This is downright terrifying…

  • brucethemoose@lemmy.world
    link
    fedilink
    arrow-up
    8
    ·
    edit-2
    19 days ago

    This is a common pattern unfortunately. Big LLMs are benchmaxxing coding and one shot answers, and multi turn conversation is taking a nosedive.

    https://arxiv.org/abs/2504.04717

    Restructure your prompts, or better yet try non-OpenAI LLMs. I’d suggest z.ai, Jamba, and Gemini Pro for multi turn. Maybe Qwen Code, though it’s pretty deep fried too.