Pith. sign in

REVIEW 1 major objections 1 minor 109 references

Mitigating Context Interference for Reliable and Efficient Search Agents

T0 review · 1 major / 1 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A multi-turn search agent's accuracy and speed are improved by refining the newest retrieved documents, not the accumulated history, and the paper shows how to train a small refiner and fold it into reinforcement learning.

desk verdict Solid empirical study of context interference in multi-turn search agents; the distilled refiner works, but the RL-training benefit is confounded with inference-time refinement. read the letter →

arxiv 2608.10743 v1 pith:VUAZDUBM submitted 2026-08-11 cs.CL

Boyang Xue , Bin Wu , Shuofei Qiao , Sheng Wang , Rui Wang , Yiming Du , Hongru Wang , Jeff Z. Pan
show 3 more authors
Emine Yilmaz Kam-Fai Wong Aldo Lipani
This is my paper · ORCID
classification cs.CL
keywords contextinterferencemulti-turnsearchagentsrefinementreinforcementlearningretrieval-augmentedgenerationquestionansweringdistillationefficiency
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Multi-turn search agents get long contexts, and the paper argues that the newest batch of retrieved documents is what distracts them most; earlier documents, earlier queries, and earlier reasoning are much less harmful. To remove that interference, the authors train a small context refiner that, given the current search query and the top documents, outputs only the query-relevant sentences, and they replace the full retrieved documents with that output at every turn. They then fold the refiner into reinforcement learning (CRRL), so training rollouts are also generated over refined contexts. Across seven single- and multi-hop question-answering benchmarks, the refined agents improve exact-match accuracy and need fewer retrieval calls, so the paper concludes that refine context and then generate is a workable recipe for cheaper, more reliable search agents.

What carries the argument

Two components carry the argument. The first is the context refiner $F$: a small instruction-tuned language model fine-tuned to map a search query $q_{i-1}$ and a retrieved document set $\mathbf{d}_i$ to a short refined text $\tilde{\mathbf{d}}_i$ containing only the information relevant to answering $q_{i-1}$, trained on distilled teacher extractions that an entailment check keeps faithful to the source. The second is CRRL, a GRPO-based RL procedure whose rollouts use refined observations $\tilde{\mathbf{d}}_i$ instead of raw documents, so the policy learns search and reasoning behaviour on interference-free contexts while token-level loss masking keeps retrieved tokens out of the policy gradient. The refiner is what removes the noise identified in the causal analysis; CRRL is what makes refinement part of learning rather than a separate preprocessing step.

What would settle it

Train Search-GRPO exactly as in the paper and then run it at inference with the context refiner switched on, without any refinement-aware training. If that control matches CRRL's exact-match scores, the paper's claim that refinement must enter RL training is not supported; if CRRL still wins, the training-time contribution is real.

Watch

Extended reading notes

Core claim

Context interference in a multi-turn search agent is caused mainly by the latest observation—the top-$K$ documents returned by the most recent retrieval—rather than by older retrieved documents, past search queries, or old reasoning steps, which the paper shows by masking segments of the history one at a time. On that basis the authors introduce a context refiner $F$ trained by supervised fine-tuning on a distilled dataset of query-relevant extractions produced by a stronger teacher and verified by an entailment model to stay strictly inside the source documents. The refiner replaces each observation $\mathbf{d}_i$ with $\tilde{\mathbf{d}}_i=F(q_{i-1},\mathbf{d}_i)$, and the resulting cleaned contexts are used both at inference and inside GRPO rollouts in the proposed CRRL; the claim is that refinement during RL training further improves reliability (exact match) and efficiency (average retrieval times, context length, and inference time) over retrieval-based and RL baselines on all seven datasets.

Load-bearing premise

The central comparison assumes that CRRL's gains come from refinement during RL training, since the refiner is switched on at CRRL inference but not at the baseline's inference, and no baseline-plus-refiner control is reported.

Editorial extensions

If this is right

  • Search-agent accuracy can be improved by fixing the input context, not only by scaling the policy model.
  • Average retrieval calls drop, so per-question latency and cost fall even after accounting for the refiner's extra inference.
  • A weak model can approximate a strong teacher's context refinement via distillation, reducing dependence on expensive teacher calls at inference time.
  • Rollout quality in RL training matters enough that context hygiene during training yields better final policies than applying refinement only at inference.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The localization of interference to the latest retrieved documents suggests a cheap diagnostic for deployed agents: if accuracy collapses just after a retrieval turn, the problem is likely the newest documents, not the accumulated history.
  • The same distillation-plus-entailment recipe could transfer to other observation-heavy agent settings, such as tool outputs or code execution logs, but the interference source would have to be re-localized for each setting rather than assumed to be the latest observation.
  • A direct control that applies the refiner at inference time to the RL-trained baseline without refinement-aware training would separate the value of training on clean contexts from the value of reading clean contexts; this is the comparison the current table does not report.
  • If the refiner is also used at inference for CRRL, its efficiency gains may partly reflect a shorter effective context length rather than a better policy, which matters for how the method would port to agents with longer action budgets.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

