TRAINING-FREE FRAMEWORK

CoreGaze: Core Subgraph-Driven Visual Gaze Diffusion for Training-Free Referring Multimodal Large Language Models (2026.acl-long)

Challenge:	Large language models face unique challenges such as domain-specific terminologies and reasoning over specialized knowledge.
Approach:	They propose a multi-disciplinary collaboration framework that leverages LLM-based agents in a role-playing setting.
Outcome:	The proposed framework excels at mining and harnessing medical expertise within LLMs, as well as extending its reasoning abilities.

Copied to clipboard

Xiaoyang Yi, Jing Chen, Yuru Bao, Jian Zhang

Learning from Emptiness: De-biasing Listwise Rerankers with Content-Agnostic Probability Calibration (2026.acl-short)

Challenge:	Existing methods rely on extensive fine-tuning to mitigate attention distraction, leading to redundant outputs or hallucinations.
Approach:	They propose a training-free framework that simulates human visual gaze diffusion for fine-grained comprehension by combining a sparse semantic graph with a core subgraph with amplified initial influence.
Outcome:	The proposed framework simulates human visual gaze diffusion for fine-grained comprehension.

Copied to clipboard

Hang Lv, Hongchao Gu, Ruiqing Yang, Liangyue Li, Zulong Chen, Defu Lian, Hao Wang, Enhong Chen

SCAIR: Schema-Conditioned Agentic Iterative Reasoning for Enterprise Knowledge Graphs (2026.acl-industry)

Challenge:	Existing methods for listwise reranking exhibit intrinsic position bias . existing methods are constrained by an inherent trade-off between efficiency and flexibility .
Approach:	They propose a training-free framework that mechanically decouples positional bias from ranking decisions.
Outcome:	a training-free framework decouples position bias from ranking decisions . evaluations show it outperforms training-based methods and outperformed expensive methods .

Copied to clipboard

Prateek Chaturvedi, Yuqicheng Zhu, Hongkuan Zhou, Dongzhuoran Zhou, Yunjie He, Steffen Staab, Fei Du, Jie Tang, Evgeny Kharlamov

Structured Data in NLP

Challenge:	Existing agentic approaches for Knowledge Graph-based Retrieval-Augmented Generation fail to generalize to real-world enterprise Knowledge graphs (KGs) dense, schema-driven, and operationally constrained, requiring a training-free framework.
Approach:	They propose a training-free framework that integrates structured planning with controlled iterative reasoning by injecting schema-conditioned structural priors and enforcing schemas during multi-hop reasoning.
Outcome:	The proposed framework significantly improves on a real-world enterprise-oriented benchmark constructed from a Configuration Management DataBase (CMDB).

REAP: Towards Effective Training-Free Chemical Reasoning with Explicit Atomic Priors (2026.findings-acl)

Copied to clipboard

Mingxu Zhang, Dazhong Shen, Qi Zhang, Ying Sun

Do Domain-specific Experts exist in MoE-based LLMs? (2026.findings-acl)

Challenge:	Current approaches to instill explicit priors into LLMs often suffer from an information bottleneck .
Approach:	They propose a training-free framework that equips LLMs with an external knowledge base, enabling them to reason over retrieved chemical priors dynamically.
Outcome:	Experiments show that REAP outperforms current reasoning methods and rivals state-of-the-art training-based models.

Copied to clipboard

Giang Do, Hung Le, Truyen Tran

TIPA: Typologically Informed Parameter Aggregation (2026.findings-eacl)

Challenge:	Existing studies on domain-specific experts in Large Language Models (LLMs) are still lacking.
Approach:	They propose a training-free framework that introduces zero additional inference cost and outperforms well-trained MoE-based LLMs.
Outcome:	The proposed framework outperforms well-trained MoE-based LLMs and strong baselines across target and non-target domains.

Copied to clipboard

Stef Accou, Wessel Poelman

Multilinguality

Challenge:	Massively multilingual language models enable cross-lingual generalization but underperform on low-resource and unseen languages.
Approach:	They propose a typologically informed framework that constructs proxy language adapters by aggregating existing ones, weighted by typological similarity.
Outcome:	The proposed framework outperforms baselines on five NLP tasks and over 230 languages.

Can Calibration of Positional Encodings Enhance Long Context Utilization? (2026.findings-eacl)

Copied to clipboard

Tom Zehle, Matthias Aßenmacher

Image Embedding Sampling Method for Diverse Captioning (2025.emnlp-main)

Challenge:	Large language models suffer from positional biases that reduce effective utilization of long contexts.
Approach:	They propose a training-free framework for calibrating Positional Encodings at inference time.
Outcome:	The proposed framework improves on needle-in-a-haystack and cross-chunk reasoning benchmarks and provides a lightweight method for improving long-context utilization.

