Localllama is pretty dead these days
The current dream is one device transcription + llm sponsor tagging and autoskipping
Japan + saurkraut - Germany = sushi
https://kinostr.com/ now has public chat per event. You won’t see it until an event was created through the website though (to publish a channel ID)
If anyone wants to try to publish something then I can test the chat area 🤣
Connect with nostr doesnt work on mobile, sign in with amber would be cool
Vibe coded a podcast app BC fountain makes you use Google play lol
✅ Download from URL
✅ Play with adjustable playback speed
🎯AI advertisement tagging + skipping
There's a magic moment when a bunch of dumb components snap together into something cool
No fee recurring buys is cool, but I just direct deposit my paycheck in and get hit with fees - feels bad, can you do something about that?
Since tomorrows payday it's probably going to 120k then crashing
No price is too high, no tax too burdensome to satisfy the statist's desire for mediocre, pothole-ridden asphalt. They would sacrifice their firstborn for a highway system that's perpetually "under construction" and recoil if you even ask if there's a better way to finance continuous strips of concrete than by compulsory payments bundled with innumerable other "benefits" instead of a-la-cart payments
Please shill technical books on machine learning.
I want things suitable for audiobooks, like pop-sci, not so much textbooks
If they aren't audiobooks it's OK I'll just kokoroTTS them to audiobooks
6969 SATs for first 10 shills
SMH Claude can't even make a podcast app even after I gave it Google tools and dozens of dollars
Steak and kefir for breakfast
Will sponsorblock be replaced by on-device AI tagging?
What's the trendy bitcoin-approved Easter candy?
Like bespoke cocoa/tallow raw-milk chocolate bunnies?
Just wormhole far away and point your telescope back at it, it's still there
Folks allegedly running Llama4 and getting 25+ tokens/sec on a 3090 and shitloads of RAM
Literally a $2.5k setup to run a top model and acceptable speeds
Yea L4 was kind of a flop but MoE looks like the future and is the most accessible architecture for self hosting right now
My boss's new technical interview is "build something cool in an hour, use anything you want"
Sorry can't drive I need to update my car
you can just vibecode things
very interested in non-feed approaches to navigating nostr
a couple tweaks and this very basic visualization could be quite powerful; right now it's just looking at quote reposts and replies of a small subset of recent events; imagine loading more interactions and going farther than tweets into the other stuff rabbit hole
https://blossom.f7z.io/bf40ff59b4f3c8bda983f57d813161bb21d8e7970c68652fdc1826a371acf4b0.mp4
Or like a time series thing to visualize the life of a note
