EXPLORATION

QueryExplorer: An Interactive Query Generation Assistant for Search and Exploration (2024.naacl-demo)

Challenge:	Large language models exhibit highly homogeneous, repetitive responses, resulting in inefficient exploration.
Approach:	They propose a method that constructs semantically consistent yet distributionally distinct prior contents to different responses and decouple the one-to-many mapping.
Outcome:	The proposed method improves absolute performance by 5.3% and increases generation diversity by 198.3% on average while significantly enhancing output diversity and test-time scaling.

Copied to clipboard

Kaustubh Dhole, Shivam Bajaj, Ramraj Chandradevan, Eugene Agichtein

NLP-KG: A System for Exploratory Search of Scientific Literature in Natural Language Processing (2024.acl-demos)

Challenge:	Formulating effective search queries can be a daunting task for users when they lack expertise in a specific domain or are not proficient in the language of the content.
Approach:	QueryExplorer is an interactive query generation, reformulation, and retrieval interface with support for Hug-gingFace generation models and PyTerrier’sretrieval pipelines and datasets.
Outcome:	QueryExplorer is an interactive query generation, reformulation, and retrieval interface with support for Hug-gingFace generation models and PyTerrier’sretrieval pipelines and datasets, and extensivelogging of human feedback.

Copied to clipboard

Tim Schopf, Florian Matthes

PROM: Pivoted and Regulated Optimization for Multilingual Instruction Learning (2025.naacl-short)

Challenge:	Existing systems for scientific literature search are typically tailored to keyword-based lookup searches, limiting possibilities for exploration.
Approach:	They propose a feature-rich system that supports the exploration of research literature in unfamiliar natural language processing fields.
Outcome:	The proposed system supports the exploration of research literature in unfamiliar natural language processing fields.

Copied to clipboard

Jaeseong Lee, Seung-won Hwang, Hojin Lee, Yunju Bak, Changmin Lee

Multilinguality

Challenge:	Existing solutions to large language models (LLMs) are English-centric, hindering their application to 6500+ existing languages.
Approach:	They propose to append English tuning data with its translated pair to solve this problem . they identify English as an internal pivot language and propose to regulate between them .
Outcome:	The proposed model is able to generalize on multiple benchmarks across different languages.

EXPLORER: Exploration-guided Reasoning for Textual Reinforcement Learning (2024.eacl-long)

Copied to clipboard

Kinjal Basu, Keerthiram Murugesan, Subhajit Chaudhury, Murray Campbell, Kartik Talamadupula, Tim Klinger

CHARD: Clinical Health-Aware Reasoning Across Dimensions for Text Generation Models (2023.eacl-main)

Challenge:	Text-based games (TBGs) combine natural language understanding with reasoning.
Approach:	They propose an exploration-guided reasoning agent for textual reinforcement learning that integrates natural language with reasoning.
Outcome:	The proposed agent outperforms baseline agents on TWG and TWC games.

Copied to clipboard

Steven Y. Feng, Vivek Khetan, Bogdan Sacaleanu, Anatole Gershman, Eduard Hovy

Small Agents, Big Gains: Journey-Aware and Critic-Guided Simulation for Long-Horizon Shopping Dialogues (2026.acl-industry)

Challenge:	Existing studies show that pretrained language models can act as knowledge bases and reason like humans.
Approach:	They propose to use pretrained language models to generate free-flow textual explanations about 52 health conditions across three clinical dimensions.
Outcome:	The proposed model can generate concise and readable text, but can be improved on medical accuracy and QA.

Copied to clipboard

Qing Ping, Changyou Chen, Binxuan Huang

MedDecXtract: A Clinician-Support System for Extracting, Visualizing, and Annotating Medical Decisions in Clinical Narratives (2025.acl-demo)

Challenge:	e-commerce assistants must support inspiration, comparison, and tool-grounded fact-checking . lack of data-coverage and verification problem hampers efficient, deployable models . eaa: "training trajectories must cover diverse user workflows with high fidelity"
Approach:	They propose a system that synthesizes diverse, faithful, and policy-aligned shopping trajectories . a small model can significantly outperform same-size baselines and surpass a large-model baseline .
Outcome:	The proposed model outperforms existing models and surpasses large models with 8 higher inference throughput.

