fractorial

3 ms·

1. 1.▲
   by fractorial 5d ago

   I am in the process of open sourcing it! Unfortunately, my employer requires a (hopefully) cursory legal review.
2. 2.▲
   by fractorial 6d ago

   I’ve rolled my own harness in Go. I have a rule to not let the production LoC exceed 50k lines. It is _very_ nice for my use cases. DeepSeek Flash V4.1 often performs at the level of GPT 5.6 Sol for non-orchestration tasks (programming and
3. 3.▲
   by fractorial 16d ago

   As a programmer, maths researcher, and painter, there’s certainly something satisfying (to me) when an artistically symmetrical piece observes no mathematical symmetry: The very asymmetry in the language used to describe it is, in and of it
4. 4.▲
   by fractorial 16d ago

   Only under gross negligence would it be unprovable: Did you use a model whose training set included user data? Did the transcripts of any of the agents include a tool call whose result including user data?
5. 5.▲
   by fractorial 17d ago

   I’m impressed you had the patience to even make it that far!
6. 6.▲
   by fractorial 17d ago

   Brilliant. Precisely the reason I stopped using Anthropic's products.
7. 7.▲
   by fractorial 22d ago

   Sir, I think you are lost.
8. 8.▲
   by fractorial 23d ago

   > Because Codex without codex is fantastic. Seriously: I roll my own harness over OpenRouter and GPT 5.6, and it is genuinely better than CC+Fable for my maths and programming use cases. The one thing Fable 5.1 still tops out on is the a
9. 9.▲
   by fractorial 29d ago

   `pathological` is being used here in the security context: For example, _a parser contains a vulnerability when pathological inputs cause a program to crash._
10. 10.▲
    by fractorial 1mo ago

    How does a _multi-trillion dollar company_ think this is good presentation?
11. 11.▲
    by fractorial 1mo ago

    This is ludicrous; also, why publicly admit that blatant violation of ToS?
12. 12.▲
    by fractorial 1mo ago

    Roll your own harness or use an open source harness with a Codex subscription. I maintain a Claude subscription for Fable but seldom use it.
13. 13.▲
    by fractorial 1mo ago

    Damn, what an epic library name.
14. 14.▲
    by fractorial 1mo ago

    Palomar requires the Comparator/formalization.yaml challenge stack. It also has an LLM-based vacuity gate backed by Tao. Theoremdb is just a slop farm.
15. 15.▲
    by fractorial 1mo ago

    To be clear, I am deep into auto-research, but hooking up slop to slop is just unlikely to produce anything valuable. Value is in how maths is communicated: The process, frustrations, triumphs, etc. We have to able to take generated formali
16. 16.▲
    by fractorial 1mo ago

    Irrespective of opinions about a project or its authors, disclosing a potential security issue in a public forum is markedly not responsible disclosure.
17. 17.▲
    by fractorial 1mo ago

    This is, plainly stated, awesome.
18. 18.▲
    by fractorial 1mo ago

    I would love for someone to give me a coherent argument as to how this isn’t tone-deaf, vacuous garbage.
19. 19.▲
    by fractorial 2mo ago

    I can’t say much for a Chief Evangelist, but the Product Manager certainly has quite the C.V.
20. 20.▲
    by fractorial 2mo ago

    Actual link to source material: https://zenodo.org/records/21871667 It appears that there is no machine-checked proof or corroboration of this claim; I would be dubious in calling this “solved.”
21. 21.▲
    by fractorial 2mo ago

    Fascinating approach; however, a nightmare to scroll on mobile.
22. 22.▲
    Go: A more robust proposal process(github.com)

    1 points by fractorial 2mo ago|0 comments
23. 23.▲
    by fractorial 2mo ago

    For context, I've rolled my own research harness in Go; I've been trying very hard to keep the non-test LOC around ~30k lines. To me, I consider this large, but necessary, especially after meticulously crafting red diffs while add
24. 24.▲
    by fractorial 2mo ago

    I’m not quite sure why the harness itself needs to be small. Isn’t the system prompt and management of system prompt the bit you want lightweight?
25. 25.▲
    by fractorial 2mo ago

    I similarly find it interesting; I do not understand why it is not the default to just generate the thing as a draft, research anything you aren’t clear on, and re-write it in your own voice.
26. 26.▲
    by fractorial 2mo ago

    Cosmically apt username given the substance of this comment.
27. 27.▲
    by fractorial 2mo ago

    It would appear that rolling my own Anthropic-free harness / serving stack with a closed-weight carve out for Codex models is an absolute win.
28. 28.▲
    by fractorial 2mo ago

    > Do anti-LLM types expect to be vindicated in an orgy of copyright lawsuits that resets the industry back to 2022? What do people pushing expect to accomplish? You weren’t kidding, huh.
29. 29.▲
    by fractorial 2mo ago

    > Google is less full of crap about the open weights stuff because of Gemma 4, but they are full of crap for a zillion other things I can't exactly feel good about them. So everyone sucks. Google catching this stray made me laugh, h
30. 30.▲
    by fractorial 2mo ago

    Pretty much this. I designed my own research harness and infra for theorem proving and conjecturing; however, you still need to review formulations!