nickpadge
Joined 5 years ago
Karma
13
Submissions
7
Member since
Nov 9
Recent submissions
Comment · 8 months ago
Comment · 1 year ago
I love lancedb. It’s the only way I’ve found to performantly and cheaply serve 50m+ records of 768 dimensions. Runs on s3 a bit too slow, but on EFS can still be a few hundred millis.
Comment · 1 year ago
Some of the sibling comments had questions around purposefully releasing defenses which don’t work. I think Carlini’s (one of the paper authors) post can add some important context: https://nicholas.carlini.com/writing/2024/why-i-attack.html.
TLDR: Once these defenses are broken, all previously protected work is perpetually unprotected, so they are flawed at a foundational level.
Ignoring these arguments and pretending they don’t exist is pretty unethical.
Comment · 1 year ago
How were you able to make the transition?
Comment · 1 year ago
I suspect this occurred to me multiple times over the last few years. I would seem to have ads on YouTube specifically related to vocal conversations I had with coworkers or family members, and far enough outside of my interests where the ads raised suspicion.
I was considering performing experiments to test if this was the case, but never got around to it. I wanted to be certain it wasn’t an instance of the Baader–Meinhof phenomenon.
More to the point, this is an incredible breach of privacy. The amounts noted in the article are pathetically small for such an intrusion, especially over the period of 10 years for impacted users.
Comment · 2 years ago
How is your experience with async SQLAlchemy with FastAPI?Ive experienced a handful of nasty bugs, specifically around how connection pooling is handled.
Comment · 3 years ago
Spawning | https://spawning.ai | Front-End/React Developer | REMOTE (US) | Full Time
Spawning is building AI tools for artists, by artists. Our mission is to empower artists to leverage AI while respecting ownership over their data. Our first product, Have I Been Trained (https://haveibeentrained.com), allows artists to flag their art for removal from the LAION-5B dataset that was used to train models like Stable Diffusion.
We’re looking for a front end engineer to help us build out our next series of products. Our front end stack is currently composed of React, TypeScript, and MUI. You’ll be working directly under the VP of Engineering to implement the front end code, and integrate it with our CI/CD pipeline and backends. Plenty of opportunity to learn backend, AI/ML, and cloud technologies, if interested. Open to new graduates.
Reach out to me directly at nick at spawning.ai for more details.
The docs quality is spotty, and the lack of parity between the async and sync python API is frustrating, but otherwise it’s been great.
The only performance issues I’ve had have been A) not rebuilding indexes on an appropriate cadence, B) not filtering the search space enough for queries which bypass the index, or C) running search against millions of vectors on an object storage and expecting millisecond latency.