I wish they published a post where we read about capabilities, quality, accuracy and other parameters
Maybe they have a big update soon as they made this one a silent update.
DeepSeek Introduces Vision
201–210 of 218 posts
Re: DeepSeek Introduces Vision
#202Direct competition to american companies like OpenAi, Anthropic proving china can also launch great models
Once we get Mythos level opensource then that would be in a league of its own.
Re: DeepSeek Introduces Vision
#203Earlier quoted context omitted.
I mean, even applied voice 'models' suck for this. For some godawful reason, Apple Maps voice directions assume that you also understand what it omits. So if it says "turn right in 500 meters" "250 meters" and then you stop at an intersection after 150 meters and it says "turn right", it expects you to understand that it doesn't mean the immediate right at the intersection, but the next one [because you still haven't…
This is one thing Waze I think seems to do better than the competition. And they have a ton of different voices. They also clearly show which voices can do street names (which is hugely helpful). For some reason the Australian and British accented voices feel more polite than the Americans
How about google maps says "keep north"..as if I am sitting in my car with a magnetic compass...gets my goat everytime
Re: DeepSeek Introduces Vision
#204Earlier quoted context omitted.
There's a dearth of research properly comparing them.
I'm talking about research pushing state of the art in computer vision. Vits have 100% become more popular than CNNs in most CV research.
Yes but not based on rigorous comparison. I'm not saying ViT is bad. But it took over mainly because it's the shiny new thing. It very bandwagon-Y even among PhD students.
Re: DeepSeek Introduces Vision
#205Earlier quoted context omitted.
Deepmind hasn't produced any frontier model since Gemini 3.0 pro though.
At IO, google said 3.5 pro would be released this month.
We'll see when it's released then! There's a chance it's going to be a very good model, but often DeepMind tend to "pre release" models that seem great, and then by the time you get to the release they've gotten worse for some reason.
They also tend to struggle with tool calls, more than the latest GPT or Opus. And around December 2025/January 2026 I remember the Gemini CLI being unusable because they were always at capacity.
But also I've seen product features built on Gemini models and they do pretty well here, especially around translation it seems.
Re: DeepSeek Introduces Vision
#206Earlier quoted context omitted.
I'm talking about research pushing state of the art in computer vision. Vits have 100% become more popular than CNNs in most CV research.
Yes but not based on rigorous comparison. I'm not saying ViT is bad. But it took over mainly because it's the shiny new thing. It very bandwagon-Y even among PhD students.
There's no 'rigorous comparison' that puts CNNs over Vits in quality and Vits unlocked more use cases easier than CNNs did. That's why they're more popular, not because it's 'bandwagon-y'.
Re: DeepSeek Introduces Vision
#207Earlier quoted context omitted.
I'm talking about research pushing state of the art in computer vision. Vits have 100% become more popular than CNNs in most CV research.
Yes but not based on rigorous comparison. I'm not saying ViT is bad. But it took over mainly because it's the shiny new thing. It very bandwagon-Y even among PhD students.
> There's no 'rigorous comparison' that puts CNNs over Vits
That’s not accurate. My team wrote a paper for school in which a resnet model out performed a ViT model of the same size on almost all metrics. These were smaller models, but depending on the use case that might be what you want.
Re: DeepSeek Introduces Vision
#208Earlier quoted context omitted.
Yes but not based on rigorous comparison. I'm not saying ViT is bad. But it took over mainly because it's the shiny new thing. It very bandwagon-Y even among PhD students.
> There's no 'rigorous comparison' that puts CNNs over Vits That’s not accurate. My team wrote a paper for school in which a resnet model out performed a ViT model of the same size on almost all metrics. These were smaller models, but depending on the use case that might be what you want.
Don't know if it's you (did you publish?). I read about something similar but it had its issies:
- Tuning hyperparameters to gain improvement on a dataset when you're constantly looking at the answers is pretty meaningless. It's basically testing on the training data.
- Eval on ImageNet1k alone (very small, useless for the real world) made me wonder if it wasn't just overfit to the training set. Would it perform better training on the datasets used for the foundation models ? I doubt it.
Well I'm not saying CNNs are bad or useless at any rate.
Re: DeepSeek Introduces Vision
#209Earlier quoted context omitted.
Yes but not based on rigorous comparison. I'm not saying ViT is bad. But it took over mainly because it's the shiny new thing. It very bandwagon-Y even among PhD students.
There's no 'rigorous comparison' that puts CNNs over Vits in quality and Vits unlocked more use cases easier than CNNs did. That's why they're more popular, not because it's 'bandwagon-y'.
What's the use case enabled vs running a ConvNeXt or EfficientNetV2 and using the resulting strided features as you would the resulting tokens of a ViT? I'm not saying that ViT is worse. Just saying that the scholarship around comparing them is very bad or nonexistent. You have to properly tune the hyperparam enters on both sides in a fair way, and use all the general modern training tricks also on the CNN side to make it fair.
Re: DeepSeek Introduces Vision
#210Earlier quoted context omitted.
> There's no 'rigorous comparison' that puts CNNs over Vits That’s not accurate. My team wrote a paper for school in which a resnet model out performed a ViT model of the same size on almost all metrics. These were smaller models, but depending on the use case that might be what you want.
Don't know if it's you (did you publish?). I read about something similar but it had its issies: - Tuning hyperparameters to gain improvement on a dataset when you're constantly looking at the answers is pretty meaningless. It's basically testing on the training data. - Eval on ImageNet1k alone (very small, useless for the real world) made me wonder if it wasn't just overfit to the training set. Would it perform bett…
Exactly. Most of the comparison papers are useless. This is hard stuff, only few people have the chops it takes to even attempt this. You can of course train some models and then post the numbers, that's not the hard part.