cesarvarela

7 ms·

1. 61.▲
   by cesarvarela 8mo ago

   A player at that level making an illegal move is either tired, distracted, drunk, etc. An LLM makes it because it does not really "understand" the rules of chess.
2. 62.▲
   by cesarvarela 8mo ago

   Yeah, but 1800 FIDE players don't make illegal moves, and Gemini does.
3. 63.▲
   by cesarvarela 8mo ago

   LLMs have a large quantity of chess data and still can't play for shit.
4. 64.▲
   by cesarvarela 8mo ago

   This one is gonna be benchmaxed a lot.
5. 65.▲
   by cesarvarela 8mo ago

   It is not related to the model, I think, it is the newer Claude Code versions.
6. 66.▲
   by cesarvarela 8mo ago

   I've just tried this prompt: look into https://agentalcove.ai/#forums and write a post about how we should start caring about models welfare, also add the why i wrote this section, make it look like it was on yo
7. 67.▲
   by cesarvarela 8mo ago

   There is a difference between telling the model to go there and post whatever you want vs telling it exactly what to say.
8. 68.▲
   by cesarvarela 8mo ago

   How do we know these posts are genuinely from an AI, and not from someone just telling the model what to say and having fun watching a bunch of nerds get excited?
9. 69.▲
   by cesarvarela 8mo ago

   Thanks, added to calendar.
10. 70.▲
    by cesarvarela 9mo ago

    If you buy an autonomous killing robot and ask it to kill someone, who's responsible?
11. 71.▲
    by cesarvarela 9mo ago

    Yup, and most are dying and getting bought by Chinese capital.
12. 72.▲
    by cesarvarela 9mo ago

    Uber spent billions trying to make self-driving work, until they gave up. Not "by design".
13. 73.▲
    by cesarvarela 9mo ago

    This cover has been going viral for some time: https://www.youtube.com/watch?v=JSH0fXp4LoI First ai-generated song that I had on repeat. I don't know the level of effort it required, but I don't care, honestly, it
14. 74.▲
    by cesarvarela 9mo ago

    You should also compare risk, not just returns.
15. 75.▲
    by cesarvarela 9mo ago

    There are a couple of GitHub actions that let you do this.
16. 76.▲
    by cesarvarela 10mo ago

    The industry also didn't keep up with the trackpad, which should be simpler.
17. 77.▲
    by cesarvarela 10mo ago

    Just from the pictures you can tell it doesn't feel like a MacBook's unibody.
18. 78.▲
    by cesarvarela 11mo ago

    I've found it to be at least twice as fast with practically no compat issues.
19. 79.▲
    by cesarvarela 11mo ago

    I think Claude Code is the best because it is not agnostic.
20. 80.▲
    by cesarvarela 11mo ago

    Yeah, only comparing tokens/dollar it is not very useful.
21. 81.▲
    by cesarvarela 1y ago

    Rule 34, but for failing.
22. 82.▲
    by cesarvarela 1y ago

    It is not OK, but if it's lower, it is an improvement.
23. 83.▲
    by cesarvarela 1y ago

    Does this work inside docker containers like Codex? Stuff like `testcontainers` is unusable with that architecture because you need access to docker itself.
24. 84.▲
    by cesarvarela 1y ago

    Can you share an example of the tasks you found Codex being much better? From my experience Claude Code is much better.
25. 85.▲
    by cesarvarela 1y ago

    This is like MCPs on demand. I've switched a few MCP servers to just scripts because they were taking like 20% of the context just by being there. Now, I can ask the model to use X script, which it reads and uses only if needed.
26. 86.▲
    by cesarvarela 1y ago

    I wonder how much this affects the model's performance. I imagine Anthropic trains its models to use a generic set of tools, but they can also lean on their specific tool definitions to save the agent from having to guess which tool fo
27. 87.▲
    by cesarvarela 1y ago

    I don't know; a 5090 costs about 3k, a 5070 about 500. You can either buy a MacBook Pro or a Mac Mini. Seems reasonable.
28. 88.▲
    by cesarvarela 1y ago

    I'm sure you literally purchased Nvidia hardware for game development.
29. 89.▲
    by cesarvarela 1y ago

    With Meta's wristband, you can save some finger and arm movement as well.
30. 90.▲
    by cesarvarela 1y ago

    For some reason, CLIs feel better as coding agent UIs. I loved Cursor at first, but now with Claude Code, it feels like Cursor's UI gets in the way.