For bounded real-valued function classes, uniform convergence at scale γ, agnostic learnability at γ/2, and finite fat-shattering dimension above γ are equivalent.
arXiv preprint arXiv:2510.15020 , year=
10 Pith papers cite this work. Polarity classification is still indexing.
10
Pith papers citing it
fields
cs.LG 10years
2026 10representative citing papers
OrderGrad supplies unbiased likelihood-ratio and reparameterization gradient estimators for finite-sample L-statistics by applying a rank-based reward transformation usable in standard policy-gradient updates.
Discrete diffusion models learn data support before frequencies because the exact reverse process decomposes edits into a dominant validity scale and a finer probability coefficient.
SFT supplies entangled compositional traces of atomic skills and routing modules; RL identifies those modules and enables recombination on novel compositions outside the SFT support.
Autocurriculum decomposition for semiautomata simulation achieves 2^O(sqrt(log T)) sample complexity under interactive feedback and relaxes reference model coverage to block length B << T under RLVR, versus Omega(T) for direct methods.
Proposes MaxPO using a Leave-Two-Out baseline for centered unbiased advantages in max@K policy gradients, with a unified derivation of finite-batch estimators.
Experiments indicate RL applied early in pre-training often matches full SFT-then-RL performance, targeted data composition outweighs scale for RL success, and averaging RL and SFT objectives outperforms sequential or single methods.
A new pipeline uses interpretability to characterize concepts in preference data and shape rewards via feature or data interventions during LM post-training.
Limited generator access in autoregressive post-training confines learners to root-start rollouts whose value is bounded by on-policy prefix probabilities, while weak prefix control unlocks richer observations and produces an exponential gap in KL-regularized outcome-reward training.
citing papers explorer
-
Scale-Sensitive Shattering: Learnability and Evaluability at Optimal Scale
For bounded real-valued function classes, uniform convergence at scale γ, agnostic learnability at γ/2, and finite fat-shattering dimension above γ are equivalent.
-
OrderGrad: Optimizing Beyond the Mean with Order-Statistic Policy Gradient Estimation
OrderGrad supplies unbiased likelihood-ratio and reparameterization gradient estimators for finite-sample L-statistics by applying a rank-based reward transformation usable in standard policy-gradient updates.
-
Support Before Frequency in Discrete Diffusion
Discrete diffusion models learn data support before frequencies because the exact reverse process decomposes edits into a dominant validity scale and a finer probability coefficient.
-
From Reasoning Traces to Reusable Modules: Understanding Compositional Generalization in Language Model Reasoning
SFT supplies entangled compositional traces of atomic skills and routing modules; RL identifies those modules and enables recombination on novel compositions outside the SFT support.
-
Learning to Reason with Curriculum II: Compositional Generalization
Autocurriculum decomposition for semiautomata simulation achieves 2^O(sqrt(log T)) sample complexity under interactive feedback and relaxes reference model coverage to block length B << T under RLVR, versus Omega(T) for direct methods.
-
On Advantage Estimates for Max@K Policy Gradients
Proposes MaxPO using a Leave-Two-Out baseline for centered unbiased advantages in max@K policy gradients, with a unified derivation of finite-batch estimators.
-
RL Excursions during Pre-Training: Re-examining Policy Optimization for LLM training
Experiments indicate RL applied early in pre-training often matches full SFT-then-RL performance, targeted data composition outweighs scale for RL success, and averaging RL and SFT objectives outperforms sequential or single methods.
-
Anatomy of Post-Training: Using Interpretability to Characterize Data and Shape the Learning Signal
A new pipeline uses interpretability to characterize concepts in preference data and shape rewards via feature or data interventions during LM post-training.
-
The Role of Generator Access in Autoregressive Post-Training
Limited generator access in autoregressive post-training confines learners to root-start rollouts whose value is bounded by on-policy prefix probabilities, while weak prefix control unlocks richer observations and produces an exponential gap in KL-regularized outcome-reward training.
- A Theoretical Framework for Statistical Evaluability of Generative Models