ilaksh

37 ms·

661.
by ilaksh 2y ago
In a functional society we should not have to run most of our transactions through private companies giving them a piece of everything. The technology is obsolete.
662.
by ilaksh 2y ago
Wow, sounds like a nightmare. Operations staff definitely have real jobs.
663.
by ilaksh 2y ago
It's not going to be about careers anymore. It's going to be about leveraging AI and robotics as very cheap labor to provide goods and services.
664.
by ilaksh 2y ago
Don't worry, even if it was "available" in your country, it's not really available. I am in the US and I just see a waitlist sign up.
665.
by ilaksh 2y ago
Actually there is a pretty significant new model announced today and available now: "MiniMax (Hailuo)Video-01-Live" https://blog.fal.ai/introducing-minimax-hailuo-video-01-live... Although I tried that and it has
666.
by ilaksh 2y ago
That's an LLM workflow and not an agent if it's on rails created by a predefined workflow and doesn't make tool calls, or does not have any choice in tool calls. The tool calls are what give it agency.
667.
by ilaksh 2y ago
There is a very simple and obvious definition: it's agentic if it uses tool calls to accomplish a task. This is the only one that makes sense. People want to conflate it with their random vague conceptions of AGI or ASI or make some ki
668.
by ilaksh 2y ago
Yes you can and there is at least one example, a web application where you enter a URL and the LLM automatically generates the page including links, you click a link, the LLM fills it in on the fly. I can't remember the name of it. Bu
669.
by ilaksh 2y ago
This is really simple and approachable. Reminds me slightly of Forth or subleq. https://colorforth.github.io/index.html https://en.wikipedia.org/wiki/Forth_(programming_language) https://git
670.
by ilaksh 2y ago
I'm kind of with you in that you could build something kind of like it based on a fast LLM. But what they are actually talking about is a new cutting edge ML model that takes a huge amount of data and compute to train.
671.
by ilaksh 2y ago
I don't know off the top of my head, only played with it a little not seriously.
672.
by ilaksh 2y ago
I think that Cerebras and Groq would be fun to experiment with using normal LLMs for generating interfaces on the fly, since they are so fast.
673.
by ilaksh 2y ago
Larger models are more robust reasoners. Is there a limit? What if you make a 5 TB model trained on a lot of multimodal data where the language information was fully grounded in videos and images etc. Could more robust reasoning be that sim
674.
Sharing new research, models, and datasets from Meta FAIR(ai.meta.com)
325 points by ilaksh 2y ago|63 comments
675.
by ilaksh 2y ago
My own agent framework. Not quite ready to share but https://github.com/runvnc/xingen
676.
by ilaksh 2y ago
Since they explained that they used an open source model and system https://github.com/Microsoft/TRELLIS , it will be possible for other developers who want to start similar businesses to launch basic competitors within
677.
by ilaksh 2y ago
I think that's a fair point that not every company owes the end user a recipe for how to reproduce their product. However, it's also a fair question on Hacker News. Again, fair if they chose not to answer it.. but many people here
678.
by ilaksh 2y ago
People are flooding this with deliberately horrible ideas for some reason. I assume the reason is that people are trash. Not really what I expected from HN.
679.
by ilaksh 2y ago
Also can you ban people who post actual trash like binary files.
680.
by ilaksh 2y ago
Can you add markdown support? I already submitted one with markdown (username 'runvnc') .. also maybe deal with longer entries like only show the first 250 characters in the box and the rest in a tooltip/hover or when you cli
681.
by ilaksh 2y ago
It seems like the latent space could be even more useful if it was trained with the transcribed videos.
682.
by ilaksh 2y ago
This is actually a different version from what they had before. What they released today is Sora Turbo.
683.
by ilaksh 2y ago
Part of it might also be capacity problems.
684.
by ilaksh 2y ago
I can't even sign up. I assume it's a capacity issue.
685.
by ilaksh 2y ago
Text to image is a very efficient way to express yourself meaningfully.
686.
by ilaksh 2y ago
I will give you the possibility that some text to image generator models of the moment may not be able to create images that compete with Quanta magazine artists. Maybe. These systems have limitations but I think if you study your own logic
687.
by ilaksh 2y ago
"In all views". "It will do everything and I will need to say nothing". Those are the pretentious parts of your reply. It's just not true that everyone feels that practicing art is the main point. Sure, it can be de
688.
by ilaksh 2y ago
There have always been downturns in the velocity of increase in capability of technology like a series of s-curves, but at a zoomed out level it's still an exponential that continues as we constantly innovate. The trajectory we are on
689.
by ilaksh 2y ago
In this view the purpose of creating art is self-development? I suspect we could drill down into what you are trying to say for a whole thread but the result is just going to be a specification for pretension.
690.
by ilaksh 2y ago
The technical explanations are going to be technical. Based on what you said it sounds like you want to search through ComfyUI tutorials to find the one that seems just right to you. AI will eventually basically make artists obsolete just l

More ›