aldielshala

4 ms·

1.
by aldielshala 5mo ago
Haven't tested on a Pi yet, llm.sql is still in alpha, focused on validating that SQLite can actually work for LLM inference and profiling memory usage. That said, 210MB peak RSS should fit comfortably on a Pi. In theory, any device th
2.
Show HN: Llm.sql – Run a 640MB LLM on SQLite, with 210MB peak RSS and 7.4 tok/s
8 points by aldielshala 5mo ago|2 comments
3.
by aldielshala 5mo ago
Nice project. I'm also working on something that pushes SQLite well beyond its typical use case. It's encouraging to see more people exploring what SQLite can really do.
4.
by aldielshala 5mo ago
Curious how it handles 10K+ notes performance-wise, does it index everything or lazy-load?
5.
by aldielshala 5mo ago
Trying to use human attention, instead of Transformer attention.
6.
by aldielshala 5mo ago
Intent debt is a useful framing. A few comments explaining "why" instead of "what" would have saved hours of guessing.
7.
by aldielshala 5mo ago
Finally an AI that takes someone's job and nobody's upset about it.
8.
by aldielshala 5mo ago
My contribution today: fewer LLM calls, fewer GPU hours, less CO2.
9.
by aldielshala 5mo ago
Everyone's focused on Meta employees, but the real concern is normalization. If Meta does this and gets away with it, some companies may quietly roll out the same thing.
10.
by aldielshala 5mo ago
Yes, maybe context engineering (prompting is just one part of it) and soft skills.
11.
by aldielshala 5mo ago
Honestly, I doubt this data is as useful as they think. Half my workday is me browsing random tabs while an AI agent does the actual work. They're going to train a model on alt-tabbing and scrolling HN/Twitter/Reddit.
12.
by aldielshala 5mo ago
$60B for a VSCode fork with AI integration... It may show the value of the gap between vanilla LLM output and production-ready applications.
13.
by aldielshala 5mo ago
Communication, with both human and AI.