A familiar face…

I finished Karpathy's NN series, decided to skip books and math, and jumped right into training models. It's really fun, and theory feels like a waste of time until I build up a decent domain intuition.

Let's make it count

They changed my flight to a Boeing

I want to share my words of appreciation for Jane. It's been almost a week since she passed away, but she'll forever remain a part of my passion for nature and anthropology. RIP

The last few years were necessary chaos. What comes next is the new beginning. I adore this feeling.

i am getting academically overloaded, i am yearning to build another simulator or compiler

Wait, I can just act on intuition when solving problems and get relatively high accuracy? I take back what I said about being a slow thinker.

There are some really amazing places to study in Warsaw

i just love programming, man

my takeaway: coding agents are incredibly powerful, but many of their solutions are not optimal, and these little things add up. the difference between a good programmer and a great programmer is that a good programmer can solve any problem, whereas a great programmer can shape

truly remarkable hardware. i can't wait to write more kernels, but first i need a decent rig that can handle it

i wrote an autograd engine in c++ backed by tenstorrent kernels, and a tiny, broken, MLP. a huge shout-out to @karpathy for teaching me ML in two days https://t.co/GyWp6hDTt7

top 3 @ tenstorrent hackathon please hire me

one billion in loss 20 minutes before submissions

tt n300s instance is running

locked in

i have just enough time to binge watch Karpathy's neural nets series

for the next few days, i'll be preparing for the tenstorrent hackathon. i'd trade a kidney for an internship, and two for a job

it's my first time working with a coding agent. let's see how it goes