this is pretty cool
Today we release FLUX 3 Action - an open-weight 7B world-action model, achieving SOTA performance and efficiency on various leaderboards such as RoboLab.
I am incredible proud of the team - and both backbone and embodiment-specific finetunes are open weight, alongside the training recipes!
1
1
285
For the nerds like me and @skalskip92: nice little boost from using my 3D-aware augmentation stack with relighting and fog. Seems to scale well with larger models, I should try some bigger ViT based things.
2
293
uɐɥdǝʇS retweeted
Pixel diffusion models spend as much compute on blank sky as on faces. Every patch is a token in every layer.
Our fine-tuned MiniT2I runs on region tokens instead. Half as many tokens, same quality, twice as fast, trained at different budgets for elastic inference.
13
37
398
49,986
Oi @ADoricko chill out over there
Los Angeles, CA (moments ago)
Local resident of 30 years has never seen the LA river reach this level following a series of thunderstorms that swept through the region this afternoon.
459
uɐɥdǝʇS retweeted
I have a soft spot for a certain kind of person. They are incredibly useful right up until an institution has to explain them.
20
15
280
18,716
huge
We made history. On just our second flight, we became the first commercial space company from Europe to deliver satellites into orbit. Mission 'Onward and Upward' successfully lifted off and deployed payloads into orbit. A new chapter for European spaceflight. More: isaraerospace.com/press/hist…
5
646