1 major / 1 minor

Summary. This paper studies context interference in multi-turn LLM search agents, addressing three questions: which parts of the context cause interference, how to refine contexts to mitigate it, and whether incorporating context refinement into RL training improves search agents. The authors report that masking previous retrieved documents, queries, and thinking steps yields only small gains, from which they conclude that the latest retrieved documents are the primary source of interference. They then distill a context refinement dataset from GPT-4 and train a small context refiner that extracts query-relevant information from retrieved documents. Finally, they integrate this refiner into a GRPO-based RL training pipeline (CRRL), where rollouts refine retrieved documents before the policy generates actions. Experiments on seven QA benchmarks with two Qwen models compare inference-time refinement methods and training-based methods, reporting improvements in Exact Match and reductions in average retrieval times, context length, and inference time. The paper also includes an analysis of recall rate versus recall accuracy and a discussion of data contamination.

Significance. If the conclusions hold, the paper provides a clear and actionable finding: in multi-turn search agents, the latest retrieved documents are the dominant source of context interference, and a distilled, relatively small context refiner can mitigate it at inference time and during RL training. The strengths are the breadth of evaluation (seven QA datasets, two model sizes, multiple efficiency metrics), the consistency of improvements in the inference-time refinement results (Table 2), the release of code, and the careful separation of training and test data. The paper also draws attention to an important problem—context interference in agentic search—that is less studied than in single-turn RAG. However, the central claim about the benefit of incorporating refinement into RL training is not cleanly isolated: the CRRL comparison in Table 3 varies both the training contexts and the inference-time context processing, and no control separates these factors. The causal attribution to 'latest documents' in Section 3.2 is also indirect.

major comments (1)
  1. [Sec. 4.3, Tables 3] The main training results are reported without variance or multiple seeds. The differences between CRRL and Search-o1 in Table 3 are small (36.6 vs 36.2 EM for Qwen2.5-7b; 31.5 vs 30.3 for Qwen2.5-3b), and the 2.0-point improvement over Search-GRPO on the 7b model could be within run-to-run noise, especially given that the Appendix D training details mention instability ('In cases where training diverges, we evaluate at the most recent stable checkpoint'). Reporting the mean and standard deviation over at least three seeds for the headline comparisons, or at least for CRRL and the strongest baselines, is necessary to support the 'significantly improve' language used in the abstract and Section 4.3.
minor comments (1)
  1. [Sec. 3.3] Minor typo: 'close performance to GPT-Refine' should be 'close performance to GPT-Refine' or 'performance close to GPT-Refine's'; also 'a distill dataset' is more naturally 'a distilled dataset'.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: the refiner is trained by distillation and compared with its teacher, and CRRL is a standard GRPO extension; the missing inference-refinement control is an experimental confound, not circularity.

full rationale

Walking the paper's derivation chain, no claim reduces by construction to its own inputs. The RQ i conclusion that interference primarily arises from the latest retrieved documents is supported by masking ablations (IRCoT-o, IRCoT-oq, IRCoT-oqp in Sec. 3.2), not by defining the latest document as the cause. The context refiner is trained via SFT on a distilled dataset produced by GPT-4 with entailment verification, and is then evaluated against GPT-Refine, GPT-Compress, and Self-Refine (Table 2); this is a standard distillation comparison, and the close match to the teacher is an empirical result rather than a circular equivalence. CRRL (Sec. 4.1) applies a fixed auxiliary refiner during GRPO rollouts and at inference; its loss is a standard group-relative policy objective over refined contexts, and no equation reintroduces the target metric as a fitted parameter. The paper's own limitation section notes that the refiner is an auxiliary module rather than being integrated into the agent's training pipeline, which accurately characterizes the method and does not create a circular step. The only substantive experimental weakness is that CRRL is evaluated with the refiner active at inference whereas Search-GRPO is not, and Table 2 shows inference-only refinement already provides most of the observed gain, so the marginal benefit of training on refined contexts is not cleanly isolated. That is a confound in attribution, not circular reasoning, and the paper's self-citations are background references rather than load-bearing justifications of the empirical claims. Therefore the circularity score is 0.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The paper introduces no new physical or mathematical entities. Its central claims rest on domain assumptions about ablation validity, attribution of the recall gap, data contamination, evaluation metrics, and entailment filtering, plus several hand-chosen experimental parameters. The most fragile assumptions are the two that support the 'latest documents are the primary source' conclusion.