Copied to clipboard

Sania Waheed, Na Min An

GrowOVER: How Can LLMs Adapt to Growing Real-World Knowledge? (2024.acl-long)

Challenge:	Currently, large-scale captioning models are less accessible for resource-constrained applications such as mobile devices and assistive technologies.
Approach:	They propose a training-free framework that enhances caption diversity and informativeness by explicitly attending to distinct image regions using a comparably small VLM as the backbone.
Outcome:	The proposed framework achieves comparable performance to larger models on MSCOCO, Flickr30k, and Nocaps test datasets while maintaining strong image-caption relevancy and semantic integrity with the human-annotated captions.

Copied to clipboard

Dayoon Ko, Jinyoung Kim, Hahyeon Choi, Gunhee Kim

Natural Language Interfaces

Challenge:	Existing knowledge-based datasets are outdated due to the rapid evolution of knowledge.
Approach:	They propose a retrieval-interactive language model framework that evaluates and reflects on its answers for further re-retrieval.
Outcome:	The proposed framework performs comparably to or surpasses continuously trained language models.

ThinkPilot: Steering Reasoning Models via Automated Think-prefixes Optimization (2026.findings-eacl)

Copied to clipboard

Sunzhu Li, Zhiyu Lin, Jiale Zhao, Shuling Yang, Chen Wei

BannerAgency: Advertising Banner Design with Multimodal LLM Agents (2025.emnlp-main)

Challenge:	Large Reasoning Models (LRMs) are powerful but still suffer from inefficient and off-target reasoning.
Approach:	They propose a training-free framework that automatically optimizes Large Reasoning Models' reasoning by generating think-prefixes that evolve driven by a taxonomy of reasoning behaviors.
Outcome:	The proposed framework significantly improves accuracy-length trade-off for efficient reasoning, drastically improves safety and improves instruction following.

Copied to clipboard

Heng Wang, Yotaro Shimose, Shingo Takamatsu

Memo-SQL: Structured Decomposition and Experience-Driven Self-Correction for Training-Free NL2SQL (2026.findings-acl)

Challenge:	Advertising banners are an instrumental medium in digital marketing campaigns.
Approach:	They propose a training-free framework for fully automated banner ad design creation that enables frontier multimodal large language models to streamline the production of effective banners with minimal manual effort.
Outcome:	The proposed framework is based on a training-free model that can be used to create fully automated banner ad design creations with minimal manual effort across diverse marketing contexts.

Copied to clipboard

Yang Zerui, Weichuan Wang, Yanwei Xu, Linqi Song, Yudai Matsuda, Wei Han, Bo Bai

Programming Languages in NLP

Challenge:	Existing NL2SQL systems rely on in-context learning with only correct examples . current test-time scaling methods often decompose questions arbitrarily, resulting in poor performance .
Approach:	They propose a structured decomposition and experience-aware self-correction framework for NL2SQL . they build a dynamic memory of successful queries and historical error–fix pairs .
Outcome:	The proposed framework achieves 68.5% execution accuracy on BIRD, setting new state of the art among open, zero-fine-tuning methods.

Look, Compare, Decide: Alleviating Hallucination in Large Vision-Language Models via Multi-View Multi-Path Reasoning (2025.coling-main)

Copied to clipboard

Xiaoye Qu, Jiashuo Sun, Wei Wei, Daizong Liu, Jianfeng Dong, Yu Cheng

VideoEraser: Concept Erasure in Text-to-Video Diffusion Models (2025.emnlp-main)

Challenge:	Large Vision-Language Models (LVLMs) have impressive capabilities in multi-modal context comprehension, but they still suffer from hallucination problems due to inconsistent outputs with the image content.
Approach:	They propose a training-free framework MVP to reduce hallucinations in Large Vision-Language Models . they propose multi-view information-seeking strategy to perceive the comprehensive information in the image .
Outcome:	The proposed framework reduces hallucinations in large vision-language models by combining multi-view multi-path reasoning with multi-vision multi-path reasoning.

Copied to clipboard

Naen Xu, Jinghuai Zhang, Changjiang Li, Zhi Chen, Chunyi Zhou, Qingming Li, Tianyu Du, Shouling Ji

Bridging the Dynamic Perception Gap: Training-Free Draft Chain-of-Thought for Dynamic Multimodal Spatial Reasoning (2025.findings-emnlp)

Challenge:	Experimental results show that VideoEraser outperforms prior methods regarding efficacy, integrity, fidelity, robustness, and generalizability.
Approach:	They propose a training-free framework that prevents T2V diffusion models from generating videos with undesirable concepts even when explicitly prompted with those concepts.
Outcome:	The proposed framework outperforms existing methods in erasure, celebrity erasion, and explicit content erasing tasks.

Copied to clipboard

Siqu Ou, Hongcheng Liu, Pingjie Wang, Yusheng Liao, Chuan Xuan, Yanfeng Wang, Yu Wang

Personalized Text Generation with Contrastive Activation Steering (2025.acl-long)

Challenge:	Existing methods for dynamic spatial reasoning are limited to text or static visual domains .
Approach:	They propose a framework that augments textual reasoning chains with dynamic visual drafts .
Outcome:	The proposed framework outperforms existing methods in dynamic spatial reasoning tasks.

