pith. sign in

arxiv: 2606.01240 · v1 · pith:CJMYCE3Nnew · submitted 2026-05-31 · 💻 cs.CL

Efficient RAG with Intent-Aware Retrieval and Semantics-Preserving Chunking

Fachrina Dewi Puspitasari , Chaoning Zhang , Jiaquan Zhang , Zhicheng Wang , Hafiz Shakeel Ahmad Awan , Rizwan Qureshi , Jewon Lee , Tae-Ho Kim
show 1 more author
Yang Yang
This is my paper

Pith reviewed 2026-06-28 17:36 UTC · model grok-4.3

classification 💻 cs.CL
keywords retrieval-augmented generationRAGintent-aware retrievalsemantics-preserving chunkingmulti-hop reasoningsmall language modelsHotPotQAFEVER
0
0 comments X

The pith

InSemRAG improves multi-hop RAG accuracy by dynamically weighting retrieval to query intent and repairing semantic damage in chunks.

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

The paper proposes InSemRAG to fix information insufficiency in standard RAG systems caused by retrieval that ignores query intent and chunks that fragment evidence. It introduces an iterative retrieve-and-check process with an intention-aware retriever that adjusts channel weights on the fly and semantics-preserving chunking that detects and fixes damaged evidence. This leads to measurable gains on tasks that need multi-hop reasoning. The system keeps overall competitiveness while cutting latency by using small language models instead of larger ones.

Core claim

InSemRAG addresses intent-agnostic retrieval and information fragmentation in conventional RAG via an iterative retrieve-and-check mechanism. The intention-aware retriever implements dynamic hybrid retrieval that adaptively weights the retrieval channels based on the query intent. The semantics-preserving chunking performs detection and reparation to the damaged evidence chunks to preserve semantic integrity. Extensive experiments show a 2.65-point F1 improvement on HotPotQA, a 1.5-point accuracy increase on FEVER, and competitive results to Multi-Hop RAG at 4.32 times lower latency when using small language models.

What carries the argument

The iterative retrieve-and-check mechanism with intention-aware retriever (IAR) for dynamic weighting and semantics-preserving chunking (SPC) for evidence repair.

If this is right

  • Gains of 2.65 F1 points on HotPotQA for multi-hop tasks.
  • 1.5-point accuracy rise on FEVER for evidence-sensitive tasks.
  • Competitive benchmark results across datasets with 4.32 times lower latency than Multi-Hop RAG.
  • Overall competitiveness to recent state-of-the-art RAG methods.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The efficiency gains from small language models could make complex RAG viable on resource-limited hardware.
  • Chunk repair may reduce downstream generation errors on queries not covered in the current benchmarks.
  • Adaptive weighting of retrieval channels might extend to other hybrid search systems outside RAG.

Load-bearing premise

The iterative retrieve-and-check mechanism with IAR and SPC can accurately detect query intent, adapt retrieval weights, and repair semantic damage in chunks without introducing new errors or requiring extensive human tuning of the SLM components.

What would settle it

A direct comparison showing zero or negative F1 change on HotPotQA or higher latency than Multi-Hop RAG after applying the full InSemRAG pipeline would falsify the central performance claims.

Figures

Figures reproduced from arXiv: 2606.01240 by Chaoning Zhang, Fachrina Dewi Puspitasari, Hafiz Shakeel Ahmad Awan, Jewon Lee, Jiaquan Zhang, Rizwan Qureshi, Tae-Ho Kim, Yang Yang, Zhicheng Wang.

Figure 1
Figure 1. Figure 1: Limitations on the conventional RAG system that relies on a fixed retrieval channel and ignores semanti [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Illustration of InSemRAG framework. In a high-level view, we first enrich the input query from multiple [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Distribution of channel weighting across dif [PITH_FULL_IMAGE:figures/full_fig_p014_3.png] view at source ↗
read the original abstract

