VeryFrugal@sh.itjust.works to A Boring Dystopia@lemmy.worldEnglish · edit-22 months agoPeople got mad after they lost their AI boyfriend after GPT-4o deprecationwww.reddit.comexternal-linkmessage-square122linkfedilinkarrow-up1501arrow-down19file-text
arrow-up1492arrow-down1external-linkPeople got mad after they lost their AI boyfriend after GPT-4o deprecationwww.reddit.comVeryFrugal@sh.itjust.works to A Boring Dystopia@lemmy.worldEnglish · edit-22 months agomessage-square122linkfedilinkfile-text
minus-squarebrucethemoose@lemmy.worldlinkfedilinkarrow-up11·edit-22 months agoThis 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.
minus-squarelichtmetzger@discuss.tchncs.delinkfedilinkarrow-up2·2 months ago Jamba Wasn’t that a ringtone service in the 90s?
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.
Wasn’t that a ringtone service in the 90s?