Copied to clipboard

Jinghao Zhang, Yuting Liu, Wenjie Wang, Qiang Liu, Shu Wu, Liang Wang, Tat-Seng Chua

Information Retrieval

Challenge:	Existing approaches to personalized text generation rely on retrieval-augmented generation and parameter-efficient fine-tuning.
Approach:	They propose a training-free framework that disentangles and represents personalized writing style as a vector in LLM’s activation-space.
Outcome:	The proposed framework achieves 8% relative improvement in personalized generation while reducing storage requirements by 1700 over PEFT method.

Confidence-guided Refinement Reasoning for Zero-shot Question Answering (2025.emnlp-main)

Copied to clipboard

Youwon Jang, Woo Suk Choi, Minjoon Jung, Minsu Lee, Byoung-Tak Zhang

BILLY: Steering Large Language Models via Merging Persona Vectors for Creative Generation (2026.eacl-long)

Challenge:	Existing frameworks that generate single-step reasoning do not improve QA reasoning .
Approach:	They propose a framework that strategically constructs and refines sub-questions and their answers (sub-QAs) they argue that sub-QA does not always enhance QA reasoning .
Outcome:	The proposed framework can be integrated with existing QA models and benchmarks.

Copied to clipboard

Tsung-Min Pai, Jui-I Wang, Li-Chun Lu, Shao-Hua Sun, Hung-yi Lee, Kai-Wei Chang

Dual Activation-Weight Sparsity: A Training-Free Framework for Efficient Large Language Model Compression (2026.acl-long)

Challenge:	Multi-LLM systems enhance creativity of large language models by simulating human collective intelligence but suffer from significant drawbacks, such as high computational costs and inference latency.
Approach:	They propose a training-free framework that captures the benefits of multi-LLM collaboration by extracting and blending multiple distinct persona vectors directly in the model’s activation space.
Outcome:	The proposed framework surpasses model prompting and traditional multi-LLM approaches while significantly reducing inference time and computational costs.

Copied to clipboard

Luoyang Sun, Guangyan Li, Cheng Deng, Haifeng Zhang, Jian Zhao, Yongqiang Tang, Wensheng Zhang, Jun Wang

LongHeads: Multi-Head Attention is Secretly a Long Context Processor (2024.findings-emnlp)

Challenge:	Large language models (LLMs) excel at natural language tasks but face deployment bottlenecks due to computational demands.
Approach:	They propose a training-free framework that exploits activation and weight sparsity . they use a three-tier routing strategy that uses magnitude-based pruning .
Outcome:	Experiments on Llama and Mistral models show that DAWS outperforms activation-weight sparsity pruning methods.

Copied to clipboard

Yi Lu, Xin Zhou, Wei He, Jun Zhao, Tao Ji, Tao Gui, Qi Zhang, Xuanjing Huang

Mitigating Cultural Bias in LLMs via Multi-Agent Cultural Debate (2026.findings-acl)

Challenge:	Large language models struggle to process lengthy inputs due to limited length generalization and attention’s quadratic computational demands.
Approach:	They propose a training-free framework that allows each head to attend to important context chunks instead of allowing each head a full sentence .
Outcome:	The proposed framework unlocks multi-head attention's untapped potential by allowing each head to attend to important context chunks instead of the full sentence.

Copied to clipboard

Qian Tan, Lei Jiang, Yuting Zeng, Shuoyang Ding, Xiaohua Xu

FiDeLiS: Faithful Reasoning in Large Language Models for Knowledge Graph Question Answering (2025.findings-acl)

Challenge:	Existing approaches to evaluate large language models fail to address cultural bias in non-Western languages . Chinese prompting shifts bias toward East Asian perspectives rather than eliminating it, authors say .
Approach:	They propose a Chinese–English bilingual benchmark and multi-agent vote frameworks that enable explicit "no bias" judgments.
Outcome:	The proposed framework achieves 57.6% average No Bias Rate on Chinese-English benchmark and 86.0% on Arabic CAMeL benchmark.

Copied to clipboard

Yuan Sui, Yufei He, Nian Liu, Xiaoxin He, Kun Wang, Bryan Hooi

Semantic-Eval : A Semantic Comprehension Evaluation Framework for Large Language Models Generation without Training (2025.acl-long)

Challenge:	Existing retrieval-based or agent-based methods are prone to generating erroneous or hallucinated outputs.
Approach:	They propose a framework to leverage knowledge graphs as external knowledge sources to improve the factuality of LLM responses by anchoring answers to verifiable reasoning steps retrieved from KGs.
Outcome:	The proposed framework improves factuality and interpretability across benchmarks and reduces computational costs.

