ilaksh

29 ms·

571.
by ilaksh 2y ago
PaddleOCR layout works, and so do some open source large language vision models
572.
by ilaksh 2y ago
They mention ZetaOffice as a new wasm version.
573.
by ilaksh 2y ago
I would like to think that we will make it, as long as they don't create actual concentration camps by trying to deport millions of immigrants at the same time. Is anyone tracking any large detainment camps? But in general we need to m
574.
by ilaksh 2y ago
Hasn't started yet. Give it a few months or a year for the Taiwan situation to start to really boil and for them to have had their fill of rounding up Hispanics.
575.
by ilaksh 2y ago
I would like to believe that those are just intellectually lazy labels. Are you sure it's not 100% accurate to say that Musk and Hegseth are neo-Nazis? Or that we are headed straight for new American Holocaust against Hispanics and a r
576.
by ilaksh 2y ago
Maybe it's not that new then.. after all, I guess the Nazis did a lot of great work on rockets.. but very unfortunate. That association clouds people's views of technology in general. Technology itself is not biased or ignorant or
577.
by ilaksh 2y ago
My take is that we actually need more efficiency in government, and from a technology standpoint, they might have some good ideas. I have a very pro-technology worldview. I'm independent leaning left but I thought that Democrats have b
578.
by ilaksh 2y ago
I think that unfortunately what's been proven with recent developments is that in some way racism is very American.
579.
by ilaksh 2y ago
Why is it sad, and why are we restricted to textbooks still?
580.
by ilaksh 2y ago
That's an interesting idea. Just to mention though: LiveScript is a really great language that compiles to JavaScript. https://livescript.net/
581.
by ilaksh 2y ago
Right and congrats on all of that. But I need to know the evals to know whether this actually makes sense versus alternatives.
582.
by ilaksh 2y ago
Are there any evals or benchmarks of this model?
583.
by ilaksh 2y ago
Has anyone run any benchmarks on the quantized (non-distilled) versions?
584.
by ilaksh 2y ago
Is the deeper connection to art just about the fact that there was another person who felt something, or is it about what the art is communicating/representing? I would argue that the real value is in what feeling or information is act
585.
by ilaksh 2y ago
Okay but can you parse my question a little more carefully? Because I said "the whole point". Also you're talking about the context of the art and it's meaning. That does not depend on any actual person being involved or
586.
by ilaksh 2y ago
That's a great quote and explanation of why the process of making art is important for those who participate in it. However, it does not claim that the _entire point of art_ is that process, which is what I am refuting. I think it&#x27
587.
by ilaksh 2y ago
I would really like an objective explanation of why this attitude "that the whole point of art is the process of making it" is anything but pretentiousness in it's purest form. Surely the actual artifact itself has some value
588.
by ilaksh 2y ago
Yep, I got an email about o3-mini in the API an hour ago.
589.
by ilaksh 2y ago
True.
590.
by ilaksh 2y ago
I have mainly been using Claude 3.5/3.6 Sonnet via API in the last several months (or since 3.5 Sonnet came out). However, I was using o1 for a challenging task at one point, but last I tested it had issues with some extra backslashes
591.
by ilaksh 2y ago
It looks like a pretty significant increase on SWE-Bench. Although that makes me wonder if there was some formatting or gotcha that was holding the results back before. If this will work for your use case then it could be a huge discount ve
592.
by ilaksh 2y ago
I don't think OpenAI is training on your data. At least they say they don't, and I believe that. I wouldn't be surprised if the NSA or something has access to data if they request it or something though. But DeepSeek clearly
593.
by ilaksh 2y ago
Replicate has Hunyuan video training now. https://replicate.com/blog/fine-tune-video Also, Kling 1.6 Elements works pretty okay if you use the same person/face for each element. Kling also has lip sync. Or this li
594.
by ilaksh 2y ago
It says they were not able to reproduce an image with the same prompt. So they just didn't know about seeds?
595.
by ilaksh 2y ago
I think people didn't understand my comment. I am very aware of this already.
596.
by ilaksh 2y ago
I know what MoE is. Maybe read my comments more carefully and give me the benefit of the doubt.
597.
by ilaksh 2y ago
I know it's a MoE and I didn't need the five year old explanation of why larger models are smarter. I'm also aware of interpretability research. You should read my question much more carefully and think about it harder.
598.
by ilaksh 2y ago
I don't think chatbots are the point of generative UI at all. The point is to provide a dynamic UI mainly as an alternative to chatting. The challenge is doing that quickly and reliably. It's easy to get in a situation where the u
599.
by ilaksh 2y ago
Their art team should definitely start integrating AI into some pieces. It seems like it's actually perfect for The Onion. Imagen 3 has spectacular prompt adherence. Throw in a face swap and some Kling 1.6 videos sometimes. Seems like
600.
by ilaksh 2y ago
Right, they are. But they still use massive amounts of energy compared to brains. So it seems that we need a new paradigm of some sort. So much investment is being announced for data centers. I assumed there would be more investments in fun

More ›