Normalizing flows are constructed by learning the velocity of a stochastic interpolant via a quadratic loss derived from its probability current, yielding an efficient ODE-based alternative to diffusion models.
Hutchinson , title =
6 Pith papers cite this work. Polarity classification is still indexing.
6
Pith papers citing it
representative citing papers
StAD distills divergence of PF-ODEs via the Langevin-Stein operator for faster, lower-variance likelihood estimation in generative models without Jacobian costs.
A covariance-aware extension of DDIM sampling for pixel-space diffusion models that uses Tweedie's formula and Fourier decomposition to model reverse-process covariance and improves sample quality at low NFE.
A forward-only Lanczos gradient approximation for Hermitian matrix function bilinear forms whose error scales with the same residual norm as the forward approximation and appears stable without reorthogonalization.
RePlaid achieves a 20x compute gap to autoregressive models, new SOTA PPL of 22.1 among continuous DLMs on OpenWebText, and competitive scaling laws by aligning architecture with modern discrete DLMs.
Approximates large matrix multiplication via truncated SVD and circulant decompositions with O(n^2 log n) complexity and ~1% relative error, including LLM operation demonstrations.
citing papers explorer
-
Building Normalizing Flows with Stochastic Interpolants
Normalizing flows are constructed by learning the velocity of a stochastic interpolant via a quadratic loss derived from its probability current, yielding an efficient ODE-based alternative to diffusion models.
-
StAD: Stein Amortized Divergence for Fast Likelihoods with Diffusion and Flow
StAD distills divergence of PF-ODEs via the Langevin-Stein operator for faster, lower-variance likelihood estimation in generative models without Jacobian costs.
-
Covariance-aware sampling for Diffusion Models
A covariance-aware extension of DDIM sampling for pixel-space diffusion models that uses Tweedie's formula and Fourier decomposition to model reverse-process covariance and improves sample quality at low NFE.
-
Fast and Stable Gradient Approximation for Bilinear Forms of Hermitian Matrix Functions
A forward-only Lanczos gradient approximation for Hermitian matrix function bilinear forms whose error scales with the same residual norm as the forward approximation and appears stable without reorthogonalization.
-
Continuous Diffusion Scales Competitively with Discrete Diffusion for Language
RePlaid achieves a 20x compute gap to autoregressive models, new SOTA PPL of 22.1 among continuous DLMs on OpenWebText, and competitive scaling laws by aligning architecture with modern discrete DLMs.
-
Efficient approximations of matrix multiplication using truncated decompositions
Approximates large matrix multiplication via truncated SVD and circulant decompositions with O(n^2 log n) complexity and ~1% relative error, including LLM operation demonstrations.