The demand for powerful instruction following and reasoning capability of large language models (LLMs) has promoted rapid development of retrieval-augmented generation (RAG). The RAG system assists LLM generation by retrieving chunks of query-fit supplementary knowledge from an external database. Conventional RAG systems, however, suffer from information insufficiency due to two factors, which are intent-agnostic retrieval and information fragmentation. Our work proposes a RAG framework, termed InSemRAG, that addresses these challenges via an iterative retrieve-and-check mechanism with two supporting modules, an intention-aware retriever (IAR) and semantics-preserving chunking (SPC). IAR implements a dynamic hybrid retrieval method that adaptively weights the retrieval channels based on the query intent, while SPC performs detection and reparation to the damaged evidence chunks to preserve the semantic integrity. To alleviate the computational latency brought by our iterative mechanism, we leverage small language models (SLMs). Extensive experiments across several benchmark datasets consistently demonstrate the competitiveness of our method against recent state-of-the-art RAG mechanisms. Particularly, our method achieves significant gains on multi-hop and evidence-sensitive tasks, with a 2.65-point improvement in F1 on HotPotQA and a 1.5-point increase in accuracy on FEVER. Our method also achieves competitive performance to Multi-Hop RAG with 4.32$\times$ lower latency with the utilization of SLM.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The paper proposes InSemRAG, a RAG framework using an iterative retrieve-and-check mechanism with an intent-aware retriever (IAR) that dynamically weights retrieval channels and semantics-preserving chunking (SPC) that detects and repairs damaged evidence chunks. It leverages SLMs to mitigate latency and reports competitive or superior results on multi-hop and evidence-sensitive benchmarks, including a 2.65-point F1 gain on HotPotQA and 1.5-point accuracy gain on FEVER, while achieving 4.32× lower latency than Multi-Hop RAG.

Significance. If the performance claims hold under rigorous controls, the work offers a practical advance in RAG by addressing intent-agnostic retrieval and chunk fragmentation while maintaining efficiency via SLMs. The emphasis on iterative intent detection and semantic repair could influence designs for complex reasoning tasks, provided the gains prove robust.

major comments (2)
  1. [Experiments] Experiments section: the reported gains (e.g., 2.65 F1 on HotPotQA, 1.5 accuracy on FEVER) are presented without error bars, statistical significance tests, or ablation studies isolating the iterative retrieve-and-check component, IAR weighting, and SPC repair; this directly limits verification of the central claim that IAR and SPC drive the improvements.
  2. [Methods] Methods section: concrete details on the SLM prompting or fine-tuning for intent classification, retrieval weighting, and chunk repair are insufficient to assess whether the iterative mechanism avoids introducing new errors or requires extensive tuning, which is load-bearing for the weakest assumption that the mechanism functions reliably.
minor comments (1)
  1. [Abstract] Abstract and §3: the latency comparison (4.32× lower) should specify the exact baseline implementation and hardware to allow direct replication.

Simulated Author's Rebuttal

2 responses · 0 unresolved

Thank you for the constructive feedback. We address each major comment below and will incorporate revisions to strengthen the experimental reporting and methodological transparency.

read point-by-point responses

  1. Referee: [Experiments] Experiments section: the reported gains (e.g., 2.65 F1 on HotPotQA, 1.5 accuracy on FEVER) are presented without error bars, statistical significance tests, or ablation studies isolating the iterative retrieve-and-check component, IAR weighting, and SPC repair; this directly limits verification of the central claim that IAR and SPC drive the improvements.

    Authors: We agree that the absence of error bars, statistical tests, and component-specific ablations limits the ability to verify the source of gains. In the revised version we will report means and standard deviations over multiple random seeds, include paired statistical significance tests against baselines, and add ablation tables that remove or isolate the iterative retrieve-and-check loop, the IAR weighting mechanism, and the SPC repair step individually. revision: yes

  2. Referee: [Methods] Methods section: concrete details on the SLM prompting or fine-tuning for intent classification, retrieval weighting, and chunk repair are insufficient to assess whether the iterative mechanism avoids introducing new errors or requires extensive tuning, which is load-bearing for the weakest assumption that the mechanism functions reliably.

    Authors: We will expand the Methods section with the exact prompt templates used by the SLM for intent classification, dynamic channel weighting, and chunk repair, together with any fine-tuning details, temperature settings, and stopping criteria for the iterative loop. We will also add a short discussion of observed error rates across iterations and the safeguards employed to limit error accumulation. revision: yes

Circularity Check

0 steps flagged

No significant circularity in derivation chain

full rationale

The paper proposes InSemRAG with IAR and SPC modules for RAG, evaluated via empirical gains on external benchmarks (HotPotQA F1, FEVER accuracy, latency vs Multi-Hop RAG). No equations, fitted parameters, or self-citations are presented that reduce claimed performance to inputs by construction. The iterative mechanism and SLM usage are described as design choices, not derived from the target metrics. Central claims rest on benchmark results rather than self-referential fitting or renamed ansatzes, making the derivation self-contained against external data.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

