Searcharxiv⌕ Search
Home/Archive/arXiv

arXiv · 2611.01987

Same Trajectory, Different Knowledge: Information-State-Aware On-Policy Distillation for Search Agents

Minhao Sun·Xian-Ling Mao·Eric Tao·Zi-Ao Ma·Zhijing Wu
View original source ↗Read source document ↗Export metadata ↓

Abstract

On-policy distillation trains search agents with teacher feedback on student-generated trajectories. Recent methods augment the teacher's context with supporting evidence, reference answers, or hindsight information. However, this privileged information may allow the teacher to rely on facts that the student cannot reliably recover from its current state. The resulting information-state mismatch can lead the teacher to undervalue retrieval that helps the student establish those facts. To address this mismatch, we introduce Information-State-Aware Distillation (ISA-Distill), which adapts teacher feedback to the student's factual recoverability. ISA-Distill estimates how reliably the student can recover action-relevant knowledge and derives teacher feedback for the same sampled action from two complementary views. The fact-use view supplies the teacher with the relevant fact, producing feedback that guides the student in using knowledge. The evidence-seeking view instead presents the information need, producing feedback that guides the student in acquiring supporting evidence. The two signals are weighted toward fact use when recoverability is high and toward evidence seeking when it is low, and the resulting feedback guides policy optimization. Experiments across diverse QA benchmarks and model scales demonstrate improved answer accuracy and retrieval efficiency. Our code and model checkpoints will be open-sourced.

Explore related subjects

cs.IR

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Minhao Sun, Xian-Ling Mao, Eric Tao, Zi-Ao Ma, Zhijing Wu. 2026-11-01. Same Trajectory, Different Knowledge: Information-State-Aware On-Policy Distillation for Search Agents. https://arxiv.org/abs/2611.01987

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

DAT: Dynamic Alpha Tuning for Hybrid Retrieval in Retrieval-Augmented Generation

Hybrid retrieval techniques in Retrieval-Augmented Generation (RAG) systems enhance information retrieval by combining dense and sparse (e.g., BM25-based) retrieval methods. However, existing approaches struggle with adaptability, as fixed weighting schemes fail to adjust to different queries. To address this, we propose DAT (Dynamic Alpha Tuning), a novel hybrid retrieval framework that dynamically balances dense retrieval and BM25 for each query. DAT leverages a large language model (LLM) to evaluate the effectiveness of the top-1 results from both retrieval methods, assigning an effectiveness score to each. It then calibrates the optimal weighting factor through effectiveness score normalization, ensuring a more adaptive and query-aware weighting between the two approaches. Empirical results show that DAT consistently significantly outperforms fixed-weighting hybrid retrieval methods across various evaluation metrics. Even on smaller models, DAT delivers strong performance, highlighting its efficiency and adaptability.

cs.IR↗
arXiv2026

Explaining Group Recommendations via Counterfactuals

Group recommender systems help users make collective choices but often lack transparency, leaving group members uncertain about why items are suggested. Existing explanation methods focus on individuals, offering limited support for groups where multiple preferences interact. In this work, we introduce and formalize the problem of finding counterfactual explanations for group recommendations. We introduce a comprehensive set of item- and explanation-level metrics that capture different aspects, such as relevance, popularity, explanation utility, interpretability, and fairness. We design five algorithms that employ techniques such as search-space pruning, greedy selection, early stopping, and incremental refinement to reduce computational cost and the number of recommender calls while producing high-quality explanations. The algorithms consider different trade-offs; for example, some prioritize computational efficiency at the expense of explanation quality, while others focus on improving explanation quality at a higher computational cost. Experiments on real-world datasets analyze these trade-offs in terms of explanation size, computational cost, utility, and fairness.

cs.IR↗
arXiv2026

MuonRec: Shifting the Optimizer Paradigm Beyond Adam in Scalable Generative Recommendation

Recommender systems (RecSys) are increasingly emphasizing scaling, leveraging larger architectures and more interaction data to improve personalization. Yet, despite the optimizer's pivotal role in training, modern RecSys pipelines almost universally default to Adam/AdamW, with limited scrutiny of whether these choices are truly optimal for recommendation. In this work, we revisit optimizer design for scalable recommendation and introduce MuonRec, the first framework that brings the recently proposed Muon optimizer to RecSys training. Muon performs orthogonalized momentum updates for 2D weight matrices via Newton-Schulz iteration, promoting diverse update directions and improving optimization efficiency. We develop an open-source training recipe for recommendation models and evaluate it across both traditional sequential recommenders and modern generative recommenders. Extensive experiments demonstrate that MuonRec reduces converged training steps by an average of 32.4\% while simultaneously improving final ranking quality. Specifically, MuonRec yields consistent relative gains in NDCG@10, averaging 12.6\% across all settings, with particularly pronounced improvements in generative recommendation models. These results consistently outperform strong Adam/AdamW baselines, positioning Muon as a promising new optimizer standard for RecSys training. Our code is available.

cs.IR↗