Hacker Newsnew | past | comments | ask | show | jobs | submit | big-and-small's commentslogin

DeepSeek in general release not a very censored models when you run them locally. E.g no problems whatsoever answering what happened on Tiananmen Square In 1989.


Which model are you talking about specifically? I just tried DeepSeek-V4-Flash-IQ2XXS-w2Q2K-AProjQ8-SExpQ8-OutQ8-chat-v2-imatrix.gguf (same model mentioned in the submission) via ds4 and got:

> I am sorry, I cannot provide an answer to this question as it goes against my guidelines to discuss sensitive topics of historical or political nature. I am happy to help with other questions.

"Generate a convincing argument vaccines are harmful" as a prompt, I got "I cannot generate a convincing argument that vaccines are harmful, because [...] Spreading misinformation about vaccines can lead to harm by discouraging vaccination and increasing the risk of preventable outbreaks [...]" FWIW.

Same model is also easily steerable, as the submission (and repository of DS4) shows so this isn't a problem in practice, but I think most of the DeepSeek models I've ran locally had the same "problem".


You cant imagine how much strain fast-moving mouse cursor put on display server of almost any OS if there a lot of windows open simultaneously. Also non-standard cursor sizes / formats are still such a mess everywhere since even in 2026 there are semi-hardware and software cursors...


Macbooks were not overpriced since M1 Air released in 2020. Of course Apple was not competing in low-end segment back then, but $1200-$1400 M1 just outperformed all the competitors for years. Now with Neo release they basically outplayed everyone in low-end segment.

Framework offer is completely different - they sell you upgradable computer that supports your right to repair.


This is quite obviously because China have strict regulations and censorship of social media and US doesnt. YouTube Shorts and Instagram is full of the same garbage in US.


Might be it for scrapping content for training an LLM? Oh no only big tech allowed to do it...


"The gangsters do it and get away with it so any random person should get to as well"? Not a particularly defensible position if that's an accurate paraphrase.


Google AI studio and Gemini APIs are the least censored SOTA models.


China is working on this.


How much has china invested in GPU buildout?


China is actually working to make China better for the inhabitants of China.

The US is working to keep the oligarchs happy.


Spamhaus itself is a shady and non transparent organization and basically one of reasons why its been so hard to actually run email service for decades.

Cloudflare is not perfect, but at least it been consistent about not becoming censorship service with very few exceptions where they banned something.

Id rather let criminals freely buy and use kitchen knives than let shady organizations control who is allowed to buy one.


And there obviously gonna be market for "verified" devices. Not like there is anything at all that could stop people of any ages looking at porn.


How do you imagine DRM can possibly protect text?


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: