Need myself that Gobi Desert drip

I've created a neural network activator function that converges ~about 3x as fast as a standard LeakyReLU. After a full epoch of FashionMNIST its ~about 5% more accurate. Need to test it on a broader set of network types but looks promising, I'm excited. pic is loss vs

The MI-24 is EASILY my favorite helicopter

imagine driving this to the liquor store

“Uhhhhh… Let me be clear”

Middle management

I figured this out in my freshman year of college, if I didn’t write a note with a pencil, I didn’t learn it at all

I just caught myself doomscrolling LinkedIn. I think I may have a problem

-Be me -SSH into a remote linux ai system -install nvidia drivers -install vllm -create a server for a llm -run testing software I wrote I am completely Unemployable

Spent 2 days wrestling bugs while having and LLM help me code a transformer attention head algorithm idea. >got nowhere In frustration, I spent 30 minutes just carefully doing it by hand >worked

This PSN outage is complete NONSENSE! Sony! FIX YOUR stuff!!!

The fact that something that has ingested the entirety of human literature can't figure out how to generalize multiplication past 13 digits is actually a sign of the fact that it has no understanding of what a multiplication algorithm is.

Peter is indeed walkin, perhaps vibing

Drew a Peter griffin at work today, keeping the office interesting