber epithelial processing?

an ai product???? no no no no product why would release a product? if you show a product people will ask about benchmarks and it will never be enough frontier labs that were the 100X becomes the 2X saas dog but if you have no product you can say you building superintelligence a potential pure play

a thousand followers already, hell yeah man you folks rock

me supporting the wife in whatever interpersonal issue there is with 0 context because I love that woman

I’m calling it ma man is going to become a frontier LLM pessimist the amount of legit heavy weight machine learning folks that are deflating the hype to his face is just too much he’s gonna go full neuromorphic-latent-reasoning tin foil hat in like 2 podcast

guys btw just want to say that adam was developed in university of toronto so basically all of the performance of current frontiers llms are thanks to canada. I want every one of you guys say “thank you canada 🇨🇦” in the comments and order a can of maple syrup on amazon thanks.

okay real important question here which will brown do you prefer: 1. morgan stanley will brown 2. prime-intellect will brown very very crucial for my research I need answer asap thank you all

I’m very impressed by the quality of followers I have not only are they intellectually obviously top percentile of the human population these folks are also very good looking it’s incredible

here’s my growth strategy: - step 1: find an interesting niche deep learning topic that consume my whole life for a week - step 2: get sidetracked by random side quests by my followers - step 3: ??? - step 4: find back the initial topic and post about it also identity theft

I always knew there was something special about me

if you give the first frame of this video to an LLM there is literally 0% chance they can predict what will happen next

dear sweet jesus no wonders deep learning took time to take off

what are you 5 years goal folks

weighting on this subject with tact because I hate to see two of my education heros butting heads: no math for deep learning. I’m not saying learn it later or use your high school math no. math. at. all. math in machine learning is a psyops from Big Statistics to slow us down

I might get crucified for this but in-person handwritten code exams is the best way to assess computer-science skills