free parameters (4)
  • Number of retrieved documents K = 3
    Top-K passages returned per search round; chosen by the authors and directly controls how much potential noise enters the context.
  • Ranking filter thresholds = 0.2, 0.5
    Used for the alternative ranking baseline in Appendix E.2; the authors note performance depends on the threshold choice.
  • Training corpus size for CRRL vs baselines = 40k for CRRL; 60k for training baselines
    Sec. 4.2 says CRRL uses 40k samples, while Appendix D says training baselines use 60k. This mismatch is unclarified and could affect comparison fairness.
  • GRPO hyperparameters = beta=0.001, clip=0.2, temperature=0.7, top_p=1.0, max action budget=8
    Standard RL hyperparameters chosen by hand; they influence training stability and final performance.
assumptions (5)
  • domain assumption Generated search queries and retrieved documents in a history are mutually independent with rarely sequential dependencies, so masking older components is a valid way to isolate interference sources.
    Invoked in Sec. 3.2 before defining the masked variants. If removing previous documents or queries changes later retrieval trajectories, the ablations do not cleanly measure interference.
  • domain assumption The persistence of the recall-rate versus recall-accuracy gap after removing older documents and queries is attributable to context interference in the latest observation, rather than to the LLM's inability to extract an embedded answer from relevant documents.
    Used in Sec. 3.2 and Figure 3 to support the 'latest documents primary cause' conclusion without directly ablating the latest observation.
  • domain assumption The 2018 Wikipedia dump is timeline-synchronized with the test sets, so the retrieved knowledge base serves as ground truth and does not contaminate parametric knowledge.
    Appendix C.1 argues that questions carry clear timeline information; this is the basis for the no-contamination claim.
  • domain assumption Exact Match and Average Retrieval Times are sufficient proxies for reliability and efficiency.
    Used throughout; EM is brittle for free-form answers, and ART ignores the computational cost of refinement and other non-retrieval overhead.
  • domain assumption The entailment model correctly verifies that refined texts are fully entailed by the original documents and add no extra knowledge.
    Sec. 3.3 relies on this verification to filter the distilled context refinement dataset.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Mitigating Context Interference for Reliable and Efficient Search Agents." pith.science (2026). https://pith.science/paper/VUAZDUBM

@misc{pith2026260810743,
  author       = {Pith},
  title        = {Pith review of: Mitigating Context Interference for Reliable and Efficient Search Agents},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VUAZDUBM}},
  note         = {Machine review of arXiv:2608.10743}
}
read the original abstract

Recent research empowers Large Language Models (LLMs) as multi-turn search agents to iteratively retrieve and generate outputs until complex tasks are solved. However, the contexts of multi-turn search agents are lengthy and complex. For example, the retrieved set of documents in each turn would inevitably introduce irrelevant information that distracts LLMs, referring to \textit{context interference}, potentially hindering the reliability and efficiency of search agents. Therefore, we conduct a systematic study on context interference in multi-turn search agents, focusing on investigating i) which parts of the context of search agents will contribute to the context interference, ii) how to refine the contexts of search agents to mitigate the interference, and iii) can incorporating context refinement into search agent training yield further improvements. We reveal that interference primarily arises from the latest retrieved documents. Based on the explored findings, we then introduce a distill-based context refiner to dynamically mitigate context interference for multi-turn search agents. Finally, we validate that incorporating context refinement into RL training pipelines of search agents can significantly enhance both reliability and efficiency. This study highlights the importance of mitigating context interference of search agents, inspiring a novel paradigm of ``refine context and then generate'' for AI agents.

Figures

Figures reproduced from arXiv: 2608.10743 by the authors.

Figure 1
Figure 1. Demonstration of how context interference [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Examples of the search agent with (a) con [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Demonstrations of context interference effects on four IRCoT variants of search agents. Results are [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Training pipeline for the distill-based context refiner. [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Averaged context lengths of search agents [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: Demonstration of our proposed CRRL method. [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: Demonstration of how contextual interference affects the performance of LLM search agents. “recall [PITH_FULL_IMAGE:figures/full_fig_p015_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

109 extracted references · 6 canonical work pages

  1. [1]

    Benchmarking Knowledge Boundary for Large Language Models: A Different Perspective on Model Evaluation

    Yin, Xunjian and Zhang, Xu and Ruan, Jie and Wan, Xiaojun. Benchmarking Knowledge Boundary for Large Language Models: A Different Perspective on Model Evaluation. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2024. doi:10.18653/v1/2024.acl-long.124

    doi:10.18653/v1/2024.acl-long.124 2024

  2. [2]

    Knowledge Boundary of Large Language Models: A Survey

    Li, Moxin and Zhao, Yong and Zhang, Wenxuan and Li, Shuaiyi and Xie, Wenya and Ng, See-Kiong and Chua, Tat-Seng and Deng, Yang. Knowledge Boundary of Large Language Models: A Survey. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025. doi:10.18653/v1/2025.acl-long.256

    doi:10.18653/v1/2025.acl-long.256 2025

  3. [3]

    Give Me the Facts! A Survey on Factual Knowledge Probing in Pre-trained Language Models

    Youssef, Paul and Kora. Give Me the Facts! A Survey on Factual Knowledge Probing in Pre-trained Language Models. Findings of the Association for Computational Linguistics: EMNLP 2023. 2023. doi:10.18653/v1/2023.findings-emnlp.1043

    doi:10.18653/v1/2023.findings-emnlp.1043 2023

  4. [4]

