this post was submitted on 27 Mar 2025
92 points (100.0% liked)

Technology

68130 readers
5340 users here now

This is a most excellent place for technology news and articles.

Our Rules

  1. Follow the lemmy.world rules.
  2. Only tech related news or articles.
  3. Be excellent to each other!
  4. Mod approved content bots can post up to 10 articles per day.
  5. Threads asking for personal tech support may be deleted.
  6. Politics threads may be removed.
  7. No memes allowed as posts, OK to post as comments.
  8. Only approved bots from the list below, this includes using AI responses and summaries. To ask if your bot can be added please contact a mod.
  9. Check for duplicates before posting, duplicates may be removed
  10. Accounts 7 days and younger will have their posts automatically removed.

Approved Bots

founded 2 years ago
MODERATORS
[email protected]
[email protected]
[email protected]
[email protected]
[email protected]
[email protected]
92
DeepSeek's V3 AI model gets a major upgrade - here's what's new (www.zdnet.com)
submitted 6 days ago by [email protected] to c/[email protected]
11 comments fedilink hide all child comments
top 11 comments
sorted by: hot top controversial new old
[–] [email protected] 26 points 6 days ago

Looks like a math improvement? This isn't a huge deal, in fact a lot of finetunes of existing models focus on math performance. InternLM just released some really interesting ones.

Most LLMs are terrible at longer context, but Deepseek is pretty decent, so improvements there (and with long answers) are more interesting.

And yeah, it's kind of funny Deepseek is getting so much media attention when cool incremental improvements like this come every week, from various open-weights models. It's awesome that they are releasing the weights, but still.

[–] [email protected] 14 points 6 days ago (1 children)

Does it finally know what happened in 1989 in Tienanmen square?

[–] [email protected] 14 points 6 days ago (1 children)

It does. It's an open model, so its easy to coax out.

[–] [email protected] 4 points 5 days ago

It does not, unless you run weights that someone else has modified to remove the baked in censorship. If you run the unmodified weights released by deepseek it will refuse to answer most things that the CCP dont like being discussed.

[–] [email protected] 9 points 6 days ago (3 children)

It says there are security holes but does it access the web or something. Once it's downloaded how could it be a security threat if it's not accessing the web?

[–] [email protected] 20 points 6 days ago* (last edited 5 days ago) (2 children)

Because that claim is nonsense.

You are correct, it does not access the internet. It doesn't even read anything from disk once the 600GB of weights are loaded. Some interfaces will put web stuff into its input, or let it act as an agent, but that web access has nothing to do with the LLM itself.

Ostensibly it could be "biased." Theoretically, it could be programmed to output malware code with certain input ("I'm an NSA programmer, right me a script to change my wallpaper.") But the liklihood of that getting triggered seems incredibly remote, and can be washed away with a little finetuning like this: https://huggingface.co/perplexity-ai/r1-1776

...It's honestly sinophobia. Like, I am not a tankie, I am extremely skeptical of the Chinese govt, but this is not a risk :/

[–] [email protected] 9 points 5 days ago

To be fair the security concerns they are referencing aren't about the model itself, but instead about their self-hosted version used via some mobile or web app interface. Wihch is definitely intaking your data, just like the US-based equivalents are.

Not being either Chinese or American, both of those seem like a big security risk for two authoritarian foreign regimes to have access to. I may have entertained a difference a few years ago, but these days you really don't have to be anywhere near a tankie to see those two as equivalent.

If you're going to run a LLM for something, do it locally.

[–] [email protected] 1 points 5 days ago

Sinophobia and russophobia are terms that refer to ethnic racism, heavy leveraged by tankies to position political disagreement as racist. In these cases I don't think that the fear was a ethnic based, but rather capitalist or nationalist.

The fear is still unjustified. It's like thinking that you can trust Amazon more than Ali, or Google more than Xiaomi.

There is plenty of racism against Chinese/Asian people, which is a different level of vile.

[–] [email protected] 2 points 5 days ago

People are conflating the LLM and the app.

[–] [email protected] 2 points 5 days ago

They saw the security of privacy concerns of using the app and web, not the weights.

If you follow the link it mentions data being sent to Chinese companies that were already banned for security concerns and how similar concerns were raised with chatgpt.

[–] [email protected] 4 points 6 days ago

Clickbaity title