Jesse Bridgewater

What do I do if I need to get to room 735?

Good morning SF

ML / DS people: just a reminder that this is our moment and the next few years are going to be huge.

Don't settle for low-impact work.

Your work should have line-of-sight to revolutionize something.

I loved using Strang's Linear Algebra textbook as an undergrad.

Fucking legend.

Fidget work product

One of the biggest winners of the AI era are senior leaders who used to write code a lot.

Being rusty isn't much of a barrier anymore. And you can do simple things in 5 minutes between meetings.

Sparse GPT:

https://arxiv.org/abs/2301.00774

Interesting work on zeroing out LLM parameters with modest loss of model performance.

Frugal GPT:
https://arxiv.org/abs/2305.05176

Interesting performance optimizations.

I used to believe that representative democracy was needed due to the complexity of the work and the sheer volume of legislation.

But now gridlock means that so little is getting done that I think direct democracy might work just as well if not better.

And obviously this hasn’t been a problem here yet, but might be a good time to mention that pro-gun, anti-regulation comments here will also result in an immediate block.
NOBODY needs an assault weapon and high-capacity mag to “protect themselves.”

Get in we're skeeting our old cheese tweets.

The Full Story of Large Language Models and RLHF Large Language Models have been in the limelight since the release of ChatGPT, with new models being announced seemingly every week. This guide walks through the essential ideas of how these models came to be.

@vickiboykis.com not sure how your paper is coming, but this post is the best I’ve seen so far.

Leaked Google document: “We Have No Moat, And Neither Does OpenAI” SemiAnalysis published something of a bombshell leaked document this morning: Google “We Have No Moat, And Neither Does OpenAI”. The source of the document is vague: The text below is …

s/tw/sk

https://simonwillison.net/2023/May/4/no-moat/

Lots of good ideas regardless of if it is true or made up by. GPT-4

having been on twitter a little more recently it really is remarkable how the blue check is basically now a badge screaming “i am an irredeemable chud”

Nice hat

I see it differently. You create an adverse selection bias for people who could believe that narrative. If you can predict that kind of thing why are you working as a recruiter?

Understanding Large Language Models A Cross-Section of the Most Relevant Literature To Get Up to Speed

Overhauled the "Understanding LLMs -- A Cross-Section of the Most Relevant Literature To Get Up to Speed" reading list!

Now w. parameter-efficient methods, and insights via Pythia, among others to round it from 13 to 20!

https://magazine.sebastianraschka.com/p/understanding-large-language-models

Bluesky is cool like MJQ (the lounge in Atlanta in the 90s that turned Brad Pitt away).

One of the biggest potential benefits of AI is repealing Brandolini's Law. (cost to refute bullshit is 10x the cost to create it)

The cost of refuting bullshit can be eventually outsourced to AI and take much less human effort.

guy who got on this site a week before me: the barbarians have entered our sacred space. they have breached the threshold without care. this is a bad omen. we must be wary

me: i’m a small gnome making dandelion cordial in my little cave

LLM eyecandy!

(via https://arxiv.org/abs/2304.13712)

posting my first thirst trap. check out these thicc tails babe

the old hellsite is dying. the new hellsite struggles to be born. this is the time of posts.

You can make Nutella sandwiches with the back of a spoon. Take note Nobel committee.

The kids are "aight"

Bluesky makes everything better, even @dril.

Me: Pursue mastery, not status

Also me: I need to post some bangers on here so I can get more invite codes to lord over the plebes stuck on the birdsite