SALT is a subspace-adaptive plug-in for GRPO that decomposes group-relative coefficients into shared and residual channels using mini-batch Gram geometry and amplifies residuals to mitigate signed cancellation in RLVR.
Effective dimensionality: A tutorial.Multivariate Behavioral Research, 56(3):527–542, 2021
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
years
2026 2verdicts
UNVERDICTED 2representative citing papers
Modifies Gibbs sampler for GP state-space models, introduces CFA measurement structure, and validates software via simulation-based calibration to enable reliable learning of nonlinear latent dynamics.
citing papers explorer
-
SALT: When More Rollouts Don't Help in Group-Based Policy Optimization and How to Make Them Matter
SALT is a subspace-adaptive plug-in for GRPO that decomposes group-relative coefficients into shared and residual channels using mini-batch Gram geometry and amplifies residuals to mitigate signed cancellation in RLVR.
-
Learning Nonlinear Dynamics: Improving the Estimation Efficiency and Reliability of Gaussian Process State-Space Models
Modifies Gibbs sampler for GP state-space models, introduces CFA measurement structure, and validates software via simulation-based calibration to enable reliable learning of nonlinear latent dynamics.