No mathematical derivations or explicit axioms appear in the abstract; the work is empirical and relies on standard benchmark assumptions plus unstated details of intent classification and chunk repair logic.

pith-pipeline@v0.9.1-grok · 5821 in / 1073 out tokens · 21243 ms · 2026-06-28T17:36:37.879759+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

79 extracted references · 35 canonical work pages · 14 internal anchors

  1. [1]

    Aho and Jeffrey D

    Alfred V. Aho and Jeffrey D. Ullman , title =. 1972

    1972

  2. [2]

    Publications Manual , year = "1983", publisher =

    1983

  3. [3]

    Chandra and Dexter C

    Ashok K. Chandra and Dexter C. Kozen and Larry J. Stockmeyer , year = "1981", title =. doi:10.1145/322234.322243

    work page doi:10.1145/322234.322243 1981

  4. [4]

    Scalable training of

    Andrew, Galen and Gao, Jianfeng , booktitle=. Scalable training of

  5. [5]

    Dan Gusfield , title =. 1997

    1997

  6. [6]

    Tetreault , title =

    Mohammad Sadegh Rasooli and Joel R. Tetreault , title =. Computing Research Repository , volume =. 2015 , url =

    2015

  7. [7]

    A Framework for Learning Predictive Structures from Multiple Tasks and Unlabeled Data , Volume =

    Ando, Rie Kubota and Zhang, Tong , Issn =. A Framework for Learning Predictive Structures from Multiple Tasks and Unlabeled Data , Volume =. Journal of Machine Learning Research , Month = dec, Numpages =

  8. [8]

    Advances in neural information processing systems , volume=

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

  9. [9]

    Retrieval-Augmented Generation for Large Language Models: A Survey

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

    work page internal anchor Pith review Pith/arXiv arXiv

  10. [10]

    Advances in Neural Information Processing Systems , volume=

    Augmenting language models with long-term memory , author=. Advances in Neural Information Processing Systems , volume=

  11. [11]

    arXiv preprint arXiv:2501.06713 , year=

    Minirag: Towards extremely simple retrieval-augmented generation , author=. arXiv preprint arXiv:2501.06713 , year=

    work page arXiv

  12. [12]

    arXiv preprint arXiv:2410.20753 , year=

    Plan* rag: Efficient test-time planning for retrieval augmented generation , author=. arXiv preprint arXiv:2410.20753 , year=

    work page arXiv

  13. [13]

    LightRAG: Simple and Fast Retrieval-Augmented Generation

    Lightrag: Simple and fast retrieval-augmented generation , author=. arXiv preprint arXiv:2410.05779 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  14. [14]

    arXiv preprint arXiv:2404.00610 , year=

    Rq-rag: Learning to refine queries for retrieval augmented generation , author=. arXiv preprint arXiv:2404.00610 , year=

    work page arXiv

  15. [15]

    arXiv preprint arXiv:2411.13154 , year=

    Dmqr-rag: Diverse multi-query rewriting for rag , author=. arXiv preprint arXiv:2411.13154 , year=

    work page arXiv

  16. [16]

    URL https://arxiv

    Precise zero-shot dense retrieval without relevance labels, 2022 , author=. URL https://arxiv. org/abs/2212.10496 , year=

    work page arXiv 2022

  17. [17]

    arXiv preprint arXiv:2310.06117 , year=

    Take a step back: Evoking reasoning via abstraction in large language models , author=. arXiv preprint arXiv:2310.06117 , year=

    work page arXiv

  18. [18]

    arXiv preprint arXiv:2402.03367 , year=

    Rag-fusion: a new take on retrieval-augmented generation , author=. arXiv preprint arXiv:2402.03367 , year=

    work page arXiv

  19. [19]

    arXiv preprint arXiv:2503.23013 , year=

    DAT: Dynamic alpha tuning for hybrid retrieval in retrieval-augmented generation , author=. arXiv preprint arXiv:2503.23013 , year=

    work page arXiv

  20. [20]

    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=

  21. [21]

    arXiv preprint arXiv:2409.04701 , year=

    Late chunking: contextual chunk embeddings using long-context embedding models , author=. arXiv preprint arXiv:2409.04701 , year=

    work page arXiv

  22. [22]

    Findings of the Association for Computational Linguistics: NAACL 2025 , pages=

    Is semantic chunking worth the computational cost? , author=. Findings of the Association for Computational Linguistics: NAACL 2025 , pages=

    2025

  23. [23]

    The twelveth international conference on learning representations , year=

    Self-rag: Learning to retrieve, generate, and critique through self-reflection , author=. The twelveth international conference on learning representations , year=

