tehsauce

9 ms·

121.
by tehsauce 4y ago
This is so true! But we should keep trying :)
122.
by tehsauce 4y ago
It does exist! Those most notable are vast.ai and runpod.io I've been able to run ML experiments that would have cost 10s of thousands of dollars for a tiny fraction of that using vast.
123.
by tehsauce 4y ago
The closest well-known example of something like this is deepmind's "perceiver" architecture. https://www.deepmind.com/blog/building-architectures-that-ca...
124.
by tehsauce 4y ago
There has been pretty significant steady progress for almost the last 10 years, but only recently finally crossed into the realm of broad usefulness. This has driven an influx of people and funding to the field which increases the rate of p
125.
by tehsauce 4y ago
“There has also been a wide variety of accuracy-degrading performance optimizations like Xformers and Flash Attention, which are great tools if you are open to trading accuracy for performance” This is incorrect. Those optimizations do iden
126.
by tehsauce 4y ago
Super well done! The transition between levels is beautiful.
127.
by tehsauce 4y ago
I’d love an interactive guide that reflects real life experience with css, where every property mysteriously doesn’t work.
128.
by tehsauce 4y ago
This is great! It would be very interesting to see this approach applied to continuous CA
129.
by tehsauce 4y ago
A system only being able to perform tasks that you understand is a huge limitation. This will only continue as civilization transitions to silicon based life.
130.
by tehsauce 4y ago
Hi! Your guess is correct. I monitor prices on vast.ai and runpod to get the best possible GPU power per dollar. I updated the link you mentioned with some of this info!
131.
by tehsauce 4y ago
A Shameless plug, if anyone is interested in building apps using stable diffusion and wants to keep things as cheap as possible, I built a very user-friendly API that is 1/4 the cost of the official stable diffusion API. There is also
132.
by tehsauce 4y ago
Does anyone know anything about the size of this model? Number of parameters or flops? I couldn't find anything in the paper.
133.
by tehsauce 4y ago
I think the opposite is true. I think we're closer to the beginning of a kind of arms race.
134.
by tehsauce 4y ago
Hard to blame them for pricing when they can sell the exact same chip to a datacenter for $30k.
135.
by tehsauce 4y ago
Finally fixed this! I forgot that browsers can react to system light-mode/dark-mode settings.
136.
by tehsauce 4y ago
Strange, I’ll check it out. Thanks for the feedback!
137.
by tehsauce 4y ago
The background shows up white? What browser are you using? It's supposed to be black.
138.
by tehsauce 4y ago
> Image-to-Prompt Isn't this called captioning?
139.
How is computerender 10x cheaper than other services hosting Stable Diffusion?(computerender.com)
5 points by tehsauce 4y ago|6 comments
140.
by tehsauce 4y ago
Have you tried this? Do you know how well it works?
141.
by tehsauce 4y ago
I'm trying to compile a list of all crowd-sourced compute platforms that aren't exclusively for crypto purposes. So far I have: https://vast.ai https://rentaflop.com https://www.sheepit-renderfar
142.
by tehsauce 4y ago
I am currently building this! You can try it out right now at https://computerender.com
143.
by tehsauce 4y ago
There is https://vast.ai which is a generic GPU crowdsourcing platform. There are limitations that make it difficult to set up their machines as infrastructure however.
144.
by tehsauce 4y ago
Creator here. One of my favorite features of this API is that just the act of sending or posting a link to an image will cause it to be generated. For example: https://api.computerender.com/generate/origins-of-humanity,
145.
Show HN: The Simplest API for Stable Diffusion(computerender.com)
10 points by tehsauce 4y ago|1 comments
146.
by tehsauce 4y ago
It was about $200k in compute.
147.
by tehsauce 4y ago
Oh cool, that looks really nice! One of the fun things that our library does is that while it has built-in abstractions for SDFs, you can also just write pretty much whatever kind of shader (2D or anything) you want.
148.
by tehsauce 4y ago
Shamelessly mentioning our project shaderpark, which is also a live-coding interface for working with SDFs but using javascript! https://shaderpark.com/
149.
by tehsauce 4y ago
Agreed. The article claims the scenario is “mathematics vs circumstance”, but really it’s really naive math vs math which takes into account the nonlinear utility.
150.
by tehsauce 4y ago
Economies of scale and years of optimization, some of which likely have not been publicly shared.

More ›