> According to the latest regulations, we can only operate in this way. We need to strictly abide by the necessary process.

The thread references regulations put in place by Microsoft about LLMs.

Makes me want it even more.

I don't think that the statement implies the model is anything close to being uninhibited, just that only internal and well known public test suites and data were used to rate and soften its ability to discuss topics.

So even if the model is adjusted afterwards, likely the changes will only be slight to cover those few category of cases they didn't anticipate and probably won't make much difference.

good thing I downloaded it yesterday

I don’t understand how a group allegedly tied to Microsoft AI according to their paper is able to have their own twitter and do their own PR?