Copied to clipboard

Shusheng Li, Jiale Li, Yifei Qu, Xinwei Shi, Yanliang Guo, Ziyi He, Yubo Wang, Wenjun Tan

ReKG-MCTS: Reinforcing LLM Reasoning on Knowledge Graphs via Training-Free Monte Carlo Tree Search (2025.findings-acl)

Challenge:	Large language models (LLMs) have emerged as key drivers of progress in the field of natural language processing.
Approach:	They propose a framework that assesses LLM-generated text based on semantic understanding.
Outcome:	The proposed framework surpasses traditional evaluation metrics and lags behind GPT-4.

Copied to clipboard

Xiaozhuang Song, Shufei Zhang, Tianshu Yu

Embracing Imperfection: Simulating Students with Diverse Cognitive Levels Using LLM-based Agents (2025.acl-long)

Challenge:	Existing approaches to combining knowledge graphs with large language models face limitations in path exploration strategies or excessive computational overhead.
Approach:	They propose a training-free framework that synergizes Monte Carlo Tree Search with LLM capabilities to enable dynamic reasoning over KGs.
Outcome:	The proposed framework outperforms existing training-free methods and achieves competitive performance compared to fine-tuned baselines.

Copied to clipboard

Tao Wu, Jingyuan Chen, Wang Lin, Mengze Li, Yumeng Zhu, Ang Li, Kun Kuang, Fei Wu

Challenge:	Large language models (LLMs) are becoming increasingly popular in education, enabling researchers to simulate students' learning patterns and learning patterns.
Approach:	They propose a training-free framework for student simulation that takes into account student cognitive diversity and realism.
Outcome:	The proposed model outperforms baseline models and achieves 100% improvement in simulation accuracy and realism.

A Cognitive Writing Perspective for Constrained Long-Form Text Generation (2025.findings-acl)

Copied to clipboard

Kaiyang Wan, Honglin Mu, Rui Hao, Haoran Luo, Tianle Gu, Xiuying Chen

Text Generation

Challenge:	Large Language Models struggle to generate high-quality long-form text in a single pass . a new framework that trains LLMs to write human-like writing capabilities is needed .
Approach:	They propose a framework that equips large language models with human-like cognitive writing capabilities . they use a planning agent and multiple Generation Agents to generate long-form text in parallel .
Outcome:	CogWriter surpasses GPT-4o by 22% in complex instruction completion accuracy . the framework can generate coherent text in a single pass with fluency that rivals human writers .

VCSearch: Bridging the Gap Between Well-Defined and Ill-Defined Problems in Mathematical Reasoning (2025.emnlp-main)

Copied to clipboard

Shi-Yu Tian, Zhi Zhou, Kun-Yang Yu, Ming Yang, Lin-Han Jia, Lan-Zhe Guo, Yu-Feng Li

Reasoning

Challenge:	Existing studies have improved the performance of Large language models on well-defined mathematical benchmarks, but they often overlook ill-defined problems.
Approach:	They develop a large-scale benchmark that contains over 5,000 ill-defined mathematical problems.
Outcome:	The proposed framework improves the accuracy of identifying unsolvable problems by at least 12% across different LLMs, thus achieving stronger robust mathematical reasoning ability.

Efficient Test-Time Scaling via Temporal Reasoning Aggregation (2026.findings-acl)

Copied to clipboard

Jiakun Li, Xingwei He, Kefan Li, Hongzheng Chai, Hongyue Yu, Yuan Yuan

AICA-Bench: Holistically Examining the Capabilities of VLMs in Affective Image Content Analysis (2026.findings-acl)

Challenge:	Existing dynamic early-exit methods rely on single-step confidence signals . existing approaches are unreliable for detecting reasoning convergence in multi-step settings .
Approach:	They propose a training-free framework for efficient test-time scaling that determines when to terminate reasoning based on temporal aggregation of multi-step evidence rather than instantaneous signals.
Outcome:	Experiments show that TRACE reduces reasoning token usage by 25% on average while maintaining accuracy within 1–2% of full-length reasoning.

Copied to clipboard

Dong She, Xianrong Yao, Liqun Chen, Jinghe Yu, Yang Gao, Zhanpeng Jin

Sentiment Analysis

Challenge:	Recent studies have focused on factual correctness, semantic grounding, visual reasoning, or multimodal large language models.
Approach:	They propose a benchmark to assess AICA, which integrates perception, reasoning, and generation into a unified framework.
Outcome:	The proposed framework corrects intensity errors and significantly enhances descriptive depth.

From Query to Logic: Ontology-Driven Multi-Hop Reasoning in LLMs (2026.findings-acl)

Copied to clipboard

Haonan Bian, Yutao Qi, Rui Yang, Yuanxi Che, Jiaqian Wang, Heming Xia, Ranran Zhen

Contextual Experience Replay for Self-Improvement of Language Agents (2025.acl-long)

Challenge:	Large Language Models (LLMs) exhibit limitations in complex multi-hop question answering tasks that necessitate non-linear, structured reasoning.
Approach:	They propose an ontology-driven reasoning and chain framework that combines LLMs’ generative capabilities with the structural benefits of knowledge graphs.
Outcome:	Extensive experiments across a diverse set of models and standard MQA benchmarks demonstrate that the proposed framework achieves competitive performance while producing more interpretable reasoning chains.

