Daniel Z Kaplan retweeted
We @TeraflopAI have worked together with @johngfriedman and @daftengine to open-sourced all major filings from SEC EDGAR completely for free on @huggingface. It is now more important than ever to push for open dataset releases.
Given the increasingly closed-source nature of the U.S. AI ecosystem, it is now more important than ever to push for the proliferation of open model and dataset releases. Datamule (@johngfriedman), @TeraflopAI, and @daftengine collaborated to release 43 Billion Tokens of SEC EDGAR data.
Daniel Z Kaplan retweeted
Happy to release the Common Pile, an 8TB, 1 Trillion Token Dataset of Public Domain and Openly Licensed Text in collaboration with @AiEleuther, @VectorInst, @allen_ai, @huggingface, and DPI by @ShayneRedford. We provisioned a subset of the Common Pile, consisting only of public domain data for 99% of US precedential caselaw from the Caselaw Access Project and CourtListener.
Daniel Z Kaplan retweeted
Are you ever fed up with the lack of sufficiently fine-grained control in T2I diffusion models?
It turns out diffusion models already knew how to do it, we just have to learn how to leverage this capability.
Project page: compvis.github.io/attribute-… (1/n)
Daniel Z Kaplan retweeted
@TeraflopAI is excited to help support the @caselawaccess and @HarvardLIL, in the release of over 6.6 million state and federal court decisions published throughout U.S. history.
Daniel Z Kaplan retweeted
authors:
@RiversHaveWings* @StefanABaumann* myself*
@iScienceLuvr @DanielZKaplan @EnricoShippole
*equal contribution
it's been great collaborating with people I met through @AiEleuther.
with this, Birchlabs is finally a lab!
Daniel Z Kaplan retweeted
Happy to share a new paper I worked on!:
"Scalable High-Resolution Pixel-Space Image Synthesis with Hourglass Diffusion Transformers"
abs: arxiv.org/abs/2401.11605
website: crowsonkb.github.io/hourglas…
A quick thread about the paper! ↓ (1/11)
