Unverified50% confidenceFactExact time
The spherical Brownian diffusion paper proposes Brownian motion as the most natural isotropic diffusion process on the sphere, and uses it as the forward process of a continuous diffusion language model.
1
Sources
50%
Confidence
Long-term
Relevance
10/27/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedClassical diffusion models (such as DDPM and Score-Based Generative Model) rely on Brownian motion driven by Gaussian noise as the forward corruption process.69% similarUnverifiedDiffusion-LM is widely recognized as the pioneering work in the direction of diffusion language models.63% similarUnverifiedAutoregressive diffusion is a fusion variant built on diffusion models that introduces conditional dependence on historical time steps at each step of the reverse denoising process, causing the generation process to unfold along the time axis.63% similarUnverifiedBrown's thinking advances the AI alignment discussion from the question of 'whether a single model is aligned' to the level of 'whether the evolutionary process of alignment is stable,' requiring both the measurement of alignment levels and the monitoring of alignment trends.62% similarUnverifiedDiscrete flow matching is technically an extension of diffusion models into the discrete domain, migrating the denoising approach from the continuous pixel space to the discrete token space of text.60% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1102200API
curl https://kongchang.com/api/v1/knowledge/claims/1102200MCP
get_claim(id=1102200)