    Investigating the Factual Knowledge Boundary of Large Language Models with Retrieval Augmentation

    Ren, Ruiyang and Wang, Yuhao and Qu, Yingqi and Zhao, Wayne Xin and Liu, Jing and Wu, Hua and Wen, Ji-Rong and Wang, Haifeng. Investigating the Factual Knowledge Boundary of Large Language Models with Retrieval Augmentation. Proceedings of the 31st International Conference on Computational Linguistics. 2025

    2025

  5. [5]

    2025 , eprint=

    Fast, Slow, and Tool-augmented Thinking for LLMs: A Review , author=. 2025 , eprint=

    2025

  6. [6]

    2025 , eprint=

    LLMs Get Lost In Multi-Turn Conversation , author=. 2025 , eprint=

    2025

  7. [7]

    2025 , eprint=

    Beyond Single-Turn: A Survey on Multi-Turn Interactions with Large Language Models , author=. 2025 , eprint=

    2025

  8. [8]

    2025 , eprint=

    Enhancing Robustness in Large Language Models: Prompting for Mitigating the Impact of Irrelevant Information , author=. 2025 , eprint=

    2025

Show all 109 references
  1. [9]

    2025 , eprint=

    MA-RAG: Multi-Agent Retrieval-Augmented Generation via Collaborative Chain-of-Thought Reasoning , author=. 2025 , eprint=

    2025

  2. [10]

    2024 , eprint=

    RankRAG: Unifying Context Ranking with Retrieval-Augmented Generation in LLMs , author=. 2024 , eprint=

    2024

  3. [11]

    2025 , eprint=

    Context Engineering for Multi-Agent LLM Code Assistants Using Elicit, NotebookLM, ChatGPT, and Claude Code , author=. 2025 , eprint=

    2025

  4. [12]

    2025 , eprint=

    A Survey on Recent Advances in LLM-Based Multi-turn Dialogue Systems , author=. 2025 , eprint=

    2025

  5. [13]

    2018 , eprint=

    High-Dimensional Continuous Control Using Generalized Advantage Estimation , author=. 2018 , eprint=

    2018

  6. [14]

    Li and Y

    Zhihong Shao and Peiyi Wang and Qihao Zhu, Runxin Xu and Junxiao Song and Mingchuan Zhang and Y.K. Li and Y. Wu and Daya Guo , title =. CoRR , volume =. 2024 , url =

    2024

  7. [15]

    2023 , eprint=

    ReAct: Synergizing Reasoning and Acting in Language Models , author=. 2023 , eprint=

    2023

  8. [16]

    2025 , eprint=

    DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning , author=. 2025 , eprint=

    2025

  9. [17]

    2025 , eprint=

    Search-R1: Training LLMs to Reason and Leverage Search Engines with Reinforcement Learning , author=. 2025 , eprint=

    2025

  10. [18]

    2025 , eprint=

    R1-Searcher: Incentivizing the Search Capability in LLMs via Reinforcement Learning , author=. 2025 , eprint=

    2025

  11. [19]

    Proceedings of the ACM on Web Conference 2025 , pages =

    Dong, Guanting and Zhu, Yutao and Zhang, Chenghao and Wang, Zechen and Wen, Ji-Rong and Dou, Zhicheng , title =. Proceedings of the ACM on Web Conference 2025 , pages =. 2025 , isbn =. doi:10.1145/3696410.3714717 , abstract =

    2025

  12. [20]

    2023 , eprint=

    In-context Interference in Chat-based Large Language Models , author=. 2023 , eprint=

    2023

  13. [21]

    LLM Task Interference: An Initial Study on the Impact of Task-Switch in Conversational History

    Gupta, Akash and Sheth, Ivaxi and Raina, Vyas and Gales, Mark and Fritz, Mario. LLM Task Interference: An Initial Study on the Impact of Task-Switch in Conversational History. Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing. 2024. doi:10....

    2024 doi

  14. [22]

    2025 , eprint=

    Unable to Forget: Proactive Interference Reveals Working Memory Limits in LLMs Beyond Context Length , author=. 2025 , eprint=

    2025

  15. [23]

    NPJ digital medicine , volume=

    A study of generative large language model for medical research and healthcare , author=. NPJ digital medicine , volume=. 2023 , publisher=

    2023

  16. [24]

    Proceedings of the Fourth ACM International Conference on AI in Finance , pages =

    Li, Yinheng and Wang, Shaofei and Ding, Han and Chen, Hang , title =. Proceedings of the Fourth ACM International Conference on AI in Finance , pages =. 2023 , isbn =. doi:10.1145/3604237.3626869 , abstract =

    2023

  17. [25]

    ACM computing surveys , volume=

    Survey of hallucination in natural language generation , author=. ACM computing surveys , volume=. 2023 , publisher=

    2023

  18. [26]

    ACM Transactions on Information Systems , volume=

    Dense text retrieval based on pretrained language models: A survey , author=. ACM Transactions on Information Systems , volume=. 2024 , publisher=

