Chatting with the Chinese LLM

what's incredibly entertaining is that I'm sure they really tried to keep it out of the training data and yet there's clearly enough context elsewhere to impute what's going on here which is fascinating

I heard the joke was that Chinese AI can't count to 10 because you get to 8-9 and it thinks it's a Tianamen Square reference

really easy to laugh at this, but absolutely terrifyingly dystopian that this is the government of the country with the most people on earth

I'm surprised it didn't skip Tiananmen Square altogether

so it’s inference time, not training.

i guffawed at the end

Lol

Lol pulled an open source model (or model trained elsewhere 👀) and just slapped some keyword checks on the end… toddler UX