akshay326

14 ms·

1. 1.▲
   TIL sun is energy efficient at quantum entanglement(opg.optica.org)

   1 points by akshay326 1mo ago|0 comments
2. 2.▲
   Show HN: Using RasPi and Qwen 2.5 VL to find out when birds visit my balcony(akshay326.com)

   4 points by akshay326 6mo ago|0 comments
3. 3.▲
   by akshay326 7mo ago

   Cloudflare cloned Next.js because the source was right there . Most closed products have 80% of their complexity locked behind auth, encrypted state, and prerequisite flows — you can't reverse-engineer what you can't observe
4. 4.▲
   by akshay326 7mo ago

   thanks for shipping gepa! read about it today morning, impressed by its prompt optimization results! PS - the github.io link is truncated
5. 5.▲
   More than one-third of cancer cases are preventable, a study finds(scientificamerican.com)

   3 points by akshay326 7mo ago|0 comments
6. 6.▲
   by akshay326 8mo ago

   > if you're worried the model is reinforcing your biases..... i agree, i don't understand many domains well enough, yet i feel there's value in calling out assumptions, irrespective of how hard verification is
7. 7.▲
   by akshay326 8mo ago

   > Sometimes I go a step further and ask the question with the opposite bias..... curious to try this have you ever found it biasing you in the opposite direction tho?
8. 8.▲
   by akshay326 8mo ago

   wow, i wonder how bulletined & concise the outputs of your prompt might be! have you ever felt this prompt being restrictive in some sense? or found a raw LLM call without this preamble better?
9. 9.▲
   by akshay326 8mo ago

   > Have the model play the Devil's advocate. i've tried this sometimes. only issue being dumb me skipping to add similar phrasing every time i open claude or gemini have you found a way to consistently auto-nudging the model by
10. 10.▲
    by akshay326 8mo ago

    Ah interesting. i like actor-critic models! do you use it just for coding or non-technical chats too?
11. 11.▲
    Ask HN: How to prevent Claude/GPT/Gemini from reinforcing your biases?

    30 points by akshay326 8mo ago|22 comments
12. 12.▲
    by akshay326 8mo ago

    > There is an angle for doubt, for sorrow, for hate, for joy, for contemplation, and for devotion. I’m so intrigued - what was going on inside Hansen's brain?
13. 13.▲
    by akshay326 8mo ago

    Dogs?
14. 14.▲
    by akshay326 8mo ago

    way outdated but i mumble a few things every now and then -- https://akshay326.com/
15. 15.▲
    by akshay326 8mo ago

    What would you cover not-continuous? Best methods I’ve observed -progressive loading (claude skills) & symbolic search (serena mcp)
16. 16.▲
    by akshay326 8mo ago

    trouble self-hosting? checkout our 1-click Railway setup: https://github.com/seer-engg/seer
17. 17.▲
    Show HN: Open-source welcome email automation: Supabase + Gmail + Seer(youtube.com)

    1 points by akshay326 8mo ago|1 comments
18. 18.▲
    by akshay326 8mo ago

    100% agreed - i think its hard to have both comfort and growth/discomfort at once. for the 'chill' part, you can sandbag and make some money at FMAANG or whatever the new acronym is
19. 19.▲
    Show HN: 3D Visualizing YC Advice(github.com)

    2 points by akshay326 8mo ago|0 comments
20. 20.▲
    by akshay326 8mo ago

    This is pretty cool, I wanted to do something with my Readwise Reader too. Jinx today claude code created a 3D Neo4J visualizer tool for YC advice + my quotes collection. Code here - https://github.com/akshay326/quote-v
21. 21.▲
    by akshay326 8mo ago

    I’ve used mini for synthetic dataset generation extensively. Never tried Mistral; will check it out
22. 22.▲
    by akshay326 8mo ago

    true both - i've observed i end up spending more tokens + time with linting, than without
23. 23.▲
    by akshay326 8mo ago

    which simple models have you found good?
24. 24.▲
    by akshay326 8mo ago

    thanks for the idea! https://x.com/akshay326_/status/2009856179854561476
25. 25.▲
    by akshay326 8mo ago

    LLMs try to cheat. all sorts of evasive ways or smart tricks in some cases to avoid working on context-heavy tasks. i've constantly observed if left unchecked it tries to loosen the lint settings
26. 26.▲
    by akshay326 8mo ago

    thanks i've not used PostToolUse but will checkout. i'm excited about Rust's autofixable issues promise. curious how effective they are, and how deep of a issue can they solve
27. 27.▲
    by akshay326 8mo ago

    currently starting to do the same over seer's frontend, i didn't realise how simple yet effective this technique / guardrail could be!
28. 28.▲
    by akshay326 8mo ago

    amen! that's my bitter lesson for the time being, unless claude gets eerily better
29. 29.▲
    by akshay326 8mo ago

    i agree, the tool is indeed broken. its simultaneously stupid and smart in different ways. but i think there's some value in continuing to use and evaluate it
30. 30.▲
    by akshay326 8mo ago

    Accurate TL;DR. Probably should've led with that instead of burying it 380 lines deep in an autopsy report :)