@realfigmafiend @linhxeffect I'll starting poasting my experiments when I get home
English
Robin Hylands
547 posts
Robin Hylands
@Robin_Hylands
Studying math/cs. Building AI tech. Tapping the universal cosmic logos. ⁂ @momentum_place
London Ontario Katılım Aralık 2023
129 Takip Edilen73 Takipçiler
Launching Serendipity: the home for builders to think out loud.
if you’re a designer, a host, a builder, a niche hobbyist, etc…
here’s the superior place to post.
• doomscroll projects
• curate feed to only see stuff you like
• document progress w/o the pressure of the twitter algo
• give & get meaningful feedback
• find your cofounder
pls only join if you’re legit.
🔗 link in comments
English
Robin Hylands retweetledi
Today, @chris_oka_ and I are proud to share that we've raised an oversubscribed $1.56M Pre-Seed round, supported by @N49PVC, @StandUpVentures,@GarageCapital, @backboneangels, @skanwar, @Wehuns, and @tobi.
Community builders are at the forefront of the next big wave of entrepreneurship.
We're building the economic rails, professional network, and discovery layer they need to meet the skyrocketing demand for in-person experiences.
We're hiring across dev, design, and storytelling -- reach out if you feel there's a fit!!
English
Robin Hylands retweetledi
Robin Hylands retweetledi
Top 3 favourite questions to ask dudes:
> what sport would you play if you were athletic?
> do you ever wish your neck was shorter?
> did you know that you can buy shoe lift inserts?
English
English
@evanliin if you're worried about that just be curious about datacenter construction
English
The greatest compliment your dance partner can give you is to laugh with joy.
English
Was cleaning my hard drive and noticed I still had 10GB of food photos for training GANS
English
@RajabRehanDev Pack as little as possible if you're hiking >5km a day. I'd try to aim for 50lbs. The greatest luxury is light weight. Bring a wide brim hat and some mosquito netting to go over it. Breathable long sleeve button up shirts are ideal for hiking as they shield you from bugs and sun.
English
Either interp. is built into predictive modelling or each piece of data that updates the predictive model must also be able to update the AEs to the model's newly learned features.
English
A problem as I see it with mech interp. with SAEs is that the AEs are trained on frozen models; while true AGI will consist of continuously changing models. So mech interp. needs to be able to extract features of an ever changing model.
English
should I take it as a compliment or an insult if someone I meet thinks I’m a SWE?
English
@brianluidog @quantian1 You can't lose every bet if you bet for the most likely colour. Every wrong bet increases p(next bet) is correct. If you keep guessing wrong eventually you'll get 26 correct guesses in a row. I wrote a program to simulate 20m trials and the lowest number of correct bets was 26.
English
@quantian1 You can only guarantee that you make $1. You do this by betting all of it on the last card. Because earlier on, what if you lost every bet? You can't guarantee anything there.
English
A question posed at a quant dinner party: I have a 52 card deck I draw from sequentially. Each round, I offer you 1:1 odds on the color (red/black) of the next card, in any size you want. If your initial bankroll is $1, how much can you guarantee you can make playing this game?
English