DivniyB to LocalLLaMA@poweruser.forumEnglish · 2 years agoDoes OpenAI ToS prohibit generating datasets for open source LLMs?imagemessage-square18linkfedilinkarrow-up11arrow-down10
arrow-up11arrow-down1imageDoes OpenAI ToS prohibit generating datasets for open source LLMs?DivniyB to LocalLLaMA@poweruser.forumEnglish · 2 years agomessage-square18linkfedilink
minus-squareMonkey_1505BlinkfedilinkEnglisharrow-up1·2 years agoYou’ll get better datasets IMO using GPT to filter real datasets for quality rather than purely synthetic (which in theory would compound LLM flaws). But it’s a dumb law. It’s hard to even tell whats in a models dataset for sure, let alone where it came from.
You’ll get better datasets IMO using GPT to filter real datasets for quality rather than purely synthetic (which in theory would compound LLM flaws).
But it’s a dumb law. It’s hard to even tell whats in a models dataset for sure, let alone where it came from.