    2024

  19. [27]

    2024 , eprint=

    Retrieval-Augmented Generation for Large Language Models: A Survey , author=. 2024 , eprint=

    2024

  20. [28]

    Advances in neural information processing systems , volume=

    Retrieval-augmented generation for knowledge-intensive nlp tasks , author=. Advances in neural information processing systems , volume=

  21. [29]

    2025 , eprint=

    RAG-Gym: Systematic Optimization of Language Agents for Retrieval-Augmented Generation , author=. 2025 , eprint=

    2025

  22. [30]

    Advances in Neural Information Processing Systems , volume=

    Toolformer: Language models can teach themselves to use tools , author=. Advances in Neural Information Processing Systems , volume=

  23. [31]

    Interleaving Retrieval with Chain-of-Thought Reasoning for Knowledge-Intensive Multi-Step Questions

    Trivedi, Harsh and Balasubramanian, Niranjan and Khot, Tushar and Sabharwal, Ashish. Interleaving Retrieval with Chain-of-Thought Reasoning for Knowledge-Intensive Multi-Step Questions. Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Vo...

    2023 doi

  24. [32]

    ICLR , author =

    ReAct: Synergizing Reasoning and Acting in Language Models , url =. ICLR , author =. 2023 , pages=

    2023

  25. [33]

    Journal of artificial intelligence research , volume=

    Reinforcement learning: A survey , author=. Journal of artificial intelligence research , volume=

  26. [34]

    Advances in neural information processing systems , volume=

    Training language models to follow instructions with human feedback , author=. Advances in neural information processing systems , volume=

  27. [35]

    Advances in neural information processing systems , volume=

    Direct preference optimization: Your language model is secretly a reward model , author=. Advances in neural information processing systems , volume=

  28. [36]

    Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers),

    Arash Ahmadian and Chris Cremer and Matthias Gall. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers),

  29. [37]

    2025 , eprint=

    ReSearch: Learning to Reason with Search for LLMs via Reinforcement Learning , author=. 2025 , eprint=

    2025

  30. [38]

    and Favre, Benoit

    Jacqmin, L \'e o and Rojas Barahona, Lina M. and Favre, Benoit. ``Do you follow me?'': A Survey of Recent Approaches in Dialogue State Tracking. Proceedings of the 23rd Annual Meeting of the Special Interest Group on Discourse and Dialogue. 2022. doi:10.18653/v1/2022.sigdial-1.33

    2022 doi

  31. [39]

    R e2 G : Retrieve, Rerank, Generate

    Glass, Michael and Rossiello, Gaetano and Chowdhury, Md Faisal Mahbub and Naik, Ankita and Cai, Pengshan and Gliozzo, Alfio. R e2 G : Retrieve, Rerank, Generate. Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: ...

    2022 doi

  32. [40]

    LLML ingua: Compressing Prompts for Accelerated Inference of Large Language Models

    Jiang, Huiqiang and Wu, Qianhui and Lin, Chin-Yew and Yang, Yuqing and Qiu, Lili. LLML ingua: Compressing Prompts for Accelerated Inference of Large Language Models. Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing. 2023. doi:10.18653/v1/2...

    2023 doi

  33. [41]

    2025 , eprint=

    Cats Confuse Reasoning LLM: Query Agnostic Adversarial Triggers for Reasoning Models , author=. 2025 , eprint=

    2025

  34. [42]

    arXiv preprint arXiv:1803.05457 , year=

    Think you have solved question answering? try arc, the ai2 reasoning challenge , author=. arXiv preprint arXiv:1803.05457 , year=

    arXiv

  35. [43]

    arXiv preprint arXiv:2009.03300 , year=

    Measuring massive multitask language understanding , author=. arXiv preprint arXiv:2009.03300 , year=

    2009 arXiv

  36. [44]

    Dense Passage Retrieval for Open-Domain Question Answering

    Karpukhin, Vladimir and Oguz, Barlas and Min, Sewon and Lewis, Patrick and Wu, Ledell and Edunov, Sergey and Chen, Danqi and Yih, Wen-tau. Dense Passage Retrieval for Open-Domain Question Answering. Proceedings of the 2020 Conference on Empirical Methods in Natural Language Pr...

    2020 doi

  37. [45]

    Transactions of the Association for Computational Linguistics , volume=

    Natural questions: a benchmark for question answering research , author=. Transactions of the Association for Computational Linguistics , volume=. 2019 , publisher=

    2019

  38. [46]

    T rivia QA : A Large Scale Distantly Supervised Challenge Dataset for Reading Comprehension

    Joshi, Mandar and Choi, Eunsol and Weld, Daniel and Zettlemoyer, Luke. T rivia QA : A Large Scale Distantly Supervised Challenge Dataset for Reading Comprehension. Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). ...

