jnnnthnn

6 ms·

1. 31.▲
   by jnnnthnn 2y ago

   I got some feedback that GPT-3.5 was letting people down so improved my caching strategy and defaulted everything to GPT-4o!
2. 32.▲
   by jnnnthnn 2y ago

   https://julius.ai is great
3. 33.▲
   by jnnnthnn 2y ago

   True! I had "things that store my data in the cloud" in mind.
4. 34.▲
   by jnnnthnn 2y ago

   > Instead of downloading files to your desktop and then uploading them to ChatGPT, you can now add various file types directly from your Google Drive or Microsoft OneDrive. This allows ChatGPT to understand your Google Sheets, Docs, Slid
5. 35.▲
   by jnnnthnn 2y ago

   Oh yeah! What I meant is having Ollama run on the user's machine. Might not work for the use case you're trying to build for though :)
6. 36.▲
   by jnnnthnn 2y ago

   Could you sidestep inference altogether? Just return the top N results by cosine similarity (or full text search) and let the user find what they need? https://ollama.com models also works really well on most modern hardware
7. 37.▲
   by jnnnthnn 2y ago

   You might not want them to have that information, but I think Google's history search now supports that for Chrome users: https://myactivity.google.com/myactivity
8. 38.▲
   by jnnnthnn 2y ago

   Thank you! And... got a link to share? :)
9. 39.▲
   by jnnnthnn 2y ago

   Yay! Thank you for trying it out, so glad it's been useful!
10. 40.▲
    by jnnnthnn 2y ago

    Thank you for trying it out!
11. 41.▲
    Show HN: I built a LLM-powered Ask HN: like Perplexity, but for HN comments(hackersearch.net)

    92 points by jnnnthnn 2y ago|39 comments
12. 42.▲
    by jnnnthnn 2y ago

    Thanks! I couldn't see it on Chrome due to React Devtools interference. Turns out the page was ending up in bfcache ( https://web.dev/articles/bfcache ), which in turn led to the button's loading state being pr
13. 43.▲
    by jnnnthnn 2y ago

    Blog post is now live: https://jnnnthnn.com/building-hacker-search.html
14. 44.▲
    Building Hacker Search – Learnings from building a semantic search engine for HN(jnnnthnn.com)

    2 points by jnnnthnn 2y ago|1 comments
15. 45.▲
    by jnnnthnn 2y ago

    Snap! I can't repro this. Which browser are you using?
16. 46.▲
    by jnnnthnn 2y ago

    There's a place for both, yes! I think Hacker Search should offer each ( https://news.ycombinator.com/item?id=40241332 ) and ideally intelligently figuring out what to do based on your query.
17. 47.▲
    by jnnnthnn 2y ago

    Thank you!
18. 48.▲
    by jnnnthnn 2y ago

    Thanks for trying it out! > What was the biggest thing you learned while implementing this? How much the quality of the data and resulting indexes matter. My impression based on this experience is that "RAG" might be a cohesive
19. 49.▲
    by jnnnthnn 2y ago

    I definitely think there are opportunities to provide more useful & personalized search than what Google offers for at least some queries. Quality aside, I think the primary challenge is in figuring out the right UX for delivering that
20. 50.▲
    by jnnnthnn 2y ago

    Thanks! Yup, totally feasible. I might add that!
21. 51.▲
    by jnnnthnn 2y ago

    Thank you! To learn your interests we'd at a minimum need to know what HN stories you tend to click or comment on, e.g. by a different reader view or using a browser extension. Presumably your comments and submissions could provide u
22. 52.▲
    by jnnnthnn 2y ago

    Only indexing stories for now, not yet comments. That too will get fixed!
23. 53.▲
    by jnnnthnn 2y ago

    Try any of the examples listed on the landing page. You can easily access the HN/Algolia search equivalent by clicking on "Try Hacker News search instead" to the right (on desktop). Keyword searches on the Algolia engine will
24. 54.▲
    by jnnnthnn 2y ago

    Thanks for the feedback! I made a cutoff at stories with <50 points when ingesting content, and it looks like that cut off what you were looking for. Definitely not great. I'm planning to fix that in short order, feel free to sign u
25. 55.▲
    by jnnnthnn 2y ago

    Totally! I've come to deeply dislike thumbs up/down UXes, so I'm collecting that by recording clicks on results!
26. 56.▲
    by jnnnthnn 2y ago

    Relatedly, the "Follow up queries" in the righthand floating insert are an attempt at finding a decent balance between recovering from failures & giving users enough control on the queries themselves :) See https://
27. 57.▲
    by jnnnthnn 2y ago

    Sweet! Thanks for sharing. A prior implementation had HyDE running on user searches, but I found the results to be hit-or-miss depending on the query type. I definitely want to re-explore that though; I think it should be possible to do so
28. 58.▲
    by jnnnthnn 2y ago

    Hi Simon! Big fan of your blog. I actually generate two summaries: one is part of the ingestion pipeline and used for indexing and embedding, and another is generated on-the-fly based on user queries (the goal there is to "reconcile&qu
29. 59.▲
    by jnnnthnn 2y ago

    In fairness, it's probably impossible to unambiguously determine what the intended/desired interpretation is (though intuitively it seems like k-means should be lower likelihood)!
30. 60.▲
    by jnnnthnn 2y ago

    Welp, I'm so glad I wrote "at least one person", and not "one person" in my note in the OP :D