HackerTrans
トップ新着トレンドコメント過去質問紹介求人

daymos

no profile record

投稿

Model collapse – how LLMs become worse when trained on their own output

ibm.com
2 ポイント·投稿者 daymos·5 か月前·1 コメント

コメント

daymos
·5 か月前·議論
Why can't you wait for that? I picture that as everyone sounding the same
daymos
·5 か月前·議論
I predict the future of the web will be no-bot-allowed spaces a bit like discord is right now. Once you know what you are reading was created by a bot, it looses all its appeal, except when its purely informational
daymos
·5 か月前·議論
I overheard a conversation between a uni professor and a phd student the other day. Professor was complaining 99% his students use chatgpt to write essays in uni. He seemed genuinely distressed about the effect this was having on all of them.