    2017 doi

  39. [47]

    When Not to Trust Language Models: Investigating Effectiveness of Parametric and Non-Parametric Memories

    Mallen, Alex and Asai, Akari and Zhong, Victor and Das, Rajarshi and Khashabi, Daniel and Hajishirzi, Hannaneh. When Not to Trust Language Models: Investigating Effectiveness of Parametric and Non-Parametric Memories. Proceedings of the 61st Annual Meeting of the Association f...

    2023 doi

  40. [48]

    Proceedings of the 2013 conference on empirical methods in natural language processing , pages=

    Semantic parsing on freebase from question-answer pairs , author=. Proceedings of the 2013 conference on empirical methods in natural language processing , pages=

    2013

  41. [49]

    H otpot QA : A Dataset for Diverse, Explainable Multi-hop Question Answering

    Yang, Zhilin and Qi, Peng and Zhang, Saizheng and Bengio, Yoshua and Cohen, William and Salakhutdinov, Ruslan and Manning, Christopher D. H otpot QA : A Dataset for Diverse, Explainable Multi-hop Question Answering. Proceedings of the 2018 Conference on Empirical Methods in Na...

    2018 doi

  42. [50]

    Constructing A Multi-hop QA Dataset for Comprehensive Evaluation of Reasoning Steps

    Ho, Xanh and Duong Nguyen, Anh-Khoa and Sugawara, Saku and Aizawa, Akiko. Constructing A Multi-hop QA Dataset for Comprehensive Evaluation of Reasoning Steps. Proceedings of the 28th International Conference on Computational Linguistics. 2020. doi:10.18653/v1/2020.coling-main.580

    2020 doi

  43. [51]

    Transactions of the Association for Computational Linguistics , volume=

    MuSiQue: Multihop Questions via Single-hop Question Composition , author=. Transactions of the Association for Computational Linguistics , volume=. 2022 , publisher=

    2022

  44. [52]

    Measuring and Narrowing the Compositionality Gap in Language Models

    Press, Ofir and Zhang, Muru and Min, Sewon and Schmidt, Ludwig and Smith, Noah and Lewis, Mike. Measuring and Narrowing the Compositionality Gap in Language Models. Findings of the Association for Computational Linguistics: EMNLP 2023. 2023. doi:10.18653/v1/2023.findings-emnlp.378

    2023 doi

  45. [53]

    Transactions of the Association for Computational Linguistics , volume=

    Did aristotle use a laptop? a question answering benchmark with implicit reasoning strategies , author=. Transactions of the Association for Computational Linguistics , volume=. 2021 , publisher=

    2021

  46. [54]

    arXiv preprint arXiv:2304.04675 , year=

    Multilingual machine translation with large language models: Empirical results and analysis , author=. arXiv preprint arXiv:2304.04675 , year=

    arXiv

  47. [55]

    Transactions of the Association for Computational Linguistics , volume=

    Benchmarking large language models for news summarization , author=. Transactions of the Association for Computational Linguistics , volume=. 2024 , publisher=

    2024

  48. [56]

    Advances in Neural Information Processing Systems , volume=

    Camel: Communicative agents for" mind" exploration of large language model society , author=. Advances in Neural Information Processing Systems , volume=

  49. [57]

    2017 , eprint=

    Proximal Policy Optimization Algorithms , author=. 2017 , eprint=

    2017

  50. [58]

    arXiv preprint arXiv:2501.12948 , year=

    Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning , author=. arXiv preprint arXiv:2501.12948 , year=

    arXiv

  51. [59]

    arXiv preprint arXiv:2312.14925 , volume=

    A survey of reinforcement learning from human feedback , author=. arXiv preprint arXiv:2312.14925 , volume=

  52. [60]

    Advances in neural information processing systems , volume=

    Chain-of-thought prompting elicits reasoning in large language models , author=. Advances in neural information processing systems , volume=

  53. [61]

    arXiv preprint arXiv:2501.05366 , year=

    Search-o1: Agentic search-enhanced large reasoning models , author=. arXiv preprint arXiv:2501.05366 , year=

    arXiv

  54. [62]

    Journal of Machine Learning Research , volume=

    Scaling instruction-finetuned language models , author=. Journal of Machine Learning Research , volume=

  55. [63]

    Advances in Neural Information Processing Systems , volume=

    Simpo: Simple preference optimization with a reference-free reward , author=. Advances in Neural Information Processing Systems , volume=

  56. [64]

    arXiv preprint arXiv:2412.16720 , year=

    Openai o1 system card , author=. arXiv preprint arXiv:2412.16720 , year=

    arXiv

  57. [65]

