A Holistic Approach to Undesired Content Detection in the Real World · HackerPapers discussions

A Holistic Approach to Undesired Content Detection in the Real World — mentioned by intended

Mon, 24 Jul 2023 14:59:51 GMT

Content Moderation / Trust and Safety person Open AI’s como paper, A Holistic Approach to Undesired Content Detection in the Real World. https://arxiv.org/pdf/2208.03274.pdf Lots of interesting facts are strewn around the paper. ——- The first paper that squarely talked about the language resource gap in CS/ML. Before this came out, it was hard to explain just how stark the gap between English and other languages was. Lost in Translation: Large Language Models in Non-English Content Analysis https://cdt.org/insights/lost-in-translation-large-language-... —— This paper gets in for the title: “I…

Paper and discussion history. A mention is not an endorsement.

A Holistic Approach to Undesired Content Detection in the Real World — mentioned by seanhunter

Fri, 23 Jun 2023 08:41:45 GMT

I'm not sure what exactly you mean by censorship here because nothing that I know of that they do is actually censorship. In particular they don't do the sort of prompt stuffing you're talking about, because that would just use up the context window of the LLM and you'd know because the context window is public and as a dev you can use the whole window. There is a "system" role that allows application developers to "whisper in the ear" of the LLM and provide prompts into the context that are different from the user prompts. That's how you build chat apps where you have a user providing some…

Paper and discussion history. A mention is not an endorsement.

A Holistic Approach to Undesired Content Detection in the Real World — mentioned by seanhunter

Fri, 16 Jun 2023 18:47:47 GMT

Don't believe me? OpenAI themselves use a special-purpose model and not an LLM for moderation. https://arxiv.org/abs/2208.03274 If they don't think this is a situation for an LLM, you can probably safely assume it's not.

Paper and discussion history. A mention is not an endorsement.