Thank you @SenSchumer for your leadership in framing SAFE AI innovation. It was an honor to be at @CSIS where @SenSchumer announced this. We should encourage public-private partnerships, especially in AI+Science, while minimizing risks.
@animaanandkumar
-
Earth-2: Largest Hybrid ML-Physics Dataset for Climate Modeling
By
–
This is the largest published dataset for hybrid ML-physics research! Great to be part of this work in advancing #AI + #Science for #climate The Earth-2 effort @NVIDIAAI is a massive undertaking to use the most powerful computing with #AI for climate modeling.
-
Schumer Launches SAFE Innovation Framework for AI Age
By
–
It was an honor to be at @CSIS where @SenSchumer revealed his SAFE Innovation framework emphasizing American innovation and the need to preserve it. Thank you, Tom Pritzker for highlighting the @caltech @UChicago workshop on #AI+#Science Transcript: https://
csis.org/analysis/sen-c
huck-schumer-launches-safe-innovation-ai-age-csis
… -
Diffusion Models Training 70% Faster with Competitive Performance
By
–
In fig, comparison with diffusion models on ImageNet-256×256 with and without guidance. The area of each bubble indicates the FLOPs for a single forward pass during training. Our method is more compute-efficient with competitive performance: reduces training time by ~70%
-
Asymmetric Transformer Architecture with Masked Patch Reconstruction
By
–
Ours is an asymmetric transformer encoder-decoder architecture: encoder on unmasked patches + decoder on full patches. To promote learning, we add an auxiliary task of reconstructing masked patches to denoising score-matching objective that learns the score of unmasked patches.
-
Masked Training Reduces Diffusion Model Training Costs by 70%
By
–
Reduce training cost of diffusion models by ~70% through masked training of transformer backbones. Masked training is popular for self-supervised representation learning, but we are first to show for #GenerativeAI https://
github.com/Anima-Lab/Mask
DiT
… @wn8_nie @Kay12400259 @ArashVahdat -
Reinforcement Learning Foundations for Autonomous Flight Control Systems
By
–
Congratulations @SahinLale He has laid theoretical foundations in #AI for autonomous systems by designing efficient reinforcement learning methods and applying them to challenging scenarios like flight control in turbulent conditions https://
thesis.library.caltech.edu/15219/ #Caltech2023 -
Physics-Informed Neural Networks for Molecular Modeling and Design
By
–
Congratulations @ZhuoranQ He has done trailblazing work on Physics-Informed Neural Approaches for Multiscale Molecular Modeling and Design #AI + #Chemistry with me and @tfmiller3 https://
thesis.library.caltech.edu/15077/ #Caltech2023 -
Physics-Informed Neural Networks for Molecular Modeling and Design
By
–
Congratulations @ZhuoranQ He has done trailblazing work on Physics-Informed Neural Approaches for Multiscale Molecular Modeling and Design #AI + #Chemistry with me and @tfmiller3 https://
thesis.library.caltech.edu/15077/ #Caltech2023 -
Caltech Students Graduate with Groundbreaking AI Chemistry and Autonomy Research
By
–
Congratulations @SahinLale @ZhuoranQ @Caltech #Caltech2023 Nothing brings me more joy as an advisor than to see my students graduate with flying colors! They both did ground-breaking interdisciplinary work in #AI + #Chemistry and #AI + #Autonomy