Did you know that Taiwan calls itself “Republic of China”? It’s official stance is also “one China”, but under ROC rule.
Why don’t they drop that? Like… No chance unless poo gets dementia.
Because they had a civil war, and even though the nationalists lost to the communists, they still feel entitled to the whole country. To properly understand present circumstances, you need to understand how we ended up here.
Oh, well, guess we’ll all go back to using the big, proprietary, close weighted model and hope it tells us the right answer when we ask who won the US Presidential election of 2020.
Who give af? Honestly.
also the actual model itself (at least some versions of it) aren’t even that china-biased, deepseek-r1 run locally can give perfectly factually accurate answers that make the chinese government look bad.
factually accurate answers
Their English language models are fed from the English language corpus, which is basically everything written by the people of the Five Eyes countries, with all the truths, mistruths, and biases that that entails, with all the internalized Western capitalist Cold War I & Cold War II propaganda that that entails.
LLMs aren’t repositories of truth, and just because you believe a piece of Western propaganda doesn’t necessarily mean that it’s factual or accurate.
in case you are not aware, the online version is hosted in china and has to comply with their laws… if you don’t like censorship you can locally install one of the distilled models which are: open source
The local models(full and distilled) are also censored. The models censorship is just implemented superficially to immediately close any thinking tags and refuse when detecting censored material. If there already is any token after the <think> token the model will start answering away, which also happens on the official API because it puts a new line after the <think> token for some reason. That’s why on chat.deepseek.com censored topics are first answered and then redacted by some other safeguard a few seconds later. Whilst there are some great abliterated(=technique that tries to remove parts of llms that cause refusals) versions of the distills on huggingface that prevent all refusals after a few tries, they only tackle refusals, not political opinions such as Taiwan’s status as an independent country.
What are you on about? running deepseek-r1 locally in ollama answers “censored” topics just fine, it just answers stuff like a chinese diplomat questioned on live tv