Copied to clipboard

Mohamed Elgaar, Hadi Amiri, Mitra Mohtarami, Leo Anthony Celi

Challenge:	Clinical notes contain important information about medical decisions embedded within unstructured text.
Approach:	They propose an open-source interactive system that automatically extracts medical decisions from clinical text.
Outcome:	The open-source system extracts and visualizes medical decisions from clinical text.

Scheduled Sampling for Transformers (P19-2)

Copied to clipboard

Tsvetomila Mihaylova, André F. T. Martins

Exploring Attention Attractors in Large Language Models (2026.acl-long)

Challenge:	Existing studies show that scheduled sampling can be applied to recurrent neural networks to avoid exposure bias.
Approach:	They propose to use teacher forced embeddings and model predictions to avoid exposure bias in sequence-to-sequence generation.
Outcome:	The proposed technique achieves performance close to a teacher-forcing baseline on two language pairs and is promising for future research.

Copied to clipboard

Ziheng Wang, Zihao Yue, Wenxuan Wang, Qin Jin

Valet: Rule-Based Information Extraction for Rapid Deployment (2022.lrec-1)

Challenge:	Existing studies have suggested that attention attractors function as "summary tokens" while others speculate that tokens with weaker semantics attract high attention, they act as attention sinks that offload excessive attention.
Approach:	They examine attention attractors, tokens that draw significantly high attention, in large language models.
Outcome:	The proposed models are able to capture long-range dependencies within a given context.

Copied to clipboard

Dayne Freitag, John Cadigan, Robert Sasseen, Paul Kalmar

End-to-end Deep Reinforcement Learning Based Coreference Resolution (P19-1)

Challenge:	a number of machine learning models can be trained to perform sentence-level information extraction at accuracies ranging from strong to adequate.
Approach:	They propose a Python framework for rule-based information extraction that allows for complex matching.
Outcome:	The proposed framework can be used to perform rule-based information extraction on examples.

Copied to clipboard

Hongliang Fei, Xu Li, Dingcheng Li, Ping Li

Policy Gradient as a Proxy for Dynamic Oracles in Constituency Parsing (P18-2)

Challenge:	Recent neural network models for coreference resolution are usually trained with heuristic loss functions that are computed over a sequence of local decisions.
Approach:	They propose an end-to-end reinforcement learning based coreference resolution model to directly optimize coreference evaluation metrics.
Outcome:	The proposed model achieves new state-of-the-art performance on the English OntoNotes v5.0 benchmark.

Copied to clipboard

Daniel Fried, Dan Klein

Demonstrating ViviDoc: Generating Interactive Documents through Human-Agent Collaboration (2026.acl-demo)

Challenge:	Existing models for constituency parsing are transition based, decomposing production of each parse tree into a sequence of action decisions.
Approach:	They propose to use a policy gradient method to reduce exposure bias by allowing exploration during training.
Outcome:	The proposed method outperforms static oracle likelihood training on four constituency parsers in three languages on four parser-agnostic languages.

Copied to clipboard

Yinghao Tang, Yupeng Xie, Yingchaojie Feng, Tingfeng Lan, Wei Chen

Challenge:	Recent LLM-based agents can automate content creation, but naively applying them yields uncontrollable and unverifiable outputs.
Approach:	They propose a human-agent collaborative system that generates interactive educational documents from a single topic input.
Outcome:	The proposed system generates documents comparable in quality to human-authored ones.

AWARE: Agentic Knowledge Warehousing for Contextual Intelligence (2026.findings-acl)

Copied to clipboard

Hongjin Qian, Siqi Bao, Zhao Cao, Zheng Liu

Better Simultaneous Translation with Monotonic Knowledge Distillation (2023.acl-long)

Challenge:	Large language models excel in information seeking tasks, but their knowledge is limited in coverage and timeliness.
Approach:	They propose an agentic knowledge warehousing framework that transforms unstructured data into minimal, task-conditioned knowledge representations consumable by LLMs.
Outcome:	Experiments on GAIA, WebWalker, and BrowseComp-Plus show improvements over baselines.

Copied to clipboard

Shushu Wang, Jing Wu, Kai Fan, Wei Luo, Jun Xiao, Zhongqiang Huang

A Self-verified Method for Exploring Simile Knowledge from Pre-trained Language Models (2024.lrec-main)

Challenge:	Existing methods to train offline MT models require generating target tokens before source sentence is fully consumed.
Approach:	They propose a method that leverages traditional translation models as teachers to generate monotonic yet accurate reference translations for sequence-level knowledge distillation.
Outcome:	The proposed approach improves on strong baselines and on a monotonic version of the WMT15 De-En test set.