Copied to clipboard

Yitao Liu, Chenglei Si, Karthik R Narasimhan, Shunyu Yao

Dynamic Steering With Episodic Memory For Large Language Models (2025.findings-acl)

Challenge:	Existing large language model (LLM) agents fail in complex tasks without any environment-specific experiences.
Approach:	They propose a framework that accumulates and synthesizes past experiences into a dynamic memory buffer to enable efficient self-improvement for language agents in their context window.
Outcome:	The proposed framework improves performance on WebArena and VisualWebAren . it surpasses tree search method with fewer token costs and achieves state-of-the-art performance of 31.9%.

Copied to clipboard

Van Dai Do, Quan Hung Tran, Svetha Venkatesh, Hung Le

CoG: Controllable Graph Reasoning via Relational Blueprints and Failure-Aware Refinement over Knowledge Graphs (2026.acl-long)

Challenge:	Existing activation steering methods apply a single sentence-level steering vector uniformly across all tokens, ignoring LLMs’ token-wise, auto-regressive nature.
Approach:	They propose a framework that aligns LLMs to given demonstrations by steering at the token level conditioned on the input query.
Outcome:	The proposed framework surpasses baselines across safety, style transfer, and role-playing tasks, demonstrating improved alignment as demonstration scales.

Copied to clipboard

Yuanxiang Liu, Songze Li, Xiaoke Guo, Zhaoyan Gong, Qifei Zhang, Huajun Chen, Wen Zhang

Structured Data in NLP

Challenge:	Existing approaches to large language models often exhibit cognitive rigidity, causing reasoning stagnation.
Approach:	They propose a training-free framework that mimics the interplay between intuition and deliberation.
Outcome:	The proposed framework outperforms state-of-the-art approaches on three benchmarks.

MMAPG: A Training-Free Framework for Multimodal Multi-hop Question Answering via Adaptive Planning Graphs (2025.emnlp-main)

Copied to clipboard

Yiheng Hu, Xiaoyang Wang, Qing Liu, Xiwei Xu, Qian Fu, Wenjie Zhang, Liming Zhu

HydraRAG: Structured Cross-Source Enhanced Large Language Model Reasoning (2025.emnlp-main)

Challenge:	Existing multimodal question answering models rely on sequential retrieval and reasoning, but this single-path paradigm makes them vulnerable to errors due to misleading intermediate steps.
Approach:	They propose a multimodal multi-hop question answering framework guided by an Adaptive Planning Graph . they propose modality-specific strategies that dynamically adapt to distinct data types .
Outcome:	The proposed framework outperforms existing models that rely on training.

Copied to clipboard

Xingyu Tan, Xiaoyang Wang, Qing Liu, Xiwei Xu, Xin Yuan, Liming Zhu, Wenjie Zhang

Removal of Hallucination on Hallucination: Debate-Augmented RAG (2025.acl-long)

Challenge:	Current RAG system retrieves evidence from knowledge graphs and text documents but has limitations in multi-hop reasoning, multi-entity questions, and source verification.
Approach:	They propose a training-free framework that unifies graph topology, document semantics, and source reliability to support deep, faithful reasoning in large language models.
Outcome:	The proposed framework outperforms the current hybrid model-based model-driven system by 20.3% and 30.1% on seven benchmark datasets.

Copied to clipboard

Wentao Hu, Wengyu Zhang, Yiyang Jiang, Chen Jason Zhang, Xiaoyong Wei, Li Qing

Reasoning

Challenge:	erroneous or biased retrieval can mislead generation, compounding hallucinations.
Approach:	They propose a framework that integrates multi-agent debates into retrieval and generation stages to improve retrieval reliability.
Outcome:	The proposed framework improves retrieval reliability, reduces hallucinations and significantly improves overall factual accuracy.

Adaptive Planning for Multi-Attribute Controllable Summarization with Monte Carlo Tree Search (2026.acl-long)

Copied to clipboard

Sangwon Ryu, Heejin Do, Yunsu Kim, Gary Lee, Jungseul Ok

Information Extraction & Text Mining

Challenge:	Controllable summarization is a form of outputs that tailors summaries to user-specified attributes.
Approach:	They propose an adaptive planning framework that reframes the task as planning the order of sequential attribute control with a customized Monte Carlo Tree Search.
Outcome:	The proposed framework surpasses LLM-based self-planning models and fine-tuned baselines in multi-attribute controllable summarization.

