emadm

13 ms·

121.
by emadm 3y ago
What’s better a 130bn model or 10 13bn models working together.
122.
by emadm 3y ago
The ability to run it on your own private data and train/distill it from the larger models is very useful (see Orca for example: https://arxiv.org/abs/2306.02707 ). Most of the valuable data in the world is private
123.
by emadm 3y ago
Drop in replacements except for code tasks.
124.
by emadm 3y ago
Code is main thing, it has some tradeoffs. It tunes well on code though and the code ai team at stability ai are working on stuff. We can now set and forget runs so will have a better dataset for the next 13b and different tokeniser, this w
125.
by emadm 3y ago
Had a nice chat with Yann last week, we will release complementary stuffs. I don't think 30b and 65b are useful given what we do, the key is optimising models for consumer & swarming them. As for SD.. Maybe try the bot on the disco
126.
by emadm 3y ago
RedPajama ran on the Summit supercomputer ( https://en.wikipedia.org/wiki/Summit_(supercomputer) ) NVIDIA V100s/PowerPC chips as part of the INCITE grant which necessitated variations from the LLaMA training paramet
127.
by emadm 3y ago
We (Stability AI) trained it on our TPUs with input from the OpenLM team as an OpenLLaMA collaboration. The 20b model is 780b tokens in, lots of learnings so we can optimise future runs. Hopefully these will be useful bases for continued re
128.
by emadm 3y ago
Lucidrains just put redirect up while redoing it Our actual url is www.actuallyopenai.com
129.
by emadm 3y ago
This article was a bit silly, gave corrections not run. Note Robin, Andreas, Dominik, three of the five authors of latent diffusion and stable diffusion are part of Stability AI and we are the only active developers. Lots of models to come.
130.
by emadm 3y ago
Most of the original researchers of latent diffusion are employed at Stability AI as is the co-lead of Stable Diffusion. This leaked deck was before stable diffusion release and the number refers to other open source models built by stabili
131.
by emadm 3y ago
This is dumb. We employed Eleuther team members as Stability AI employees/contractors and incubated them until the 501(c)3 was set up and we managed to bring in other funders too: https://techcrunch.com/2023/03&#x2
132.
by emadm 3y ago
The strategy was not DreamStudio but variant open models for private data. We will be open sourcing DreamStudio shortly having refactored it.
133.
by emadm 3y ago
Stability AI is 18 months old so far and has spent 100x less than OpenAI which was founded in 2015. I think it unreasonable to compare at this stage, we are still early as a company and organisation. We have revenue (to be announced) but re
134.
by emadm 3y ago
Yeah, we were always really clear about that, the GitHub is clearly CompVis for example. We fund dozens of models and collaborate on them too, the ones we assert IP over are those in our GitHub as benchmark models based on open research cod
135.
by emadm 3y ago
Most of the latent and stable diffusion authors also work at Stability AI, as do many other generative AI research leaders in media. Naming rights on the model were no part of the compute grant, we give them incredibly freely and also suppo
136.
by emadm 3y ago
I normally don't like engaging in these things but think should step in given increasing fud. Stability AI lawyers never contacted HuggingFace requesting an IP takedown. You may ask HuggingFace themselves. RunwayML agreed not to releas
137.
by emadm 4y ago
Actually the license is not the same as the prior one to remove this
138.
by emadm 4y ago
He is happy now on an n=1 case. Potentiating GABA really helped with him, with other folk it may be balancing glutamate. Our medical system isn't really set up for that which is why I designed and launched CAIAC, it assumes ergodicity
139.
by emadm 4y ago
I am not a coauthor but do know a decent amount about AI systems from a maths and computer science degree from Oxford, couple decades coding and being lead architect on https://hai.stanford.edu/watch-caiac amongst other stu
140.
by emadm 4y ago
That's pretty much how we hire at Stability AI. Contribute to LAION, Eleuther or any of the image/media generation open source notebooks and you'll get an interview pretty quick, hired dozens that way.
141.
New sota OpenCLIP model released(laion.ai)
6 points by emadm 4y ago|1 comments
142.
by emadm 4y ago
Second this, one of the biggest contributors to open source AI and someone everyone should sponsor.
143.
by emadm 4y ago
Images actually used our open LAION 400m dataset plus 400m of their own. On compute there is more than enough compute available to open source now via LAION and Eleuther AI to train these models, will just a bit of time.