Copied to clipboard

Longxuan Ma, Changxin Ke, Shuhan Zhou, Churui Sun, Wei-Nan Zhang, Ting Liu

Learning Retrieval Augmentation for Personalized Dialogue Generation (2023.emnlp-main)

Challenge:	Pre-trained language models (PLMs) have succeeded in natural language processing because they learn generic knowledge from a large corpus.
Approach:	They propose a method that allows pre-trained language models to explore simile knowledge from PLMs . they enhance PLM models with a multi-level simile recognition task that evaluates similes aplenty .
Outcome:	The proposed method can explore more accurate simile knowledge for PLMs.

Copied to clipboard

Qiushi Huang, Shuai Fu, Xubo Liu, Wenwu Wang, Tom Ko, Yu Zhang, Lilian Tang

To Code or not to Code? Adaptive Tool Integration for Math Language Models via Expectation-Maximization (2025.findings-acl)

Challenge:	Personalized dialogue generation is a popular approach for conversational AI applications . however, persona profiles may not provide comprehensive descriptions of the persona .
Approach:	They propose a method that leverages persona profiles and dialogue context to generate personalized dialogues by leveraging personas and persona profile.
Outcome:	The proposed method outperforms baselines on the CONVAI2 dataset . it is expected to generate personalized dialogues based on persona profiles and dialogue context .

Copied to clipboard

Haozhe Wang, Long Li, Chao Qu, Weidi Xu, Fengming Zhu, Wei Chu, Fangzhen Lin

Chunks as Arms: Multi-Armed Bandit-Guided Sampling for Long-Context LLM Preference Optimization (2026.acl-long)

Challenge:	Existing tools that integrate chain-of-thought reasoning and code execution lack metacognitive awareness to integrate tools.
Approach:	They propose a framework that synergizes structured exploration with off-policy RL optimization to create a cycle between metacognitive tool-use decisions and evolving capabilities.
Outcome:	The proposed framework improves over 11% on MATH500 and 9.4% on AIME without o1-like CoT.

Copied to clipboard

Shaohua Duan, Pengcheng Huang, Xinze Li, Zhenghao Liu, Xiaoyuan Yi, Yukun Yan, Shuo Wang, Yu Gu, Ge Yu, Maosong Sun

Cross-Lingual Question Answering over Knowledge Base as Reading Comprehension (2023.findings-eacl)

Challenge:	Recent studies have explored fine-tuning Large Language Models with synthetic data to enhance their long-context capabilities.
Approach:	They propose a framework that leverages a Multi-Armed Bandit rollout strategy to identify the most informative chunks from the given long context for sampling high-quality and diverse responses.
Outcome:	The proposed framework achieves 4% improvement on long-context reasoning benchmarks on Llama and Qwen.

Copied to clipboard

Chen Zhang, Yuxuan Lai, Yansong Feng, Xingyu Shen, Haowei Du, Dongyan Zhao

Can LLMs Facilitate Interpretation of Pre-trained Language Models? (2023.emnlp-main)

Challenge:	Existing high-quality xMRC datasets can be further utilized to fine-tune our model.
Approach:	They propose a cross-lingual question answering over knowledge base approach that converts KB subgraphs into passages to narrow the gap between KB schemas and questions.
Outcome:	The proposed approach outperforms baselines and achieves strong few-shot and zero-shot performance on two xKBQA datasets in 12 languages.

Copied to clipboard

Basel Mousi, Nadir Durrani, Fahim Dalvi

VoiSeR: A New Benchmark for Voice-Based Search Refinement (2021.eacl-main)

Challenge:	Existing methods to uncover knowledge encoded within pre-trained language models are limited in terms of scalability and scope of interpretation.
Approach:	They propose to use a large language model, ChatGPT, as an annotation tool . they demonstrate that ChatGPt produces accurate and semantically richer annotations .
Outcome:	The proposed method produces accurate and semantically richer annotations compared to human annotations.

