Episode Details

Today's papers: - Unified Audio Intelligence Without Regressing on Text Intelligence: https://arxiv.org/abs/2607.05196v2 - LLM-as-a-Verifier: A General-Purpose Verification Framework: https://arxiv.org/abs/2607.05391v2 - DSpark: Confidence-Scheduled Speculative Decoding with Semi-Autoregressive Generation: https://arxiv.org/abs/2607.05147v1 - Wasserstein Residuals: Learning Gradient Flows from Population Dynamics: https://arxiv.org/abs/2607.04738v1 - TILDE: TILt-based Distributional Erasure for Concept Unlearning: https://arxiv.org/abs/2607.06432v1 This podcast is from Colin Davis (colin-davis.com) using Claude & Elevenlabs.