Episode Details

Today's papers: - SPADE: Self-Play in Adaptive Synthetic Executable Environments: https://arxiv.org/abs/2608.19197v1 - Evaluating and Explaining Prompt Sensitivity of LLMs Using Interactions: https://arxiv.org/abs/2608.18539v1 - HYDRA: A Heterogeneous Chiplet DSE Framework for Serving Dynamic Hybrid LLM Workloads: https://arxiv.org/abs/2608.19395v1 - PEA-DPO: Perception-Enhanced Alignment Direct Preference Optimization for MLLMs Alignment: https://arxiv.org/abs/2608.19598v1 - In Two Minds about Lifelong Learning: Exploring Hemispheric Redundancy and Specialisation in Neural Models: https://arxiv.org/abs/2608.19514v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.