Copied to clipboard

Simone Filice, Giuseppe Castellucci, Marcus Collins, Eugene Agichtein, Oleg Rokhlenko

Multimodality

Challenge:	a new study shows that voice-based search systems are challenging to support in the context of the user intent of voice searches . support for voice-driven search, exploration, and refinement is a fundamental aspect of voice assistants .
Approach:	They propose to use crowdsourcing to collect voice-based search refinements . they use 10,000 search refinement utterances to annotate a search intent .
Outcome:	The proposed dataset shows that voice-based search refinements can support most common tasks . the study shows that the proposed dataset can support research in conversational query understanding .

From Bottom to Top: Extending the Potential of Parameter Efficient Fine-Tuning (2024.emnlp-main)

Copied to clipboard

Jihao Gu, Zelin Wang, Yibo Zhang, Ziji Zhang, Ping Gong

Nonparametric Forest-Structured Neural Topic Modeling (2022.coling-1)

Challenge:	Existing methods to fine-tune large language models primarily focus on the interaction between different layers, ignoring the fact that different layers store different information.
Approach:	They propose a Parameter Efficient Fine-Tuning method which freeze pre-trained parameters and fine-tunes only a few task-specific parameters.
Outcome:	The proposed methods reduce parameter count to nearly half by omitting fine-tuning in the middle layers.

Copied to clipboard

Zhihong Zhang, Xuewen Zhang, Yanghui Rao

The Promises and Pitfalls of Using Language Models to Measure Instruction Quality in Education (2024.naacl-long)

Challenge:	Existing hierarchical neural topic models can only extract topics at the same level.
Approach:	They propose to use self-attention mechanism to capture parent-child topic relationships and build a sparse directed acyclic graph to form a topic forest.
Outcome:	The proposed model outperforms baseline models on topic hierarchical rationality and affinity.

Copied to clipboard

Paiheng Xu, Jing Liu, Nathan Jones, Julie Cohen, Wei Ai

Event Transition Planning for Open-ended Text Generation (2022.findings-acl)

Challenge:	Existing methods to assess instruction quality require trained raters to observe classrooms based on established criteria.
Approach:	They propose to use Natural Language Processing techniques to assess multiple high-inference instructional practices in in-person K-12 classrooms and simulated performance tasks for pre-service teachers.
Outcome:	The proposed method is able to assess multiple high-inference instructional practices in two educational settings: in-person K-12 classrooms and simulated performance tasks for pre-service teachers.

Copied to clipboard

Qintong Li, Piji Li, Wei Bi, Zhaochun Ren, Yuxuan Lai, Lingpeng Kong

Text Generation

Challenge:	Open-ended text generation tasks require models to generate coherent continuation given limited preceding context.
Approach:	They propose a novel two-stage method which explicitly arranges ensuing events in open-ended text generation tasks.
Outcome:	The proposed method improves coherence and diversity of open-ended text generation tasks.

SD-E2: Semantic Exploration for Reasoning Under Token Budgets (2026.findings-eacl)

Copied to clipboard

Kshitij Mishra, Nils Lukas, Salem Lahlou

Representation Learning

Challenge:	Small language models struggle with complex reasoning because exploration is expensive under tight compute budgets.
Approach:	They propose a framework that makes exploration explicit by optimizing semantic diversity in generated reasoning trajectories.
Outcome:	The proposed framework surpasses Qwen2.5-3B-Instruct and strong GRPO baselines on GSM8K and improves on the harder AIME benchmark to 13.28% vs. base 6.74%.

What Makes a Good Reasoning Chain? Uncovering Structural Patterns in Long Chain-of-Thought Reasoning (2025.emnlp-main)

Copied to clipboard

Gangwei Jiang, Yahui Liu, Zhaoyi Li, Wei Bi, Fuzheng Zhang, Linqi Song, Ying Wei, Defu Lian

Playing Text-Adventure Games with Graph-Based Deep Reinforcement Learning (N19-1)

Challenge:	Recent advances in reasoning with large language models have popularized Long Chain-of-Thought (LCoT) a framework that converts sequential LCoTs into hierarchical tree structures enables deeper structural analysis of LLM reasoning.
Approach:	They propose a framework that converts sequential LCoTs into hierarchical tree structures and enables deeper structural analysis of LLM reasoning.
Outcome:	The proposed framework can be used to analyze LLM reasoning in a variety of tasks and models.

Copied to clipboard

Prithviraj Ammanabrolu, Mark Riedl

Structured Data in NLP

Challenge:	Text adventure games provide a platform for exploring reinforcement learning in combinatorial action space, such as natural language.
Approach:	They propose a deep reinforcement learning architecture that represents the game state as a knowledge graph which is learned during exploration.
Outcome:	The proposed architecture can learn a control policy faster than baseline alternatives.

