The paper investigates whether incorporating an evidence-support signal into retrieval evaluation for retrieval‑augmented generation (RAG) improves downstream decision‑making. Across multiple benchmarks and a TREC RAG 2025 setting, the evidence signal alters retriever rankings but its benefits vary: it does not consistently enhance retriever training, its usefulness for system selection depends on generator instructions, and it does not reliably predict answer quality on unseen topics. Human filtering of evidence‑rich passages preserves useful content, yet evaluators disagree on whether this improves final answers, indicating that evidence‑aware evaluation alone does not guarantee better downstream outcomes.
By Utshab Kumar Ghosh, Debayan Mukhopadhyay, Shubham Chatterjee
arXiv:2609.37491v1 Announce Type: cross
Abstract: Retrieval-augmented language models are expected to answer from the retrieved evidence, but in practice they often keep answering when that evidence...
By Zeyan Li, Qirong Guo, SIyuan Qiu, Hu Xu, Chun Li, Jianfeng Xu
arXiv:2609.37472v1 Announce Type: cross
Abstract: Behavioral tests measure how a language model reads evidence. We ask whether those measurements help choose a recommendation interface. We evaluate s...
By Han Chen, Yingrui Li
DynaKRAG is a unified framework that learns a state‑conditioned policy to control evidence acquisition in multi‑hop retrieval‑augmented generation. It uses a deterministic validity layer to build an action set, a learned continuation gate to decide between generating an answer or gathering more evidence, and an advantage scorer to rank evidence operations by predicted gain. Across HotpotQA, 2Wiki, and MuSiQue with various backbone models, DynaKRAG achieves top EM and F1 scores, improves token and retrieval efficiency, and enables terminal evidence compression that reduces context size while boosting answer quality.
By Chenyu Zhou, Yaqi Wu, Xiaolei Guo, Jiaqi Huang, Xianfa Zhang, Junxu Zhang, Zhuo Yu, Zhubo Shi, Jianghao Lin, Dongdong Ge
EviScope is a new paired counterfactual benchmark that evaluates grounded language models by fixing the question while manipulating evidence—adding, removing, distracting, or contradicting it. The v1.1 dataset includes 40 four‑condition quartets with repaired counterfactual claims and span‑level support labels for automated assessment. Experiments on Qwen2.5‑7B, Llama 3.1 8B, and Gemini 3.5 Flash show that paired metrics reveal grounding behaviors hidden by simple answer accuracy, such as unsupported answers, conflict blindness, and incorrect non‑answer actions.
By Suryadeep Singh Deswal
XHotpotQA is a new benchmark for cross‑lingual knowledge composition in multi‑hop question answering. It presents each instance as an evidence‑dependency graph with explicit language assignments for the question, bridge evidence, answer‑bearing evidence, and distractors, and includes 15,661 training and 7,405 validation examples with sentence‑level support supervision. The dataset reveals significant performance drops when evidence spans language boundaries, providing a diagnostic tool for systems that must integrate evidence across languages.
By Iman Barati, Arash Ghafouri, Behrouz Minaei-Bidgoli