EchoSpec: Training Search Agents to Weigh Evidence Above Echoes
Shihao Yang, Can Zeng, Xinzhe Li
Abstract
Search agents trained with reinforcement learning decide what to retrieve, when to search again, and when to answer. In real corpora, one claim is copied and paraphrased across many pages while the primary source settling a question appears once. We show that reinforcement-learned search agents grant echoes the credence of independent sources and lose their ability to abstain. Across five base models from 3B to 32B and seven released RL agents on multi-hop Wikipedia questions with primary evidence intact, eight paraphrased false-claim echoes overturn 58-73% of previously correct answers, while exact-match RL drives deleted-evidence abstention below 15%. Conditioning on retrieval localizes the failure to evidence acquisition: once primary evidence enters context, agents overrule eight echoes as reliably as one, whereas echoes fill top ranks and halt search early, so larger budgets, forced search turns, or passive passage filtering alone fail to recover displaced evidence. We introduce EchoSpec, specifying reliability across five corpus transformations with known target answers and training in two stages. Verification warm-start teaches multi-hop bridge-entity recovery past echoes, after which group RL optimizes an evidence-gated λ-reward with an outcome-independent acquisition bonus and scale-preserving advantages retaining the optimal rule λ/(1+λ). Across Qwen2.5-3B, Qwen2.5-7B, and Llama-3.1-8B backbones on nine evaluation suites, EchoSpec cuts eight-echo deference by 30-44 points on Qwen2.5-7B down to 29-31%, nearly quadruples eight-echo reach to 26-27% with zero-shot transfer to unseen BGE and BM25 retrievers, and raises standalone QA precision to 47-55%. Code is publicly available at https://github.com/unihsy/EchoSpec.
Create a lesson
Related papers
Cost-Aware Agentic Feature Optimization System for Fleet-Scale Ranking Models
Chang Yang, Yao Xuan, Wendan Yan et al.
Democratizing Cross-Project Knowledge for Transforming Water Governance and Climate Resilience: an Agentic AI Ecosystem
Oriol Alàs, Ian Palacín, Iván Cester et al.
From Sessions to Serving: Integrating LLM-Based Embeddings into Retrieval for Production Content Recommendation
Zhixiong Hu, Meng Li, Maggie Zhang et al.
A Task-Level Bias Taxonomy for Historical Image Retrieval
Omkar Gavali, Ratan Sebastian, Christoph Rippe et al.
LINEAGE: A Benchmark for Empirical Lineage Retrieval in AI Scientific Literature
Pangjing Wu, Xu Yuan, Wenqi Fan et al.
CACO: Coverage-Aware Coordinated Optimization for Continuous Generative Retrieval
Pan Zhang, Nan Zhuang, Yongkang Zhang et al.