Entity · paper

RL Post-Training Builds Compositional Reasoning Strategies

paperactiveprovisionalrl-post-training-builds-compositional-reasoning-strategies-f151dd3b·1 events·first seen Jul 9, 2026

Aliases: RL Post-Training Builds Compositional Reasoning Strategies

More like this (12)

Understanding Reasoning from Pretraining to Post-Training Learning to Reason by Analogy via Retrieval-Augmented Reinforcement Fine-Tuning Long-context Reasoning Benchmarks ExpRL: Exploratory RL for LLM Mid-Training Bridging the Gap Between Latent and Explicit Reasoning with Looped Transformers Reasoning Enhancement Reference-Augmented Training Adaptive Parallel Reasoning Pessimism's Paradox: Conservative Offline Training Amplifies Reward Hacking During Online Adaptation in Reasoning Models Compositional Reasoning Depth Predicts Clinical AI Failure Reasoning in Memory (RiM)Reasoning Language Models

Recent events (1)

6arXiv · cs.CL·Jul 9, 2026·source ↗

RL post-training composes primitive skills into higher-level reasoning strategies, not just amplification

A new arXiv paper investigates whether RL post-training creates genuinely new compositional reasoning strategies or merely amplifies latent base-model skills. Using a controlled rewrite-grammar environment with auditable traces, the authors show that RL solves problems the pretrained model rarely solves even with large sampling budgets, and that it does so by reorganizing primitive competencies into sequential and parallel compositional procedures. The key differentiator from rejection fine-tuning is selectivity: RL concentrates exploration into valid, reusable structures rather than producing high-volume but shortcut-laden rewrites.

Evaluation and Benchmarking Alignment and RLHF RL Post-Training Builds Compositional Reasoning Strategies