[md_rumpf]

every 3rd sentsnce is "the model was not trained on data from meta's products"

[keyle]

That makes sense. You probably don't want to train your LLM against your uncle's dubious claims about the government, flat earthers and the likes content :)

[esafak]

At least they're clear about provenance, unlike OpenAI.

https://www.reddit.com/r/ChatGPT/comments/1bfa7s3/openai_cto...