Large Language Models Struggle to Describe the Haystack without Human Help: A Social Science-Inspired Evaluation of Topic Models (2025.acl-long)

Copied to clipboard

Zongxia Li, Lorena Calvo-Bartolomé, Alexander Miserlis Hoyle, Paiheng Xu, Daniel Kofi Stephens, Juan Francisco Fung, Alden Dima, Jordan Lee Boyd-Graber

ALPACA AGAINST VICUNA: Using LLMs to Uncover Memorization of LLMs (2025.naacl-long)

Challenge:	a common use of NLP is to facilitate the understanding of large document collections.
Approach:	They propose to use large language models to replace probabilistic topic models in real-world applications.
Outcome:	The proposed model generates more human-readable topics and shows higher average win probabilities than traditional models for data exploration.

Copied to clipboard

Aly M. Kassem, Omar Mahmoud, Niloofar Mireshghallah, Hyunwoo Kim, Yulia Tsvetkov, Yejin Choi, Sherif Saad, Santu Rana

LLM-Microscope: Uncovering the Hidden Role of Punctuation in Context Memory of Transformers (2025.findings-naacl)

Challenge:	Existing studies have shown that pre-trained LLMs emit training data up to 150 more often than in regular operation.
Approach:	They propose a black-box prompt optimization method where an attacker LLM agent uncovers higher levels of memorization in a victim agent .
Outcome:	The proposed method shows 23.7% more overlap with training data compared to state-of-the-art baselines.

Copied to clipboard

Anton Razzhigaev, Matvey Mikhalchuk, Temurbek Rahmatullaev, Elizaveta Goncharova, Polina Druzhinina, Ivan Oseledets, Andrey Kuznetsov

Learning to Ask Conversational Questions by Optimizing Levenshtein Distance (2021.acl-long)

Challenge:	Large Language Models (LLMs) encode and store contextual information, but internal mechanisms are opaque.
Approach:	They propose a toolkit that assesses token-level nonlinearity, evaluates contextual memory, visualizes intermediate layer contributions and measures intrinsic dimensionality of representations.
Outcome:	The proposed framework assesses token-level nonlinearity, evaluates contextual memory, visualizes intermediate layer contributions, and measures the intrinsic dimensionality of representations.

Copied to clipboard

Zhongkun Liu, Pengjie Ren, Zhumin Chen, Zhaochun Ren, Maarten de Rijke, Ming Zhou

Natural Language Interfaces

Challenge:	Existing methods for estimating maximum likelihood are limited by easily learned tokens . Existing systems that generate questions based on dialogue context are limited in their ability to learn tokens.
Approach:	They propose a framework that optimizes the minimum Levenshtein distance through explicit editing actions.
Outcome:	The proposed framework outperforms state-of-the-art methods on two benchmark datasets and generalizes well on unseen data.

infoVerse: A Universal Framework for Dataset Characterization with Multidimensional Meta-information (2023.acl-long)

Copied to clipboard

Jaehyung Kim, Yekyung Kim, Karin de Langis, Jinwoo Shin, Dongyeop Kang

MiniLongBench: The Low-cost Long Context Understanding Benchmark for Large Language Models (2025.acl-long)

Challenge:	Several methods for characterizing datasets based on model-driven meta-information have been developed, but the relationship and complementary effects of these methods have received less attention.
Approach:	They propose a framework that captures multidimensional characteristics of datasets by incorporating various model-driven meta-information.
Outcome:	The proposed framework outperforms baselines in three real-world applications and can be used in a variety of real-time problems.

Copied to clipboard

Zhongzhan Huang, Guoming Ling, Shanshan Zhong, Hefeng Wu, Liang Lin

Learning Efficient Dialogue Policy from Demonstrations through Shaping (2020.acl-main)

Challenge:	Existing LCU benchmarks for large language models often result in prohibitively high evaluation costs . existing benchmarks exhibit significant redundancy, which means inefficiency in evaluation .
Approach:	They propose a data compression method tailored for long-text data with sparse information characteristics.
Outcome:	The proposed method reduces evaluation costs to 4.5% of the long-text benchmark LongBench . the proposed method is based on a long-term LCU benchmark with sparse information characteristics .