  24. [24]

    Corrective retrieval augmented generation , author=

  25. [25]

    Agentic Retrieval-Augmented Generation: A Survey on Agentic RAG

    Agentic retrieval-augmented generation: A survey on agentic rag , author=. arXiv preprint arXiv:2501.09136 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  26. [26]

    Proceedings of the 16th conference of the european chapter of the association for computational linguistics: main volume , pages=

    Leveraging passage retrieval with generative models for open domain question answering , author=. Proceedings of the 16th conference of the european chapter of the association for computational linguistics: main volume , pages=

  27. [27]

    The eleventh international conference on learning representations , year=

    React: Synergizing reasoning and acting in language models , author=. The eleventh international conference on learning representations , year=

  28. [28]

    Proceedings of the 61st annual meeting of the association for computational linguistics (volume 1: long papers) , pages=

    Interleaving retrieval with chain-of-thought reasoning for knowledge-intensive multi-step questions , author=. Proceedings of the 61st annual meeting of the association for computational linguistics (volume 1: long papers) , pages=

  29. [29]

    MultiHop-RAG: Benchmarking Retrieval-Augmented Generation for Multi-Hop Queries

    Multihop-rag: Benchmarking retrieval-augmented generation for multi-hop queries , author=. arXiv preprint arXiv:2401.15391 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  30. [30]

    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

  31. [31]

    TriviaQA: A Large Scale Distantly Supervised Challenge Dataset for Reading Comprehension

    Triviaqa: A large scale distantly supervised challenge dataset for reading comprehension , author=. arXiv preprint arXiv:1705.03551 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  32. [32]

    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

  33. [33]

    Proceedings of the 2018 conference on empirical methods in natural language processing , pages=

    HotpotQA: A dataset for diverse, explainable multi-hop question answering , author=. Proceedings of the 2018 conference on empirical methods in natural language processing , pages=

    2018

  34. [34]

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

    Constructing a multi-hop qa dataset for comprehensive evaluation of reasoning steps , author=. arXiv preprint arXiv:2011.01060 , year=

    work page internal anchor Pith review Pith/arXiv arXiv 2011

  35. [35]

    ELI5: Long Form Question Answering

    ELI5: Long form question answering , author=. arXiv preprint arXiv:1907.09190 , year=

    work page internal anchor Pith review Pith/arXiv arXiv 1907

  36. [36]

    FEVER: a large-scale dataset for Fact Extraction and VERification

    FEVER: a large-scale dataset for fact extraction and VERification , author=. arXiv preprint arXiv:1803.05355 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  37. [37]

    Text summarization branches out , pages=

    Rouge: A package for automatic evaluation of summaries , author=. Text summarization branches out , pages=

  38. [38]

    ACM Transactions on Intelligent Systems and Technology , volume=

    A comprehensive survey of small language models in the era of large language models: Techniques, enhancements, applications, collaboration with llms, and trustworthiness , author=. ACM Transactions on Intelligent Systems and Technology , volume=. 2025 , publisher=

    2025

  39. [39]

    arXiv preprint arXiv:2410.09037 , year=

    Mentor-kd: Making small language models better multi-step reasoners , author=. arXiv preprint arXiv:2410.09037 , year=

    work page arXiv

  40. [40]

    The Twelfth International Conference on Learning Representations , year=

    Ra-dit: Retrieval-augmented dual instruction tuning , author=. The Twelfth International Conference on Learning Representations , year=

  41. [41]

    Proceedings of the 2024 10th International Conference on Communication and Information Processing , pages=

    Evaluating Sparse and Dense Retrieval in Retrieval-Augmented Generation Systems: A Study , author=. Proceedings of the 2024 10th International Conference on Communication and Information Processing , pages=

    2024

  42. [42]

    International conference on machine learning , pages=

    Improving language models by retrieving from trillions of tokens , author=. International conference on machine learning , pages=. 2022 , organization=

    2022

  43. [43]

    Atlas: Few-shot Learning with Retrieval Augmented Language Models

    Few-shot learning with retrieval augmented language models , author=. arXiv preprint arXiv:2208.03299 , volume=

    work page internal anchor Pith review Pith/arXiv arXiv

  44. [44]

    Transactions of the Association for Computational Linguistics , volume=

    In-context retrieval-augmented language models , author=. Transactions of the Association for Computational Linguistics , volume=. 2023 , publisher=

    2023

  45. [45]

    Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval , pages=

    Old ir methods meet rag , author=. Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval , pages=

  46. [46]

    arXiv preprint arXiv:2401.07883 , year=

    The chronicles of rag: The retriever, the chunk and the generator , author=. arXiv preprint arXiv:2401.07883 , year=

    work page arXiv

  47. [47]

    2025 10th International Conference on Intelligent Computing and Signal Processing (ICSP) , pages=

    SLM-based Hybrid Retrieval for Resource Constrained Retrieval-Augmented Generation on Open Super-Large Crawled Data , author=. 2025 10th International Conference on Intelligent Computing and Signal Processing (ICSP) , pages=. 2025 , organization=

    2025

  48. [48]

    Findings of the Association for Computational Linguistics ACL 2024 , pages=

    Retrieval-augmented retrieval: Large language models are strong zero-shot retriever , author=. Findings of the Association for Computational Linguistics ACL 2024 , pages=

    2024

  49. [49]

    Foundations and Trends

    The probabilistic relevance framework: BM25 and beyond , author=. Foundations and Trends. 2009 , publisher=

    2009

  50. [50]

    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=

  51. [51]

    2025 IEEE 41st International Conference on Data Engineering Workshops (ICDEW) , pages=

    Structured Retrieval-Augmented Generation for Multi-Entity Question Answering over Heterogeneous Sources , author=. 2025 IEEE 41st International Conference on Data Engineering Workshops (ICDEW) , pages=. 2025 , organization=

    2025

  52. [52]

    arXiv preprint arXiv:2602.13035 , year =

    Look Inward to Explore Outward: Learning Temperature Policy from LLM Internal States via Hierarchical RL , author =. arXiv preprint arXiv:2602.13035 , year =

    work page arXiv

  53. [53]

    arXiv preprint arXiv:2512.06690 , year =

    Think-While-Generating: On-the-Fly Reasoning for Personalized Long-Form Generation , author =. arXiv preprint arXiv:2512.06690 , year =

    work page arXiv

  54. [54]

    TDA-RC: Task-Driven Alignment for Knowledge-Based Reasoning Chains in Large Language Models

    TDA-RC: Task-Driven Alignment for Knowledge-Based Reasoning Chains in Large Language Models , author =. arXiv preprint arXiv:2604.04942 , year =

    work page internal anchor Pith review Pith/arXiv arXiv

  55. [55]

    The Fourteenth International Conference on Learning Representations , year =

    Learning Global Hypothesis Space for Enhancing Synergistic Reasoning Chain , author =. The Fourteenth International Conference on Learning Representations , year =

  56. [56]

    The Fourteenth International Conference on Learning Representations , year =

    Text summarization via global structure awareness , author =. The Fourteenth International Conference on Learning Representations , year =

  57. [57]

    arXiv preprint arXiv:2603.12933 , year =

    Efficient and Interpretable Multi-Agent LLM Routing via Ant Colony Optimization , author =. arXiv preprint arXiv:2603.12933 , year =

    work page arXiv

  58. [58]

    arXiv preprint arXiv:2603.16060 , year=

    ARISE: Agent Reasoning with Intrinsic Skill Evolution in Hierarchical Reinforcement Learning , author =. arXiv preprint arXiv:2603.16060 , year =

    work page arXiv

  59. [59]

    arXiv preprint arXiv:2603.13134 , year =

    When Right Meets Wrong: Bilateral Context Conditioning with Reward-Confidence Correction for GRPO , author =. arXiv preprint arXiv:2603.13134 , year =

    work page arXiv

  60. [60]

    Inspo: Unlocking intrinsic self-reflection for llm preference opti- mization.arXiv preprint arXiv:2512.23126, 2025

    Inspo: Unlocking intrinsic self-reflection for llm preference optimization , author =. arXiv preprint arXiv:2512.23126 , year =

    work page arXiv

  61. [61]

    arXiv preprint arXiv:2508.19639 , year=

    FakeSV-VLM: Taming VLM for Detecting Fake Short-Video News via Progressive Mixture-Of-Experts Adapter , author=. arXiv preprint arXiv:2508.19639 , year=

    work page arXiv

  62. [62]

    StreamMeCo: Long-Term Agent Memory Compression for Efficient Streaming Video Understanding

    StreamMeCo: Long-Term Agent Memory Compression for Efficient Streaming Video Understanding , author=. arXiv preprint arXiv:2604.09000 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  63. [63]

