CuSearch: Curriculum Rollout Sampling via Search Depth for Agentic RAG

TL;DR

CuSearch introduces a curriculum sampling method that prioritizes deeper-search trajectories in RLVR-based agentic RAG training, leading to consistent performance improvements by leveraging search depth as a proxy for supervision density.

Contribution

The paper proposes CuSearch, a novel curriculum rollout sampling framework that dynamically reallocates training focus toward deeper trajectories based on search depth.

Findings

CuSearch improves performance by up to 11.8 exact-match points over standard methods.

Deeper-search trajectories provide denser supervision and are more informative for training.

Search depth serves as an effective, annotation-free proxy for supervision density in RLVR.

Abstract

Reinforcement Learning with Verifiable Rewards (RLVR) has emerged as a promising paradigm for training agentic retrieval-augmented generation (RAG) systems from outcome-only supervision. Most existing methods optimize policies from uniformly sampled rollouts, implicitly treating all trajectories as equally informative. However, trajectories differ substantially in search depth and are therefore not equally informative: deeper-search trajectories contain more retrieval decision points and provide denser direct supervision for the retrieval sub-policy. Moreover, this heterogeneity grows over training as the within-batch depth distribution shifts toward higher values, yet uniform rollout sampling remains blind to this shift. To address this, we propose CuSearch, a curriculum rollout sampling framework built on Search-Depth Greedy Allocation (SDGA), a batch-level operator that reallocates a fixed update budget toward deeper-search trajectories. SDGA-Auto always targets the deepest available trajectories in the current batch, yielding an implicit training-aligned curriculum as the depth distribution shifts upward. SDGA-Phase explicitly advances the curriculum threshold as deeper trajectories become sufficiently abundant. Experiments across model types and retrieval frameworks show that CuSearch consistently improves performance, achieving up to 11.8 exact-match points over standard GRPO on ZeroSearch. These results establish per-trajectory search depth as a reliable, annotation-free proxy for retrieval supervision density in RLVR-based agentic RAG training.