OpenDevin & AI Software Development, Enhancing Visual Language Models, , DDK: Refining Large Language Model Efficiency through Domain Knowledge

AI Papers Podcast

OpenDevin: An Open Platform for AI Software Developers as Generalist Agents VILA^2: VILA Augmented VILA HumanVid: Demystifying Training Data for Camera-controllable Human Image Animation PERSONA: A Reproducible Testbed for Pluralistic Alignment SV4D: Dynamic 3D Content Generation with Multi-Frame and Multi-View Consistency Scalify: scale propagation for efficient low-precision LLM training DDK: Distilling Domain Knowledge for Efficient Large Language Models