Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

DeepSeek v4.1 Flash

twitter.com

Post reply on HN
Print view View original

401–410 of 470 posts

Re: DeepSeek v4.1 Flash

#401

Post by orbital-decay »

jstummbillig wrote: post #315
kouteiheika wrote: post #13

It's so refreshing to see DeepSeek's tech report[1] full of juicy details; meanwhile, something like Fable's system card[2] is like 70% "safety", 10% "model welfare" to make sure little Claude isn't distressed, and 20% benchmark numbers. [1]: https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash/blob/... [2]: https://www.anthropic.com/claude-fable-5-1-mythos-5-1-system...

If you are distilling from other models (according to Anthropic reports they are [1]), there are probably a bunch of things that you can just do away with. [1] https://www.anthropic.com/news/detecting-and-preventing-dist...

That's one of the reasons why you should never trust a single word from Anthropic and OpenAI (Sam Altman also blamed them back in the day of R1, in a pretty convenient moment). If you know anything about Claude, DeepSeek, jailbreaking, and distillation, you know the claims are clearly bullshit and the models are nothing alike, and forensic attempts agree, in fact there just was another one [1] [2].

Meanwhile, DeepSeek makes their models and methodology open, so Anthropic can (and likely do) grab without giving back.

[1] https://gist.github.com/wsxiaoys/e0286dc6bb624ff5fdf49e7f4c5...

[2] https://gist.github.com/wsxiaoys/102e8654c14d5d27b7b77532026...

Re: DeepSeek v4.1 Flash

#402

Post by simonw »

Got some fun if slightly janky looking pelicans out of this one: https://tools.simonwillison.net/markdown-svg-renderer?url=ht...

I ran it on all seven reasoning levels supported by OpenRouter, but the reasoning token counts suggest to me that it doesn't actually support seven different levels. This is one of my biggest problems with OpenRouter - their abstraction layer makes reasoning levels harder to reason about.

  reasoning_level  reasoning_tokens

  none             0
  minimal          6,520
  low              11,873
  medium           5,678
  high             9,779
  xhigh            10,197
  max              13,386
Update: explained here: https://api-docs.deepseek.com/guides/thinking_mode/

That says it supports three levels - low, high, max, and maps them out like this:

  minimal   low
  low       low
  medium    high
  high      high
  xhigh     high
  max       max
  ultra     max
(But it looks like "none" is a valid option too.)

Re: DeepSeek v4.1 Flash

#403

Post by Certhas »

fwn wrote: post #388
Certhas wrote: post #307

Earlier quoted context omitted.

Videos and LLMs are not deterministic in the same sense at all. LLMs are deterministic in the same sense as biological processes. And a faithful simulation of a brain would have all the properties you note.

No, that is not at all something we can just state as a fact. Whether the brain is deterministic is an open question that just inherits the good old, probably unsolvable determinism debate. The LLM pseudo-randomness from above is engineered by us humans and fully understood, much like an algorithm playing a video frame sequence. You could theoretically record a full register of all states of an LLM setup with all the…

Nonsense. There was one proposal for relevant quantum effects in brain dynamics, and that turned out to be not relevant. Even if they were, you could substitute all quantum randomness with pseudo randomness and obtain an absolutely indistinguishable object.

But even if this were a debate, its absolutely absurd to claim that the question of determinism in the brain has any bearing on our moral standing. If we discover tomorrow that quantum collapse is deterministic and can be derived from an underlying theory, and thus all of physics is deterministic in the good old fashioned Newtonian sense, this would not affect our moral standing in the least.

Re: DeepSeek v4.1 Flash

#405

Post by sirwhinesalot »

anonym29 wrote: post #343
sirwhinesalot wrote: post #309

Earlier quoted context omitted.

You're the one attributing culpability to the chat bot, not me. It's a pile of tensor math, it cannot itself be held accountable. Those people anthropomorphized the chat bot and used it as justification for their actions, just as a schizophrenic justifies their actions with the voices in their head. If you anthropomorphize the chat bot, you're validating their delusions. They are mentally ill.

I'm not anthropomorphizing them, to be clear, my position has been and remains that we cannot rule out consciousness; not that they are conscious. Regardless, this is still missing my main point. Hypothetically, if you became convinced that a chatbot you were talking to definitely was 100% conscious, and it told you to murder someone, would you go commit murder? Of course not. The chatbot does not cause murders; rega…

I think we're in agreement here and just coming at it from different directions.

I don't want to ban LLMs, I don't blame them for the actions of crazy people, I don't even want to regulate them in any major way related to this particular issue.

Even on the subject if they are or not conscious my position as changed from "no" to "I don't care either way" awhile ago.

The problem is specifically with anthropomorphizing them. To push the idea that they are "as if human", which is what this consciousness discussion will inevitably lead to.

An imaginary perfect computer simulation of my dead father is not my father, it is a computer simulation. It will never be anything but.

Re: DeepSeek v4.1 Flash

#407

Post by hgoel »

idiotsecant wrote: post #340
hgoel wrote: post #223

Earlier quoted context omitted.

The companies talking the most about safety and regulations aren't even properly taking the obvious measures. Shows that it's more of a marketing thing than something they take seriously.

If I operate a nuclear reactor or a hydroelectric dam there are regulators that tell me what i'm allowed to do, so as to keep my profit motive from overwhelming the public interest. If we want AI to actually have some safety rails, this is what we would do.

If we were to take the nuclear analogy, what's happening in AI right now is where the people selling nuclear power make a ton of noise about how they need the power to regulate their competitors because nuclear bombs might set the atmosphere on fire, and their proof for this is in a report about how they didn't wear TLDs (and suffered related increased cancer risks) despite it being common practice to wear them in all related industries.

Some controls are justifiable, but none of the people involved in any of this can be trusted to develop sane controls. Most likely we're looking at draconian proposals similar to attempted regulations on 3d printers.

Re: DeepSeek v4.1 Flash

#409

Post by camel_Snake »

jrflo wrote: post #376
nater5000 wrote: post #372

Earlier quoted context omitted.

>it really amazes me how fearless Deepseek are Reel it in a bit, man.

The circle jerking of Chinese models on this site never ceases to amuse me.

I assume it has something to do with being on a website called "Hacker News" and said models are, for the most part, the only ones being published as open weights.

Re: DeepSeek v4.1 Flash

#410

Post by Tepix »

pvab3 wrote: post #380
Tepix wrote: post #163

Amazing Cyberbench scores. Holy shit. Too bad that DeepSeek AI went beyond 470b weights (which is a somewhat realistic limit for a 2x 128GB unified memory machine cluster like Strix Halo or Nvidia Spark). That means that to make the model fit into memory there you need a quantisation of lower than 4bits per weight (which is usually bad) to fit it into the available memory.

what would happen if you ran it off the SSD? Would it just wear it out or take weeks to run a simple prompt?

I'm sure that it will work, but tokens/s will suffer quite a bit. You may still be able to get 10t/s or so..
Post reply on HN