fshcat
banner
fshcat.bsky.social
fshcat
@fshcat.bsky.social
June 12, 2025 at 4:25 PM
this album was strangely emotionally hard-hitting. it feels like a glimpse into a version of humanity we failed to grow into
June 10, 2025 at 1:38 AM
June 10, 2025 at 12:38 AM
birds pulling some looney tunes shit
June 7, 2025 at 2:44 AM
June 4, 2025 at 1:10 AM
"For the purpose of this model, a human life is valued at €2.2 million"
June 2, 2025 at 3:52 AM
May 27, 2025 at 5:05 AM
artisanal o4-mini-high search benchmark courtesy of friend who's really into seals
April 17, 2025 at 2:37 AM
this place is actually pretty good now
April 7, 2025 at 12:11 AM
very good
February 12, 2025 at 5:34 PM
who is using 6 bit floats
February 8, 2025 at 5:15 AM
artisanal benchmarks
Still hasn't cracked pelican on bicycle svg
January 31, 2025 at 9:09 PM
secondary education should front-load existential crisis
December 24, 2024 at 4:57 AM
Reposted by fshcat
It is so hard to acknowledge the vast set of things you don't know and decide each day to chip away at a little bit of it instead of being paralyzed by the magnitude and intractability of the task
December 9, 2024 at 8:04 PM
catified
how legible is this pfp
December 9, 2024 at 3:28 AM
i thought my discovery feed was alright but it seems like you need to exhaust posts from your follows before the actual discovery algo takes over. its not good
November 30, 2024 at 4:07 AM
how legible is this pfp
November 28, 2024 at 3:05 AM
staking moral patiency on a phenomenal quality like "if it can suffer" is passable (under a few reasonable assumptions) as long as only biological beings are under consideration, but when that stops being the case youre going to need something more empirical
November 23, 2024 at 10:17 PM
this is really good, this is basically the main reason i use algorithmic feeds
this is one of my favorite custom feeds, Quiet Posters

it shows you posts from people you follow who don't post that much! so it surfaces friends' posts that might otherwise get drowned out in the Following feed (which is reverse chronological) 🤗
November 23, 2024 at 12:17 AM
to what extent did the tpot cluster emerge from within twitter vs. get transplanted into twitter from lesswrong or somewhere else
November 20, 2024 at 6:24 AM
when i see posts about improving/elaborating on an idea by discussing it with an LLM (or posts presenting the outcome of that process) i continue to get the sense that the LLM is providing very little value. chat models (rlhf'd) are just too agreeable for that use case
November 17, 2024 at 3:21 AM
kinda shocked that the effects of sleep debt are this linear (PVT is a behavioral alertness test)
October 25, 2023 at 6:15 PM
this place rly needs bookmarks
October 24, 2023 at 12:59 AM