Not an export in OS. Can op explain the project motivation and high-level design a bit? The readme file seems to be a feature introduction and barely touches the motivations, i.e., what problem does it aim to solve, is it designed for any specific domain or it's a general purpose OS, what gain could those feature bring in practice?
I happen to live in China right now and am using a VPN. I find your comments to be fundamentally incorrect. I live here more than 20 years and can't find any information about genocide, and I have a lot of (I suppose you are talking about) Xinjiang friends. Yes, there are extremists in Xinjiang with foreign funding. And The the government prepares extreme religious individuals for modern society. This process involves learning Chinese and learning basic skills to have a job. If people really died there, why we Chinese people don't know it? I mean I can literally drive to Xinjiang to check these information.
If you don't believe my words and still think people are oppressed in Xinjiang or whatever place, please apply a Chinese visa and visit here. If your visit is short-term, then a visa is not even needed.
Your narratives about China reflect self-projection. Western historical treatment of minorities doesn't mirror our approach. It's like saying Chinese people are too oppressed to afford bread or are forced to pick cotton—ignoring that bread isn't a Chinese staple and cotton-picking doesn't carry any other cultural meanings here.
By the way, if I said capitalism makes Americans too poor to eat rice, wouldn't you find that ridiculous? I'm sure you would, because you guys generally don't eat rice. This projection is exactly what your views about China represent.
"Yes, there are extremists in Xinjiang with foreign funding."
Radical islmaists are definitely a thing, but don't you think, the chinese government considers anyone extremist not ok with one party rule?
(I don't expect a honest answer, since you need a vpn to communicate)
And that you don't find information and that your chinese friends don't find know about controversal topic, then this is rather a obvious sign to me, that chinese censorship is working.
> Radical islmaists are definitely a thing, but don't you think, the chinese government considers anyone extremist not ok with one party rule?
This issue is essentially one of trust. If a person cannot be trusted by others, then even if they list facts, others won't believe them. Therefore, even if the Chinese government proves that every extremist is a real one, the West will still believe there are innocent people wrongfully accused among them. I think only mutual communication and firsthand witnessing can eliminate trust issues. According to this webpage[1], if you are a citizen of those listed countries, you can visit China for 30 days without needing a visa.
Chinese censorship operates with a 'turn a blind eye' approach rather than strict enforcement. The government allows citizens access to outside information while maintaining nominal restrictions. GitHub exemplifies this—officially censored but widely used by Chinese developers without consequence.
VPNs and encrypted messaging apps like Telegram are technically forbidden but practically tolerated. This selective enforcement extends to politically sensitive topics. While the Tiananmen Square events of 1989 remain officially censored, most Chinese citizens know about them through VPN access. Other historical events like the Cultural Revolution are not cencored and are actually taught in schools.
About me, I have studied in the U.S. and lived extensively in China, so I guess, from your standard, I'm one of the least brainwashed people in China? I know lots of cencored information but none of them relates to Xinjiang opperession. Therefore, as an educated person who lived in U.S. and China, I can confidently state that reported oppressions in Xinjiang are false narratives.
You also talked about one-party rule. Oh c'mon, a lot of people, including more than 50 minority groups, dislike the one-party rule. It's meaningless for the government to oppress one specific group.
Extremists face detainment in re-education camps for two specific reasons: First, their radical beliefs exceed public tolerance—unlike most Muslims, they engage in violence. Second, they combine religious extremism with separatist ambitions. These interventions target violent extremism and territorial separatism specifically, not related to the one-party rule. Furthermore, the number of extremists is very small. I don't think they constitute a large portion in Uyghurs. So it's not millions of extremists, I think it's at most thousands.
You may heard lots of such fake news from BBC, CNN, NYT newspaper, etc, and these media outlets function as propaganda tools. I can give you an evidence: BBC and CNN used to report a lot Chinese human rights violations. But during the Trump administration, they magically fliped to support China and praise China a lot. This dramatic shift proves that they are political tools, or at least part-time political tools when speaking of China, that simply convey the message what they're ordered to convey.
If Western critics truly wanted to challenge China and make Chinese people wake effectively, I can give they some suggesions. They can talk about
1. unaffordable housing prices in China,
2. persistent low incomes in China, and
3. environmental cleanliness. I mean most places, excluding big cities like SF or NYC, are cleaner than places in China.
These are real problems that resonate with Chinese people rather than fabricated oppression narratives.
If you want to check my words, feel free to visit China to see the reality for themselves.
It shouldn't be surprising that a group which commits a genocide and has the power to censor the media would censor the media about the genocide, so when you go looking at media it censors, you don't see anything about a genocide.
If you wanted to know about genocide in China you'd have to go to non-Chinese media. Germany and the USA do the same thing.
I am a first year MSc student in Computer Science, supervised by a young AP conducting research on SSL(self-supervised learning). However, I have a lot of doubts about the practical usefulness of this field. I think I'll do a PhD, but have no plans to enter academia. I want to be productive in industry.
An example of an exciting industrial application of ML is that OpenAI created ChatGpt.
But for SSL and RL(reinforcement learning), the papers I've seen are basically about programs to play Atari games and MineCraft. It's very far from being practical and changing society.
I googled the problem, and it looks like the only industrial applications for SSL and RL are autonomous driving and healthcare, but I'm not interested in those.
What other industrial applications are there for SSL/RL?
After reading this blog, I'm ashamed. I just graduated from college. In my high school, the education of physics was so boring and tiresome that I even hated it at one point. For this reason, I chose computer science rather than physics as my major in college. Later, I gradually became interested in physics, however, due to the lack of good enough study habits, atmosphere and courage (which is a self-deprecating way of saying cowardice and laziness), until now I have not taken a step forward. This is the decision I regret most in my life. I am going to the United States to study for a master's degree in CS. Maybe I can learn some physics during the freetime of the two-year program because the educational resources in the United States are more abundant(perhaps).
I feel the same way about pursuing CS instead of Physics. But at some point, a pragmatic solution had to be made. So don’t go hard on yourself for that, you probably would have felt the same way about CS too.
Yeah thanks. The difference of CS and physics, and other natural science, I think, is that it's more about engineering even though it's name is computer "science". However the pupolar deep learning may satisfy the needs of a natural science enthusiast. Since Hinton sayed the way neural network workss resembles the mechanism of human brain[^1]. It's quite exciting.
