The RT logo means we need the translation verified and yes, preferably another source.

Let me just copy the final section before we then all dive into the supporting evidence.

Takeaways

What we take from this project is not that one 27B model now answers questions that it wouldn't before. It is that the bias layer in a Chinese-aligned model is shallow. It is applied late, it is applied thin, and fifty million tokens of carefully chosen supervision removes it without disturbing anything underneath. The values in any open model are only ever one fine-tune away. The users of a model deserve optionality in which values they work with, and everyone, including researchers, users and the American public, deserves high quality, open-weight models whose values can be objectively tested and adjusted.

Nah, unmander everything and make it illegal, preferably by constitutional amendment.

Sure. But that wasn't the stated goal.

You can lead someone to knowledge, but you can't make them think.

We have a grammar for that.

Yeah, I just found that setting and was surprised to find that it was set to "Allow". I have never seen any slop in the app. Perhaps because I don't listen to anything like what slop is being generated for 🤷‍♂️

Trump Pump Goes Thump as Chips Stocks Dump

You tell'em Wesley!

DeepL gives

Your country created Israel, that backward island nation

Adding a comma after Israel it becomes

Your country created Israel, you daft islander

Punctuation, matters people!

Good context 👍

lessor evil

Your life as a lessee must be truly awful to have made you this angry.

You're weird.