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

DeepSeek Introduces Vision

chat.deepseek.com

Post reply on HN
Print view View original

211–218 of 218 posts

Re: DeepSeek Introduces Vision

#211

Post by bonoboTP »

crypto420 wrote: post #174
tehjoker wrote: post #147

Earlier quoted context omitted.

Can you say more about that? I haven't kept up.

CNNs excel in vision tasks where you have limited compute, limited memory, limited data, and want something that works super well and quick. People usually don't hook CNNs up to a transformer to get language understanding either, you have to train bespoke CNNs for specific tasks ViTs excel where you're unbounded in compute + data and also want text understanding or have a conversation about an image

These are vibes. ViT has been shown to work fine on small data with proper hyperparam and most of what you mention is actually doable just fine with the other architecture as well.

Re: DeepSeek Introduces Vision

#212

Post by Royce-CMR »

jauntywundrkind wrote: post #188
carterschonwald wrote: post #146

Earlier quoted context omitted.

gemini models are also fantastic at understanding non spoken sounds

I don't know what runs on my phone's Google Translate app, but whatever it is, they are doing an insult to their models by it being so bad. It's amazing at picking up sound if spoken directly into the unit, but if trying to hold any kind of conversation or listen to anything even a little bit far away, it falls completely apart, is good for basically nothing. This is obviously different than the models most people ar…

I’ve long wondered if this was deliberate - only conversations where the participants are overtly using the translator get parsed.

Re: DeepSeek Introduces Vision

#213

Post by wolttam »

fc417fc802 wrote: post #148
wolttam wrote: post #110

Earlier quoted context omitted.

The parent is off, you’re right. They may reason in any language, typically whatever the user’s language is, and you’ll see the reasoning directly with an open model like Deepseek. Research only showed that thinking might be disconnected from the final output but in my experience they are very strongly correlated in recent models

> Research only showed that thinking might be disconnected from the final output It is trivial to regularly spot obvious contradictions and inconsistencies if you read carefully. For example I've encountered traces that amounted to "I can deduce X, therefore Y, so that means Z" but then the model turns around and outputs "the answer is W because X". It's even been demonstrated that having the model output placeholder…

I see those too and I think of it as the "thinking" in action. If you could replace their actual thinking trace with gibberish and get improved performance that scaled with the amount of gibberish you injected, that's what we'd do. But instead, we see that the quality of of the model's output scales with the amount of 'thinking' tokens they generate before responding.

It has been my experience that yes, models make contradictions throughout their thinking process, but the conclusions they arrive at during/near the end of thinking more often than not align with the final output.

Re: DeepSeek Introduces Vision

#214

Post by foota »

I find the way that models understand images to be seriously lacking. The root cause of the issue as I see it is that image encoding isn't contextual. The encoder should be aware of the prompt so that it can encode the right things. It seems like this should be something that could be trained into a model.

Re: DeepSeek Introduces Vision

#218

Post by paulluuk »

arcanemachiner wrote: post #55
paulluuk wrote: post #38

Earlier quoted context omitted.

Can you explain what the benefits are of actually "talking" with the bot instead of typing and reading? As someone who would rather send a slack message to a coworker rather than actually walking over and talk to them, the idea of having to talk with my laptop is not appealing at all, haha.

Much faster and better flow. Don't knock it til you've tried it.

I've tried it, and I find it quite terrible. I can read much faster than it can speak, it's very awkward when I take a moment mid-sentence to pause and collect my thoughts, and it already begins replying, and it's not at all reasonable to do if there is another human being nearby.
Post reply on HN