    2025 , eprint=

    Qwen2.5 Technical Report , author=. 2025 , eprint=

    2025

  58. [66]

    arXiv preprint arXiv:2407.21783 , year=

    The llama 3 herd of models , author=. arXiv preprint arXiv:2407.21783 , year=

    arXiv

  59. [67]

    arXiv preprint arXiv:2303.18223 , volume=

    A survey of large language models , author=. arXiv preprint arXiv:2303.18223 , volume=

    arXiv

  60. [68]

    arXiv preprint arXiv:2401.14196 , year=

    DeepSeek-Coder: When the Large Language Model Meets Programming--The Rise of Code Intelligence , author=. arXiv preprint arXiv:2401.14196 , year=

    arXiv

  61. [69]

    arXiv preprint arXiv:2303.08774 , year=

    Gpt-4 technical report , author=. arXiv preprint arXiv:2303.08774 , year=

    arXiv

  62. [70]

    arXiv preprint arXiv:2403.05530 , year=

    Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context , author=. arXiv preprint arXiv:2403.05530 , year=

    arXiv

  63. [71]

    arXiv preprint arXiv:2212.10403 , year=

    Towards reasoning in large language models: A survey , author=. arXiv preprint arXiv:2212.10403 , year=

    arXiv

  64. [72]

    arXiv preprint arXiv:2309.01219 , year=

    Siren's song in the AI ocean: a survey on hallucination in large language models , author=. arXiv preprint arXiv:2309.01219 , year=

    arXiv

  65. [73]

    arXiv preprint arXiv:2312.10997 , volume=

    Retrieval-augmented generation for large language models: A survey , author=. arXiv preprint arXiv:2312.10997 , volume=

    arXiv

  66. [74]

    The Thirteenth International Conference on Learning Representations , year=

    Long-context llms meet rag: Overcoming challenges for long inputs in rag , author=. The Thirteenth International Conference on Learning Representations , year=

  67. [75]

    ICLR , pages=

    Xi Victoria Lin and Xilun Chen and Mingda Chen and Weijia Shi and Maria Lomeli and Richard James and Pedro Rodriguez and Jacob Kahn and Gergely Szilvasy and Mike Lewis and Luke Zettlemoyer and Wen-tau Yih , title=. ICLR , pages=. 2024 , cdate=

    2024

  68. [76]

    2024 , eprint=

    Text Embeddings by Weakly-Supervised Contrastive Pre-training , author=. 2024 , eprint=

    2024

  69. [77]

    Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing , pages=

    Active retrieval augmented generation , author=. Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing , pages=

    2023

  70. [78]

    International Conference on Learning Representations (ICLR) , year=

    React: Synergizing reasoning and acting in language models , author=. International Conference on Learning Representations (ICLR) , year=

  71. [79]

    arXiv preprint arXiv:2501.17161 , year=

    Sft memorizes, rl generalizes: A comparative study of foundation model post-training , author=. arXiv preprint arXiv:2501.17161 , year=

    arXiv

  72. [80]

    Advances in Neural Information Processing Systems , volume=

    Rankrag: Unifying context ranking with retrieval-augmented generation in llms , author=. Advances in Neural Information Processing Systems , volume=

  73. [81]

    2025 , howpublished=

    7B Model and 8K Examples: Emerging Reasoning with Reinforcement Learning is Both Effective and Efficient , author=. 2025 , howpublished=

    2025

  74. [82]

    Journal of Cognitive Neuroscience , volume=

    Reinforcement learning , author=. Journal of Cognitive Neuroscience , volume=

  75. [83]

    arXiv preprint arXiv:2403.13787 , year=

    Rewardbench: Evaluating reward models for language modeling , author=. arXiv preprint arXiv:2403.13787 , year=

    arXiv

  76. [84]

    Advances in Neural Information Processing Systems , volume=

    Iterative reasoning preference optimization , author=. Advances in Neural Information Processing Systems , volume=

  77. [85]

    Machine learning , volume=

    Simple statistical gradient-following algorithms for connectionist reinforcement learning , author=. Machine learning , volume=. 1992 , publisher=

    1992

  78. [86]

    arXiv preprint arXiv:2501.11651 , year=

    Advancing Language Model Reasoning through Reinforcement Learning and Inference Scaling , author=. arXiv preprint arXiv:2501.11651 , year=

    arXiv

  79. [87]

    arXiv preprint arXiv:2502.14768 , year=

    Logic-RL: Unleashing LLM Reasoning with Rule-Based Reinforcement Learning , author=. arXiv preprint arXiv:2502.14768 , year=

    arXiv

  80. [88]

    arXiv preprint arXiv:2212.09561 , year=

    Large language models are better reasoners with self-verification , author=. arXiv preprint arXiv:2212.09561 , year=

    arXiv

  81. [89]

    arXiv preprint arXiv:2409.12917 , year=

    Training language models to self-correct via reinforcement learning , author=. arXiv preprint arXiv:2409.12917 , year=

    arXiv

  82. [90]

    Frontiers of Computer Science , volume=

    Tool learning with large language models: A survey , author=. Frontiers of Computer Science , volume=. 2025 , publisher=

