iiinova — where your digital self keeps evolving

Georgia Channing

@cgeorgiaw.bsky.social

ai4science @ 🤗, PhD @ Oxford

Georgia Channing @cgeorgiaw.bsky.social · 06/07/2025

🧬 super psyched to announce a new collaboration between @hf.co and Ginkgo Datapoints to open up high-quality biological datasets for the machine learning community! Just dropped the GDPx and GDPa dataset series on the Hub (x1000 boost to AI for drug development) 🔗 huggingface.co/ginkgo-datap...

071 0

Georgia Channing @cgeorgiaw.bsky.social · 15/05/2025

Me trying to finish a NeurIPS submission 🫠

090 0

Georgia Channing @cgeorgiaw.bsky.social · 07/05/2025

They have also made a bunch of great demos for aerial flood and burn tracking—not to mention some impressive crop annotation. 🌽 demo: huggingface.co/spaces/ibm-nas…

000 0

Georgia Channing @cgeorgiaw.bsky.social · 07/05/2025

Pretty wild performance with 95% masked input:

100 0

Georgia Channing @cgeorgiaw.bsky.social · 07/05/2025

Seems like people must not know that NASA and IBM are releasing really epic (and open!) foundation models for weather and climate. This is their Prithvi WxC model doing zero-shot prediction for Hurricane Ida. Model Family on 🤗: huggingface.co/ibm-nasa-geosp…

100 0

Georgia Channing @cgeorgiaw.bsky.social · 06/05/2025

More generally though, as much as we talk about the impact that AI will have on science, we have explored so so so little of we have the power to do. So, if you’re looking for impact, you’ve found it

media.tenor.com

Waking Up For School Like GIF

ALT: Waking Up For School Like GIF

000 0

Georgia Channing @cgeorgiaw.bsky.social · 06/05/2025

🔋Check out LeMat-Bulk here: huggingface.co/datasets/LeMat… 🧬 You can also see the materials explorer here: huggingface.co/spaces/LeMater�� 🔬And a great summary of what you can do here: huggingface.co/blog/lematerial

110 0

Georgia Channing @cgeorgiaw.bsky.social · 06/05/2025

But in materials science, @entalpic.bsky.social and @hf.co are changing that. LeMaterial unifies and standardizes 3 of the largest materials science datasets. Take this as a sign to go design LEDs that never die or make biodegradable plastics or batteries that charge in second

110 0

Georgia Channing @cgeorgiaw.bsky.social · 06/05/2025

So one of the biggest problems in AI4science is that datasets are usually distributed, non-standardized, and living on some godforsaken platform that forces you to either download a 2 TB file locally or download a billion 10 MB files.

140 0