    Lightweight LLM Agent Memory with Small Language Models

    Lightweight LLM Agent Memory with Small Language Models , author=. arXiv preprint arXiv:2604.07798 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  64. [64]

    and Li, Haizhou , journal=

    Zhang, Malu and Wang, Jiadong and Wu, Jibin and Belatreche, Ammar and Amornpaisannon, Burin and Zhang, Zhixuan and Miriyala, Venkata Pavan Kumar and Qu, Hong and Chua, Yansong and Carlson, Trevor E. and Li, Haizhou , journal=. Rectified Linear Postsynaptic Potential Function for Backpropagation in Deep Spiking Neural Networks , year=

  65. [65]

    A Highly Effective and Robust Membrane Potential-Driven Supervised Learning Method for Spiking Neurons , year=

    Zhang, Malu and Qu, Hong and Belatreche, Ammar and Chen, Yi and Yi, Zhang , journal=. A Highly Effective and Robust Membrane Potential-Driven Supervised Learning Method for Spiking Neurons , year=

  66. [66]

    Toward Building Human-Like Sequential Memory Using Brain-Inspired Spiking Neural Models , year=

    Zhang, Malu and Luo, Xiaoling and Wu, Jibin and Belatreche, Ammar and Cai, Siqi and Yang, Yang and Li, Haizhou , journal=. Toward Building Human-Like Sequential Memory Using Brain-Inspired Spiking Neural Models , year=

  67. [67]

    Spike-Driven Lightweight Large Language Model With Evolutionary Computation , year=

    Zhang, Malu and Wei, Wenjie and Zhou, Zijian and Liu, Wanlong and Zhang, Jie and Belatreche, Ammar and Yang, Yang , journal=. Spike-Driven Lightweight Large Language Model With Evolutionary Computation , year=

  68. [68]

    Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) , pages=

    What is overlap knowledge in event argument extraction? APE: A cross-datasets transfer learning model for EAE , author=. Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) , pages=

  69. [69]

    Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) , pages=

    Beyond the granularity: Multi-perspective dialogue collaborative selection for dialogue state tracking , author=. Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) , pages=

  70. [70]

    Dual slot selector via local reliability verification for dialogue state tracking , author=. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers) , pages=

  71. [71]

    Proceedings of the AAAI conference on artificial intelligence , volume=

    Learning to imagine: distillation-based interactive context exploitation for dialogue state tracking , author=. Proceedings of the AAAI conference on artificial intelligence , volume=

  72. [72]

    Findings of the Association for Computational Linguistics: ACL 2025 , pages=

    HASH-RAG: bridging deep hashing with retriever for efficient, fine retrieval and augmented generation , author=. Findings of the Association for Computational Linguistics: ACL 2025 , pages=

    2025

  73. [73]

    Findings of the Association for Computational Linguistics: ACL 2025 , pages=

    Accelerating adaptive retrieval augmented generation via instruction-driven representation reduction of retrieval overlaps , author=. Findings of the Association for Computational Linguistics: ACL 2025 , pages=

    2025

  74. [74]

    Experience Transfer for Multimodal LLM Agents in Minecraft Game

    Experience Transfer for Multimodal LLM Agents in Minecraft Game , author=. arXiv preprint arXiv:2604.05533 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  75. [75]

    Transforming External Knowledge into Triplets for Enhanced Retrieval in RAG of LLMs

    Transforming External Knowledge into Triplets for Enhanced Retrieval in RAG of LLMs , author=. arXiv preprint arXiv:2604.12610 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  76. [76]

    arXiv preprint arXiv:2602.00135 , year=

    LLaVA-FA: Learning Fourier Approximation for Compressing Large Multimodal Models , author=. arXiv preprint arXiv:2602.00135 , year=

    work page arXiv

  77. [77]

    TechRxiv , volume =

    Pengcheng Zheng and Chaoning Zhang and Mingzheng Cui and Guo Chen and Qigan Sun and Jiaxin Huang and Jiaquan Zhang and Tae-Ho Kim and Caiyan Qin and Yazhou Ren and Guoqing Wang and Yang Yang , title =. TechRxiv , volume =. 2026 , doi =

    2026

  78. [78]

    arXiv preprint arXiv:2112.01488 , year=

    ColBERTv2: Effective and Efficient Retrieval via Lightweight Late Interaction , author=. arXiv preprint arXiv:2112.01488 , year=

    work page arXiv

  79. [79]

    Journal of Machine Learning Research , volume=

    Atlas: Few-shot Learning with Retrieval Augmented Language Models , author =. Journal of Machine Learning Research , volume=