Copied to clipboard

Huimin Wang, Baolin Peng, Kam-Fai Wong

CriticSearch: Fine-Grained Credit Assignment for Search Agents via a Retrospective Critic (2026.findings-acl)

Challenge:	Using reinforcement learning to learn dialogue policy requires a large volume of interactions with users.
Approach:	They propose a task-oriented dialogue agent that efficiently learns dialogue policy from demonstrations . they use an imitation model to distill knowledge from demonstration and reward shaping .
Outcome:	The proposed agent efficiently learns dialogue policy from demonstrations through policy shaping and reward shaping.

Copied to clipboard

Yaocheng Zhang, Haohuan Huang, Zijun Song, Zijie Zhao, Qichao Zhang, Yuanheng Zhu, Dongbin Zhao

STeCa: Step-level Trajectory Calibration for LLM Agent Learning (2025.findings-acl)

Challenge:	Existing search agent pipelines rely on sparse outcome rewards, leading to inefficient exploration and unstable training.
Approach:	They propose a tool-integrated reasoning framework that provides turn-level feedback via a retrospective critic mechanism.
Outcome:	The proposed framework outperforms baselines in multi-hop reasoning benchmarks and achieves faster convergence and training stability.

Copied to clipboard

Hanlin Wang, Jian Wang, Chak Tou Leong, Wenjie Li

Challenge:	Existing work focuses on behavior cloning from expert demonstrations or preference learning through exploratory trajectory sampling, but these methods often struggle to address long-horizon tasks where suboptimal actions accumulate step by step, causing agents to deviate from correct task trajectories.
Approach:	They propose a framework for LLM-based agent learning that identifies suboptimal actions through a step-level reward comparison during exploration and constructs calibrated trajectories using LLM reflection.
Outcome:	The proposed framework outperforms existing methods in long-horizon tasks where suboptimal actions accumulate step by step, causing agents to deviate from correct task trajectories.

E-ConvRec: A Large-Scale Conversational Recommendation Dataset for E-Commerce Customer Service (2022.lrec-1)

Copied to clipboard

Meihuizi Jia, Ruixue Liu, Peiying Wang, Yang Song, Zexi Xi, Haobin Li, Xin Shen, Meng Chen, Jinhui Pang, Xiaodong He

Natural Language Interfaces

Challenge:	Recent research has focused on developing conversational recommendation system (CRS), which provides valuable recommendations to users through conversations.
Approach:	They construct an authentic Chinese dialogue dataset consisting of over 25k dialogues and 770k utterances, which contains user profile, product knowledge base, and multiple sequential real conversations between users and recommenders.
Outcome:	The proposed dataset contains user profile, product knowledge base, and multiple sequential real conversations between users and recommenders.

Beyond Natural Language: LLMs Leveraging Alternative Formats for Enhanced Reasoning and Communication (2024.findings-emnlp)

Copied to clipboard

Weize Chen, Chenfei Yuan, Jiarui Yuan, Yusheng Su, Chen Qian, Cheng Yang, Ruobing Xie, Zhiyuan Liu, Maosong Sun

STARD: A Chinese Statute Retrieval Dataset Derived from Real-life Queries by Non-professionals (2024.findings-emnlp)

Challenge:	Natural language (NL) has long been the predominant format for human cognition and communication, but its utility in LLMs has not been thoroughly examined.
Approach:	They propose to allow LLMs to choose the most suitable format before reasoning or communicating, and to automate the selection process.
Outcome:	The proposed format improves reasoning efficiency and reduces token usage while maintaining communicative effectiveness.

Copied to clipboard

Weihang Su, Yiran Hu, Anzhe Xie, Qingyao Ai, Quezi Bing, Ning Zheng, Yun Liu, Weixing Shen, Yiqun Liu

Shorten After You’re Right: Lazy Length Penalties for Reasoning RL (2026.findings-acl)

Challenge:	Existing statute retrieval benchmarks emphasize formal and professional queries from sources like bar exams and legal case documents . existing retrieval approaches that lack domain-specific knowledge may struggle to capture the meanings of specialized terms accurately.
Approach:	They propose a dataset that captures the complexity and diversity of real queries from the general public.
Outcome:	The proposed dataset captures the complexity and diversity of real queries from the general public.

