swah

9 ms·

91.
by swah 8mo ago
Do you keep wondering if "a simple python bot could do this" ?
92.
The Insane Stupidity of UBI(geohot.github.io)
3 points by swah 8mo ago|7 comments
93.
by swah 8mo ago
I mean't ... parlare in italiano to LLMs looks fun!
94.
by swah 8mo ago
Speaking to LLMs looks fresh!
95.
by swah 8mo ago
If I were you, I'd just start always in what you call "dashboard mode". Its underwhelming to start it up and not see the "boxes" :)
96.
by swah 8mo ago
That's absolutely correct, I fear.. In english those looks bad/funny/lazy... But in code, its probably ok. Its idiomatic code, I guess.
97.
by swah 8mo ago
Got a bunch of errors when starting Claude.
98.
by swah 8mo ago
Your name lives forever on the orange site..
99.
by swah 8mo ago
Can you expand on your usage this way? Still haven't picked on that.
100.
by swah 8mo ago
More like a remembering things person? I envy that!
101.
Two different tricks for fast LLM inference(seangoedecke.com)
194 points by swah 8mo ago|67 comments
102.
by swah 8mo ago
Add a gif to the Github page!
103.
by swah 8mo ago
To code on the go: https://github.com/slopus/happy
104.
by swah 8mo ago
Crazy. Is it a library or ... ?
105.
by swah 8mo ago
Same - they are too low quality for me to decode more than 8 hours afterwards.
106.
by swah 8mo ago
And only the confetti shredder is worth anything: https://www.youtube.com/watch?v=fVbFMyR-yWg
107.
by swah 8mo ago
Are you recommending the podcast or this particular episode? :)
108.
Quality and understandability after AI(federicopereiro.com)
1 points by swah 8mo ago|0 comments
109.
by swah 8mo ago
See https://marginlab.ai/trackers/claude-code/ and discussion here https://news.ycombinator.com/item?id=45177018
110.
by swah 8mo ago
That runs super smoothly!
111.
The fall of the nerds(noahpinion.blog)
8 points by swah 8mo ago|1 comments
112.
by swah 8mo ago
What are some thoughts for someone just getting into thinking?
113.
LLMs as the new high level language(federicopereiro.com)
177 points by swah 8mo ago|378 comments
114.
by swah 8mo ago
Nice but if a simple bot can do it, then I'd feel bad about deploying a big, expensive agent... I guess its easy to add more behavior with plain instructions? That would be a departure from a simple bot..
115.
by swah 8mo ago
Do you mean reset as reopening? How can I get those lists? I would like AI to tell me "this note looks like an incomprehensible brain dump and needs review before you forget today's meetings"
116.
by swah 8mo ago
Or maybe "they" did this on purpose?
117.
by swah 8mo ago
How we know its not humans trolling as agents?
118.
by swah 9mo ago
Happy to see Erik Naggum on this list - its the one I really remember the posts, mostly in a very "particular style" which was very entertaining to me (reading it a few years later). I kinda miss that style of poster and underst
119.
by swah 9mo ago
I'd hate because it would interrupt me at the worst times! See also https://www.lesswrong.com/posts/RWu8eZqbwgB9zaerh/third-time...
120.
by swah 9mo ago
Thanks for the plug - the compact list view did give me the fuzzy feelings in my belly ;) Moving from NetNewsWire to this for a while...

More ›