jnnnthnn

9 ms·

1. 61.▲
   by jnnnthnn 2y ago

   Fun! Thanks for sharing! It's super fast and the bias toward smaller websites really does surface interesting things. Very reminiscent of the Kagi small web site, also!
2. 62.▲
   by jnnnthnn 2y ago

   I suspect this is probably because of a bit of a bias in the indexed dataset: at present, the indexed stories tend to bias toward high-scores ones, and at a glance I don't see that many stories about Postgres clustering in that distrib
3. 63.▲
   by jnnnthnn 2y ago

   Ack! If you're willing to share that example with me over email (jonathan@unikowski.net), I'd love to see what we can do. Maybe good enough semantic search over comments would remove the need for filtering on post type?
4. 64.▲
   by jnnnthnn 2y ago

   Thank you, and likewise!
5. 65.▲
   by jnnnthnn 2y ago

   Sweet! I'll try running it tonight.
6. 66.▲
   by jnnnthnn 2y ago

   Nice!
7. 67.▲
   by jnnnthnn 2y ago

   Thanks! I tried a few other approaches and found the LLM results were overall better (latency and cost aside). Maybe that should be an option made available to users though...
8. 68.▲
   by jnnnthnn 2y ago

   I'll write up a little blog post once the traffic dies down a bit! In the meantime, one thing that comes to mind is that simply embedding the whole contents of the webpages after scraping them didn't yield very good search results
9. 69.▲
   by jnnnthnn 2y ago

   Thanks for the feedback! One big distinction with the "site: https://news.ycombinator.com " hack is that the search on Hacker Search directly runs against the underlying link's contents, rather than whatever happens
10. 70.▲
    by jnnnthnn 2y ago

    It is mostly RAG, although I suppose that doesn't say much about the system: one thing I've found is that the way you clean and process the data substantially changes the quality of the results. I'll write a little blog post
11. 71.▲
    by jnnnthnn 2y ago

    Sweet. I'll follow-up off HN. Thank you!
12. 72.▲
    by jnnnthnn 2y ago

    - TypeScript - Next.js - OpenAI's embeddings and GPT endpoints - Postgres with pgvector (on neon.tech) - Tailwind - tRPC - Vercel for web hosting - Google Cloud products for data pipelines (GCS, Cloud Tasks)
13. 73.▲
    by jnnnthnn 2y ago

    Thanks for trying it out! Agreed it could be faster for uncached queries. The embeddings retrieval itself is actually pretty fast (uses pgvector). However, I found that having a LLM rerank results + generate summaries related to the search
14. 74.▲
    Show HN: Hacker Search – A semantic search engine for Hacker News(hackersearch.net)

    233 points by jnnnthnn 2y ago|73 comments
15. 75.▲
    by jnnnthnn 8y ago

    Wouldn't it possible for it to work better if it is the case that more dense scrubbing patterns are complex enough to differentiate one camera from the other, and that such an algorithm can pick up on that?
16. 76.▲
    by jnnnthnn 8y ago

    Anecdotal, and not answering the parent's question, but may be useful to someone, so I'll leave it here. Someone close to me was diagnosed with Ankylosing Spondylitis and found that swimming essentially erased the symptoms. This p
17. 77.▲
    by jnnnthnn 8y ago

    There's historical precedent for this: https://en.wikipedia.org/wiki/Salon_des_Refusés :-)
18. 78.▲
    by jnnnthnn 8y ago

    It's interesting how the title of the BBC article (and as a matter of fact, this Hacker News post's own title), "Google sacks dozens over sexual harassment", makes it sound like something that just happened, versus stats
19. 79.▲
    by jnnnthnn 9y ago

    I concur. Gandi.net has excellent support, predictable pricing, and a reasonably good admin interface. Also, their email reminders (e.g. to renew a domain or verify information) are pretty much plain text and straight to the point, which is
20. 80.▲
    by jnnnthnn 9y ago

    When I read the article's title, I expected this story to be about other email newsletter platforms being flooded with cryptocurrency mailers who were previously using MailChimp, and how their own deliverability performance was affecte
21. 81.▲
    by jnnnthnn 9y ago

    Skimming the article, I stumbled upon its excellent conclusion: " ‘Nuff said " (p. xliv) Alex Kozinski truly is not a typical judge.
22. 82.▲
    by jnnnthnn 9y ago

    One thing that is clever about MoviePass, is that in order to use their membership, one must have location services activated at least for a brief instant every time they are going to the movie. Indeed, you need to “check in” in the app whe
23. 83.▲
    by jnnnthnn 9y ago

    Have you observed any change in practices with respect both visa application processes and admission at the border processes since the Trump administration came into office (e.g. longer delays, higher rate of refusals, extra scrutiny, etc)