Loading activity…
@animaanandkumar
AI+Science, Co-Founder @accelerated_u, Bren Professor @caltech, Time100, Fmr Sr Director of #AI research @nvidia Fmr Principal Scientist @awscloud
AI
Its creators say it predicts HSE06 bands for nearly 6,000 chemically diverse three-dimensional materials using lower-cost PBE calculations.
AI
A researcher behind ReRoute says it fine-tunes pretrained models on existing data, without new intervention data.
AI
An attendee says discussions with Secretary-General António Guterres and the UN team explored new topics for scientific briefs.
AI
Anima Anandkumar says her Supercomputing keynote will explore what she calls a new scaling law beyond large language models, focused on AI that simulates and understands the physical world.
AI
The researchers behind PosteriorBench say better reconstruction accuracy can coincide with worse recovery of the distribution of possible solutions. Even strong generative samplers, they report, often underestimate uncertainty.
AI
A user shared that they were celebrating TIME’s TIME100 AI at the San Francisco museum.
AI
Anima Anandkumar says she collaborated with Prineha Narang on the project. A separate promotional post credits Narang’s research with encouraging Nvidia’s AI-and-quantum push.
AI
The team reports that CTO is, on average, 500 times faster than diffusion models while scoring 6.02 dB higher on PSNR, an image-quality metric.
AI
Anandkumar says her team starts with AI-discovered candidates rather than human-constructed ones, then turns numerical results into certified bounds for analytical stability arguments.
AI
Co-founders discuss physical world models days after company emerges from stealth.