ExploraCoder: Advancing Code Generation for Multiple Unseen APIs via Planning and Chained Exploration (2025.acl-long)

Copied to clipboard

Yunkun Wang, Yue Zhang, Zhen Qin, Chen Zhi, Binhua Li, Fei Huang, Yongbin Li, Shuiguang Deng

Programming Languages in NLP

Challenge:	Large language models face intrinsic limitations in coding with unseen APIs in training corpora.
Approach:	They propose a training-free framework that empowers LLMs to invoke multiple unseen APIs in code solution by planning a complex problem into several API invocation subtasks and experimenting with correct API usage at intermediate steps.
Outcome:	The proposed framework significantly improves performance for models lacking prior API knowledge, achieving 11.99% over retrieval-based approaches and 17.28% over pretraining-based methods in pass@10.

Beyond Uniform SVD: Dual-Level Optimization across Columns and Modules for LLM Compression (2026.findings-acl)

Copied to clipboard

Lin Xv, Xian Gao, Ting Liu, Yuzhuo fu

Less is More: Improving LLM Reasoning with Minimal Test-Time Intervention (2026.acl-long)

Challenge:	Existing methods for low-rank decomposition overlook decomposing errors and suboptimal approximation.
Approach:	They propose a low-rank decomposition framework that integrates low-level optimization at column and module levels.
Outcome:	The proposed framework outperforms state-of-the-art methods and baselines in SVD and pruning.

Copied to clipboard

Zhen Yang, Mingyang Zhang, Feng Chen, Ganggui Ding, Liang Hou, Xin Tao, Ying-Cong Chen

PlaM: Training-Free Plateau-Guided Model Merging for Better Visual Grounding in MLLMs (2026.findings-acl)

Challenge:	Recent advances in large language models (LLMs) have focused on test-time scaling to improve reasoning quality but at the cost of efficiency.
Approach:	They propose a training-free framework that enhances reasoning accuracy and stability with minimal overhead.
Outcome:	The proposed framework yields consistent gains across general, coding, and STEM tasks while remaining highly efficient.

Copied to clipboard

Zijing Wang, YongKang Liu, Mingyang Wang, Ercong Nie, Deyuan Chen, Zhengjie Zhao, Shi Feng, Daling Wang, Xiaocui Yang, Yifei Zhang, Hinrich Schuetze

Training-Free Adaptive Speculative Decoding via Linguistic Priors (2026.findings-acl)

Challenge:	Multimodal instruction fine-tuning degrades textual reasoning capability, undermining multimodal performance.
Approach:	They propose a plateau-guided model merging method that selectively injects base language model parameters into MLLMs to mitigate this degradation.
Outcome:	The proposed framework reduces multimodal instruction fine-tuning degradation by incorporating a plateau-guided model merging method into MLLMs.

Copied to clipboard

Jingyi Wang, Jiaqi Huang, Zunnan Xu, Jun Zhou, Kehong Yuan, Xiang Qian

Syntactic Text Processing

Challenge:	Speculative decoding (SPD) is a promising technique to accelerate Large Language Models (LLMs). current approaches neglect the inherent heterogeneity of natural language and fail to distinguish between semantically-rich content and structurally-predictable syntax.
Approach:	They propose a training-free framework that leverages linguistic priors to enable adaptive drafting and verification.
Outcome:	The proposed framework significantly accelerates inference without additional training.

Stop Hardening Everything: A Training-Free Neuron-Level Defense for Neural Ranking Models (2026.acl-long)

Copied to clipboard

Yu-An Liu, Ruqing Zhang, Hongru Song, Jiafeng Guo, Yixing Fan, Xueqi Cheng

Robustness in NLP

Challenge:	Existing defenses for neural ranking models are data-centric and require retraining and adversarial data generation.
Approach:	They propose a model-centric defense that addresses vulnerability at its architectural source without costly retraining or adversarial data generation.
Outcome:	The proposed approach outperforms state-of-the-art models on MS MARCO and TREC 19 while maintaining strong performance on clean data.

Beyond Query Bias: Candidate-Aware Iterative Refinement for Zero-Shot Composed Image Retrieval (2026.findings-acl)

Copied to clipboard

Nan Sun, Jing Tang, Lei Sun, Rui Chen, Yuxing Lu, Xiangxiang Chu, Hefei Ling, Yujun Cai

Adaptive Spatial and Temporal Redundancy Optimization for Efficient Reasoning in Large Language Models (2026.acl-long)

Challenge:	Existing methods to retrieve target images suffer from inherent cognitive bias due to unknown candidate distribution.
Approach:	They propose a training-free framework that reframes ZS-CIR as a self-correcting process . they propose to use retrieved results as feedback to perceive the candidate distribution .
Outcome:	Experiments on public benchmarks show that CoRR outperforms other SOTA methods.

Copied to clipboard

Tianle Chen, Pengyu Cheng, Qiyuan Zhu, Jiacheng Wang, Bei Liu, Hao Gu, Ruijie Shen, Xiaofeng Hou, Sirui Han, Jiacheng Liu

Waterfall: Scalable Framework for Robust Text Watermarking and Provenance for LLMs (2024.emnlp-main)

Challenge:	Existing research to improve CoT efficiency falls into three categories, each with distinct limitations.
Approach:	They propose a training-free framework that addresses both dimensions of CoT reasoning by applying a progressive precision reduction strategy coupled with an entropy-based confidence mechanism for adaptive termination.
Outcome:	Empirical results show that the proposed framework achieves 11.3 efficiency gain without compromising accuracy.

