3 ms·
0 books. The huggingface hosted gpt-neo 125 model is capable of doing basic question/answer analysis to a sufficient level for most moderation. No finetuning ne
by robbedpeter 5y ago
0 books. The huggingface hosted gpt-neo 125 model is capable of doing basic question/answer analysis to a sufficient level for most moderation. No finetuning needed if you have it work out yes or no zero shot answers to things like "is this text bullying, excessively negative, political, sexist..." etc, and then iterate over cases and log results. Modify your prompts to handle edge cases and it would probably be capable of replacing 80% of human moderation.
- EamonnMR 5y agoHmm, I'll still need to scrounge up enough context to understand most of that, but thanks.
- robbedpeter 5y agohttps://jalammar.github.io/illustrated-gpt2/ https://jalammar.github.io/illustrated-gpt2/ There are several easily found resources like this, and previous hn threads contain papers, documentation, and guide links spanning the spectrum from beginner to PhD researcher. You can get to a high level of proficiency without any books, but you'll need to learn the jargon for searches. It's a niche enough subject that results don't get too muddied on almost any search engine - good luck!