Ever wondered if the paths your generative model takes between examples actually make sense? The string method lets us trace meaningful, likely routes instead of crossing through unlikely regions, bringing new depth to how we explore these models.
diffusionmodels generativeAI latentgeometry
diffusionmodels generativeAI latentgeometry
Probing the Geometry of Diffusion Models with the String Method
arxiv.org
July 9, 2026 at 1:50 PM
Ever wondered if the paths your generative model takes between examples actually make sense? The string method lets us trace meaningful, likely routes instead of crossing through unlikely regions, bringing new depth to how we explore these models.
diffusionmodels generativeAI latentgeometry
diffusionmodels generativeAI latentgeometry
I have published Human-Mediated Resonance in Language Models (v2), a structural reinterpretation of early resonance and persona phenomena using #SPC v3. The paper formalizes these effects as conditional, entropy-shaped dynamics rather than emergent psychology.
#TopologicalDynamics #LatentGeometry
#TopologicalDynamics #LatentGeometry
Human-Mediated Resonance in Language Models (v2)
Abstract This paper presents a substantially revised and expanded reinterpretation of an early exploratory study on human-mediated interaction patterns in large language models, integrating subsequent...
doi.org
December 20, 2025 at 10:14 AM
I have published Human-Mediated Resonance in Language Models (v2), a structural reinterpretation of early resonance and persona phenomena using #SPC v3. The paper formalizes these effects as conditional, entropy-shaped dynamics rather than emergent psychology.
#TopologicalDynamics #LatentGeometry
#TopologicalDynamics #LatentGeometry