Copied to clipboard

Gregory Kang Ruey Lau, Xinyuan Niu, Hieu Dao, Jiangwei Chen, Chuan-Sheng Foo, Bryan Kian Hsiang Low

Robustness in NLP

Challenge:	Existing text watermarking methods are not robust enough against paraphrasing attacks . existing methods lack robustness to paraphrases and are not scalable to millions of users .
Approach:	They propose a training-free framework for robust and scalable text watermarking . they propose to use large language models as paraphrasers and a combination of techniques .
Outcome:	The proposed framework improves scalability, verifiability and computational efficiency compared to existing methods.

RACC: Regret-Aware Confidence Calibration for Consistent Masked Discrete Diffusion Decoding (2026.findings-acl)

Copied to clipboard

Qinglin Zeng, Jusheng Zhang, Jing Yang, Ningyuan Liu, Keze Wang

Safety Arithmetic: A Framework for Test-time Safety Alignment of Language Models by Steering Parameters and Activations (2024.emnlp-main)

Challenge:	Masked Discrete Diffusion Models (MDMs) enable parallel generation via iterative refinement, but their current decoding paradigms are static and myopic.
Approach:	They propose a Regret-Aware Confidence Calibration framework that aligns decoding decisions with the model’s latent self-correction capabilities.
Outcome:	The proposed framework aligns decoding decisions with model’s latent self-correction capabilities.

Copied to clipboard

Rima Hazra, Sayan Layek, Somnath Banerjee, Soujanya Poria

MEXA: Towards General Multimodal Reasoning with Dynamic Multi-Expert Aggregation (2025.findings-emnlp)

Challenge:	Current alignment methods struggle with dynamic user intentions and complex objectives, making models vulnerable to harmful content.
Approach:	They propose a training-free framework that enhances LLM safety across different scenarios.
Outcome:	The proposed framework significantly improves safety measures, reduces over-safety, and maintains model utility, outperforming existing methods in ensuring safe content generation.

Copied to clipboard

Shoubin Yu, Yue Zhang, Ziyang Wang, Jaehong Yoon, Mohit Bansal

RedundancyLens: Revealing and Exploiting Visual Token Processing Redundancy for Efficient Decoder-Only MLLMs (2025.findings-acl)

Challenge:	MEXA is a training-free framework that performs modality- and task-aware aggregation of multiple expert models to enable effective multimodal reasoning across diverse domains.
Approach:	MEXA is a training-free framework that performs modality- and task-aware aggregation of multiple expert models.
Outcome:	MEXA performs modality- and task-aware aggregation of multiple expert models . it generates interpretable textual reasoning outputs and reasons over them using a Large Reasoning Model (LRM) MEX A consistently delivers performance improvements over strong multimodal benchmarks .

Copied to clipboard

Hongliang Li, Jiaxin Zhang, Wenhui Liao, Dezhi Peng, Kai Ding, Lianwen Jin

Stop When Enough: Adaptive Early-Stopping for Chain-of-Thought Reasoning (2026.acl-long)

Challenge:	Current decoder-only architectures achieve higher performance but lower efficiency . cross-attention-based architectures skip visual token computations .
Approach:	They propose a training-free framework for analyzing trained MLLMs to investigate redundancy . they propose 'probe-activated Dynamic FFN and Hollow Attention' algorithms for visual token reductions and a layer ranking algorithm for inference acceleration.
Outcome:	The proposed framework achieves comparable performance to or better than state-of-the-art methods while remaining compatible with them.

Copied to clipboard

Renliang Sun, Wei Cheng, Dawei Li, Haifeng Chen, Wei Wang

Low-Resource NLP

Challenge:	Chain-of-Thought reasoning has driven recent gains of large language models (LLMs) on reasoning-intensive tasks by externalizing intermediate steps.
Approach:	They propose a training-free framework that adaptively determines when to stop reasoning to mitigate overthinking.
Outcome:	The proposed framework reduces token usage by 20-55% while maintaining or improving accuracy compared to standard CoT prompting.

Sparse Activation Editing for Reliable Instruction Following in Narratives (2025.emnlp-main)

Copied to clipboard

Runcong Zhao, Chengyu Cao, Qinglin Zhu, Xiucheng Ly, Shun Shao, Lin Gui, Ruifeng Xu, Yulan He

Challenge:	Existing benchmarks fail to capture the challenges of instruction following in complex narrative contexts.
Approach:	They propose a training-free framework that identifies and edits instruction-relevant neurons using only natural language instructions without requiring labelled data.
Outcome:	The proposed framework improves instruction following by identifying and editing instruction-relevant neurons using only natural language instructions, without requiring labelled data.

