So looking forward to get on stage again tomorrow to talk about my research! I will be at the semi-finals of the German Science Slam Championship in the Theaterhaus Stuttgart, 8pm. Excited about the amazing line up & the massive audience 🔥 Sandy Engelhardt Universitätsklinikum Heidelberg Informatics4life

Using the DINOv2 open source model from Meta FAIR, Virgo developed EndoDINO, a foundation model that delivers SOTA performance across a range of GI endoscopy tasks ➡️ go.fb.me/h0q0ke

Anyway I do find Black Mirror to be very tech-optimistic in that the plots usually assume that the technology works pretty seamlessly

Now would be a good time to read/watch ‘A scanner darkly’

gaussian splats making it to Superman before half of computer vision Twitter figured out what they were

easy

radiology ai is 10% training, 90% opening a DICOM on windows 98 with IE6… after Citrix and an SAP login.

great feature

funny, I don’t see the actual CV researcher categories here: ‘fix my loss function so it doesn’t collapse in 2 epochs’ (40%) ‘explain why my GPU dies at batch size = 1’ (35%) ‘convince me my architecture isn’t just ResNet in makeup’ (25%)

c’mon guys this can’t be real

neurips applying dropout at test time because the venue context window is too small

once indian uncles get this it’s over

pov you’re a squirrel watching sutton on dwarkesh: field notes on the persistent anthropocentric bias in reinforcement learning