HDMI: Advanced Inference Time Causal Probing in LLMs
Discover HDMI, a probe-free method enhancing causal probing accuracy in large language models for reliable, context-aware AI text generation.
Parallel Lifted Planning with Semi-Naive Datalog Evaluation
Enhance AI planning efficiency using parallel lifted planning via semi-naive Datalog evaluation. Achieve faster, scalable automated planning solutions.
Open-Ended Task Discovery with Bayesian Optimization
Discover how Bayesian Optimization and the Generate-Select-Refine framework enhance task discovery and optimization in scientific workflows.
Vision-Language Models: Bridging Images and Text
Explore how Vision-Language Models combine computer vision and NLP to enhance AI's understanding of images and language seamlessly.
Multi-Environment POMDPs: Finite-Horizon Strategies & Algorithms
Explore new algorithms and PSPACE-completeness in Multi-Environment POMDPs with finite-horizon objectives for advanced AI decision-making.