    2025

  83. [91]

    arXiv preprint arXiv:2410.04343 , year=

    Inference scaling for long-context retrieval augmented generation , author=. arXiv preprint arXiv:2410.04343 , year=

    arXiv

  84. [92]

    arXiv preprint arXiv:2409.19256 , year=

    Hybridflow: A flexible and efficient rlhf framework , author=. arXiv preprint arXiv:2409.19256 , year=

    arXiv

  85. [93]

    arXiv preprint arXiv:2207.06300 , year=

    Re2G: Retrieve, rerank, generate , author=. arXiv preprint arXiv:2207.06300 , year=

    arXiv

  86. [94]

    arXiv preprint arXiv:2412.11919 , year=

    Retrollm: Empowering large language models to retrieve fine-grained evidence within generation , author=. arXiv preprint arXiv:2412.11919 , year=

    arXiv

  87. [95]

    arXiv preprint arXiv:2410.23214 , year=

    Grounding by trying: Llms with reinforcement learning-enhanced retrieval , author=. arXiv preprint arXiv:2410.23214 , year=

    arXiv

  88. [96]

    2024 , publisher=

    Self-rag: Learning to retrieve, generate, and critique through self-reflection , author=. 2024 , publisher=

    2024

  89. [97]

    Large Language Models for Mathematical Reasoning: Progresses and Challenges

    Ahn, Janice and Verma, Rishu and Lou, Renze and Liu, Di and Zhang, Rui and Yin, Wenpeng. Large Language Models for Mathematical Reasoning: Progresses and Challenges. Proceedings of the 18th Conference of the European Chapter of the Association for Computational Linguistics: St...

    2024 doi

  90. [98]

    2025 , eprint=

    Toward a Theory of Agents as Tool-Use Decision-Makers , author=. 2025 , eprint=

    2025

  91. [99]

    2023 , eprint=

    Scaling Relationship on Learning Mathematical Reasoning with Large Language Models , author=. 2023 , eprint=

    2023

  92. [100]

    Advances in Information Retrieval: 47th European Conference on Information Retrieval, ECIR 2025, Lucca, Italy, April 6–10, 2025, Proceedings, Part III , pages =

    Razavi, Amirhossein and Soltangheis, Mina and Arabzadeh, Negar and Salamat, Sara and Zihayat, Morteza and Bagheri, Ebrahim , title =. Advances in Information Retrieval: 47th European Conference on Information Retrieval, ECIR 2025, Lucca, Italy, April 6–10, 2025, Proceedings, P...

    2025 doi

  93. [101]

    Ask Again, Then Fail: Large Language Models' Vacillations in Judgment

    Xie, Qiming and Wang, Zengzhi and Feng, Yi and Xia, Rui. Ask Again, Then Fail: Large Language Models' Vacillations in Judgment. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2024. doi:10.18653/v1/2024.acl-long.577

    2024 doi

  94. [102]

    2025 , eprint=

    Reinforced Internal-External Knowledge Synergistic Reasoning for Efficient Adaptive Search Agent , author=. 2025 , eprint=

    2025

  95. [103]

    Proceedings of the AAAI Conference on Artificial Intelligence , author=

    MemoryBank: Enhancing Large Language Models with Long-Term Memory , volume=. Proceedings of the AAAI Conference on Artificial Intelligence , author=. 2024 , month=. doi:10.1609/aaai.v38i17.29946 , abstractNote=

    2024 doi

  96. [104]

    Proceedings of the Twentieth European Conference on Computer Systems , pages =

    Sheng, Guangming and Zhang, Chi and Ye, Zilingfeng and Wu, Xibin and Zhang, Wang and Zhang, Ru and Peng, Yanghua and Lin, Haibin and Wu, Chuan , title =. Proceedings of the Twentieth European Conference on Computer Systems , pages =. 2025 , isbn =. doi:10.1145/3689031.3696075 ...

    2025

  97. [105]

    Proceedings of the ACM SIGOPS 29th Symposium on Operating Systems Principles , year=

    Efficient Memory Management for Large Language Model Serving with PagedAttention , author=. Proceedings of the ACM SIGOPS 29th Symposium on Operating Systems Principles , year=

  98. [106]

    2025 , eprint=

    Cognitive Kernel-Pro: A Framework for Deep Research Agents and Agent Foundation Models Training , author=. 2025 , eprint=

    2025

  99. [107]

    2025 , eprint=

    Explore to Evolve: Scaling Evolved Aggregation Logic via Proactive Online Exploration for Deep Research Agents , author=. 2025 , eprint=

    2025

  100. [108]

    CoRR , volume =

    Mukai Li and Qingcheng Zeng and Tianqing Fang and Zhenwen Liang and Linfeng Song and Qi Liu and Haitao Mi and Dong Yu , title =. CoRR , volume =. 2026 , url =. doi:10.48550/ARXIV.2602.03412 , eprinttype =. 2602.03412 , timestamp =

    2026 doi arXiv

  101. [109]

    Search-o1: Agentic Search-Enhanced Large Reasoning Models

    Li, Xiaoxi and Dong, Guanting and Jin, Jiajie and Zhang, Yuyao and Zhou, Yujia and Zhu, Yutao and Zhang, Peitian and Dou, Zhicheng. Search-o1: Agentic Search-Enhanced Large Reasoning Models. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing...

    2025 doi

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.