Aaron Havens
@aaronjhavens.bsky.social
PhD student at UIUC looking at control theory and generative modeling. previously intern at FAIR NY and Preferred Networks Tokyo.
We specialize Adjoint Matching—originally designed for reward fine-tuning—to the sampling setting.
By exploiting a factorization of the optimal transition density (a Schrödinger bridge), our new loss enables heavy reuse of simulations and energy evaluations.
By exploiting a factorization of the optimal transition density (a Schrödinger bridge), our new loss enables heavy reuse of simulations and energy evaluations.
May 1, 2025 at 1:34 AM
We specialize Adjoint Matching—originally designed for reward fine-tuning—to the sampling setting.
By exploiting a factorization of the optimal transition density (a Schrödinger bridge), our new loss enables heavy reuse of simulations and energy evaluations.
By exploiting a factorization of the optimal transition density (a Schrödinger bridge), our new loss enables heavy reuse of simulations and energy evaluations.
New paper out with FAIR(+FAIR-Chemistry):
Adjoint Sampling: Highly Scalable Diffusion Samplers via Adjoint Matching
We present a scalable method for learning to sample *conditionally* from unnormalized densities beyond classical force fields.
📄: arxiv.org/abs/2504.11713
Adjoint Sampling: Highly Scalable Diffusion Samplers via Adjoint Matching
We present a scalable method for learning to sample *conditionally* from unnormalized densities beyond classical force fields.
📄: arxiv.org/abs/2504.11713
May 1, 2025 at 1:34 AM
New paper out with FAIR(+FAIR-Chemistry):
Adjoint Sampling: Highly Scalable Diffusion Samplers via Adjoint Matching
We present a scalable method for learning to sample *conditionally* from unnormalized densities beyond classical force fields.
📄: arxiv.org/abs/2504.11713
Adjoint Sampling: Highly Scalable Diffusion Samplers via Adjoint Matching
We present a scalable method for learning to sample *conditionally* from unnormalized densities beyond classical force fields.
📄: arxiv.org/abs/2504.11713