Reverse Prompt Engineering: A Zero-Shot, Genetic Algorithm Approach to Language Model Inversion (2025.emnlp-main)

Copied to clipboard

Hanqing Li, Diego Klabjan

DiMo-GUI: Advancing Test-time Scaling in GUI Grounding via Modality-Aware Visual Reasoning (2025.emnlp-main)

Challenge:	Existing methods rely on a large number of outputs for training and inference, and they can produce garbled text.
Approach:	They propose a training-free framework that reconstructs prompts using only a limited number of text outputs from a language model.
Outcome:	The proposed framework generates high-quality prompt recovery and more semantically and functionally aligned with the originals than current state-of-the-art methods.

Copied to clipboard

Hang Wu, Hongkai Chen, Yujun Cai, Chang Liu, Qingwen Ye, Ming-Hsuan Yang, Yiwei Wang

LLM×MapReduce: Simplified Long-Sequence Processing using Large Language Models (2025.acl-long)

Challenge:	DiMo-GUI is a training-free framework for GUI grounding that splits input into textual elements and iconic elements, allowing the model to reason over each modality independently using general-purpose vision-language models.
Approach:	They propose a training-free framework for GUI grounding that leverages two core strategies: dynamic visual grounding and modality-aware optimization.
Outcome:	The proposed framework splits the input into textual elements and iconic elements, allowing the model to reason over each modality independently using general-purpose vision-language models.

Copied to clipboard

Zihan Zhou, Chong Li, Xinyi Chen, Shuo Wang, Yu Chao, Zhili Li, Haoyu Wang, Qi Shi, Zhixing Tan, Xu Han, Xiaodong Shi, Zhiyuan Liu, Maosong Sun

Calibrated Speculative Decoding: Frequency-Guided Candidate Selection for Efficient Inference (2026.acl-long)

Challenge:	Existing studies have focused on extending the context length of large language models (LLMs) due to their quadratic computational complexity and a lack of high-quality long training examples, most LLMs are trained with a limited window size.
Approach:	They propose a training-free framework that enables large language models to effectively process long texts using a divide-and-conquer strategy for comprehensive document understanding.
Outcome:	The proposed framework outperforms open-source and commercial long-context LLMs and is compatible with several models.

Copied to clipboard

null Zhouxuwen, Fangxin Liu, Chao Wang, Xiao Zheng, Hao Zheng, Min He, Li Jiang, Haibing Guan

Parallel Context-of-Experts Decoding for Retrieval Augmented Generation (2026.findings-acl)

Challenge:	Speculative decoding (SD) is a powerful and efficient way to accelerate autoregressive generation.
Approach:	They propose a training-free framework that recovers valid tokens discarded by standard verification . they use online correction memory and Semantic Consistency Gating to analyze rejections .
Outcome:	The proposed framework outperforms existing methods and achieves peak throughput speedup of 2.33x.

Copied to clipboard

Giulio Corallo, Paolo Papotti

Information Retrieval

Challenge:	Retrieval Augmented Generation relies on concatenating documents into a long context prompt, causing prefill bottlenecks.
Approach:	They propose a training-free framework that shifts evidence aggregation from attention to decoding . they treat retrieved documents as isolated "experts", synchronizing their predictions via a retrieval-aware extension of context-awful decoding.
Outcome:	The proposed framework shifts evidence aggregation from attention to decoding . it treats retrieved documents as isolated experts, synchronizing their predictions .

Not All Citations Are Equal:Entropy-Guided Citation Selection for Noise-Resistant Medical LLM (2026.findings-acl)

Copied to clipboard

Minyu Gao, Hanlin Xiao, Ruoyu Wang, Shuai Yang, YeXuan Zhang, Xin Wu, Xingyu Liu

River-LLM: Large Language Model Seamless Exit Based on KV Share (2026.acl-long)

Challenge:	Large language models have demonstrated extensive potential in medical applications . however, their practical deployment in healthcare faces significant challenges .
Approach:	They propose a training-free multi-turn reasoning framework and a post-training methodology that provides external knowledge support for large language models.
Outcome:	The proposed framework elicits internal thought, external thought, and fusion thought, with an entropy-based reward that encourages selective citation of beneficial external knowledge while penalizing noisy citations.

Copied to clipboard

Yingtao Shen, An Zou

LoRA on the Go: Instance-level Dynamic LoRA Selection and Merging (2026.acl-long)

Challenge:	Existing methods to reduce latency and speed up early exits are costly and impose significant cost and energy consumption.
Approach:	They propose a lightweight KV-Shared Exit River framework that allows the backbone’s missing KV cache to be naturally generated and preserved during the exit process.
Outcome:	The proposed framework achieves 1.71 to 2.16 speedup while maintaining high generation quality.

Copied to clipboard

Seungeon Lee, Soumi Das, Manish Gupta, Krishna P. Gummadi