Harness-G: A Graph-Structured Harness for Search Agents

• URL: http://arxiv.org/abs/2607.27652v2
• Date: Sat, 01 Aug 2026 05:12:49 GMT
• Title: Harness-G: A Graph-Structured Harness for Search Agents
• Abstract summary: Reinforcement learning (RL) search agents commonly model retrieval as free-form natural-language query generation.<n>We propose Harness-G, a graph-structured retrieval framework that redesigns this interface.<n>It reformulates free-form query generation as finite action selection, while the environment constructs the menu, tracks retrieval state, and validates and executes each choice.<n>It achieves the highest average F1 at both evaluated model scales, outperforming the strongest baseline, Graph-R1, by 10.74 points at 1.5B and 3.98 points at 3B.
• Score: 17.554676400013424
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Reinforcement learning (RL) search agents commonly model retrieval as free-form natural-language query generation and optimize multi-turn interactions using final-answer rewards. Current studies mainly improve training with denser or more structured credit signals, but rarely examine whether retrieval is properly formulated at the policy-environment interface. We observe pronounced retrieval aliasing during Search-R1 training: rollouts for the same question continue to generate distinct query strings, yet their accumulated evidence sets increasingly overlap. We call this phenomenon retrieval-equivalence collapse; in this regime, trajectories approach utility equivalence with respect to retrieval decisions, leaving within-group returns with little effective retrieval contrast. To address this problem, we propose Harness-G, a graph-structured retrieval framework that redesigns this interface. It reformulates free-form query generation as finite action selection: the policy selects an evidence sentence or entity, or chooses to answer, while the environment constructs the menu, tracks retrieval state, and validates and executes each choice. This interface reduces linguistic aliasing and makes same-state alternatives directly comparable. Building on this interface, we introduce Structured Non-myopic Credit (SNC), which uses a frozen answer scorer to compare the selected action with its alternatives and assigns downstream gains to the earlier actions that enabled them. Across six QA benchmarks, Harness-G achieves the highest average F1 at both evaluated model scales, outperforming the strongest baseline, Graph-R1, by 10.74 points at 1.5B and 3.98 points at 3B.

Related papers

• Learning from Failures: Retrieval-Centric CoT via Hard Negatives for Unified Multimodal Retrieval [17.695551248974215]
  UniME-R1 is an embedder-adviser framework that learns to reason over initially retrieved candidates.<n>UniME-R1 consistently improves retrieval performance over strong baselines.
  arXiv  Detail & Related papers  (2026-08-06T14:04:56Z)
• MMAgent-R$^2$: Learning to Rerank and Reject for Agentic mRAG [50.46218163777298]
  Knowledge-based Visual Question Answering (KB-VQA) requires models to retrieve visual entities matching the query image from large-scale encyclopedic knowledge bases and answer related questions.<n>We propose MMAgent-R$2$, an agentic mRAG framework that integrates visual reranking and active rejection as its internal verification mechanism.
  arXiv  Detail & Related papers  (2026-07-08T13:15:39Z)
• An Embarrassingly Simple Graph Heuristic Reveals Shortcut-Solvable Benchmarks for Sequential Recommendation [50.09718257952108]
  Sequentialsolv has increasingly shifted toward generative recommenders that combine sequential patterns with semantic item information.<n>These methods are often evaluated on a small set of widely used benchmarks, raising a key question: do these benchmarks actually require the advanced modeling capabilities that modern generative recommenders claim to provide?<n>We conduct a benchmark audit with an intentionally simple graph. Starting from only the last one or two items, it retrieves candidates from a few-hop item-transition graph and ranks them by item-feature similarity.
  arXiv  Detail & Related papers  (2026-05-08T02:00:11Z)
• R$^3$AG: Retriever Routing for Retrieval-Augmented Generation [59.47703698994575]
  R$3$AG decomposes retriever capability into two learnable dimensions: retrieval quality and generation utility.<n>R$3$AG consistently outperforms both the best individual retrievers and state-of-the-art static routing methods.
  arXiv  Detail & Related papers  (2026-04-22T06:51:20Z)
• CoSearch: Joint Training of Reasoning and Document Ranking via Reinforcement Learning for Agentic Search [51.911048955965136]
  CoSearch is a framework that jointly trains a multi-step reasoning agent and a generative document ranking model.<n>Our results show that joint training of the reasoning agent and retrieval system is both feasible and strongly performant.
  arXiv  Detail & Related papers  (2026-04-19T17:48:17Z)
• Retrieval as Generation: A Unified Framework with Self-Triggered Information Planning [32.86954579218087]
  We revisit retrieval-augmented generation (RAG) by embedding retrieval control directly into generation.<n>Instead of treating retrieval as an external intervention, we express retrieval decisions within token-level decoding.<n>GRIP is a unified framework in which the model regulates retrieval behavior through control-token emission.
  arXiv  Detail & Related papers  (2026-04-13T12:53:17Z)
• Breaking the Static Graph: Context-Aware Traversal for Robust Retrieval-Augmented Generation [12.71443292660797]
  We propose CatRAG, Context-Aware Traversal for robust RAG.<n>CatRAG builds on the HippoRAG 2 architecture and transforms the static KG into a query-adaptive navigation structure.<n> Experiments across four multi-hop benchmarks demonstrate that CatRAG consistently outperforms state of the art baselines.
  arXiv  Detail & Related papers  (2026-02-02T11:13:38Z)
• Chain-of-Retrieval Augmented Generation [91.02950964802454]
  This paper introduces an approach for training o1-like RAG models that retrieve and reason over relevant information step by step before generating the final answer.<n>Our proposed method, CoRAG, allows the model to dynamically reformulate the query based on the evolving state.
  arXiv  Detail & Related papers  (2025-01-24T09:12:52Z)
• Autoregressive Search Engines: Generating Substrings as Document Identifiers [53.0729058170278]
  Autoregressive language models are emerging as the de-facto standard for generating answers. Previous work has explored ways to partition the search space into hierarchical structures. In this work we propose an alternative that doesn't force any structure in the search space: using all ngrams in a passage as its possible identifiers.
  arXiv  Detail & Related papers  (2022-04-22T10:45:01Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.