Copied to clipboard

Danlong Yuan, Tian Xie, Shaohan Huang, Huishuai Zhang, Zhuocheng Gong, Chong Luo, Furu Wei, Dongyan Zhao

CityEQA: A Hierarchical LLM Agent on Embodied Question Answering Benchmark in City Space (2025.emnlp-main)

Challenge:	Existing shortening methods for long reasoning models rely on additional supervision or multi-stage post-training.
Approach:	They propose a lazy length penalty that imposes length pressure on models without extra training stages.
Outcome:	The proposed method significantly reduces response length without extra training stages while maintaining or improving performance.

Copied to clipboard

Yong Zhao, Kai Xu, Zhengqiu Zhu, Yue Hu, Zhiheng Zheng, Yingfeng Chen, Yatai Ji, Chen Gao, Yong Li, Jincai Huang

Natural Language Interfaces

Challenge:	Embodied Question Answering (EQA) tasks are primarily focused on indoor environments, leaving the complexities of urban settings unexplored.
Approach:	They propose a task where an embodied agent answers open-vocabulary questions in dynamic city spaces.
Outcome:	The proposed agent achieves 60.7% of human-level answering accuracy compared to baselines . the proposed agent outperforms existing agents in open-ended city spaces .

Designing Logic Pattern Templates for Counter-Argument Logical Structure Analysis (2024.findings-emnlp)

Copied to clipboard

Shoichi Naito, Wenzhi Wang, Paul Reisert, Naoya Inoue, Camélia Guerraoui, Kenshi Yamaguchi, Jungmin Choi, Irfan Robbani, Surawat Pothong, Kentaro Inui

Living in the Moment: Can Large Language Models Grasp Co-Temporal Reasoning? (2024.acl-long)

Challenge:	Despite their effectiveness, the logical attack structure of counterarguments remains unexplored due to its complexity.
Approach:	They propose a task to analyze logical attack structure of counterarguments in relation to their corresponding opponent argument using 10 new CA logic patterns.
Outcome:	The proposed task achieves high annotator agreement and coverage and high coverage on a dataset of 778 CAs.

Copied to clipboard

Zhaochen Su, Juntao Li, Jun Zhang, Tong Zhu, Xiaoye Qu, Pan Zhou, Yan Bowen, Yu Cheng, Min Zhang

Has It All Been Solved? Open NLP Research Questions Not Solved by Large Language Models (2024.lrec-main)

Challenge:	Current temporal reasoning datasets are limited to questions about single or isolated events, falling short in mirroring the realistic temporal characteristics involving concurrent nature and intricate temporal interconnections.
Approach:	They propose a co-temporal Question Answering benchmark that contains four co-time scenarios with 4,748 samples for evaluating the co-timing abilities of large language models.
Outcome:	The proposed benchmarks show that current LLMs struggle on CoTempQA tasks even when enhanced with Chain of Thought methodologies.

Copied to clipboard

Oana Ignat, Zhijing Jin, Artem Abzaliev, Laura Biester, Santiago Castro, Naihao Deng, Xinyi Gao, Aylin Ece Gunal, Jacky He, Ashkan Kazemi, Muhammad Khalifa, Namho Koh, Andrew Lee, Siyang Liu, Do June Min, Shinka Mori, Joan C. Nwatu, Veronica Perez-Rosas, Siqi Shen, Zekun Wang, Winston Wu, Rada Mihalcea

CluHTM - Semantic Hierarchical Topic Modeling based on CluWords (2020.acl-main)

Challenge:	Recent advances in large language models have led to misleading public discourse that “it’s all been solved.”
Approach:	They identify 14 research areas encompassing 45 research directions that require new research and are not directly solvable by LLMs.
Outcome:	The research areas identified are 45 research directions that require new research and are not directly solvable by LLMs.

Copied to clipboard

Felipe Viegas, Washington Cunha, Christian Gomes, Antônio Pereira, Leonardo Rocha, Marcos Goncalves

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

Challenge:	Hierarchical Topic modeling (HTM) exploits latent topics and relationships among them as a powerful tool for data analysis and exploration.
Approach:	They propose a hierarchical matrix factorization that exploits latent topics and relationships among them to create a powerful tool for data analysis and exploration.
Outcome:	The proposed method outperforms baselines and datasets in the vast majority of cases.

Copied to clipboard

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

1+1>2: Can Large Language Models Serve as Cross-Lingual Knowledge Aggregators? (2024.emnlp-main)

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

Yue Huang, Chenrui Fan, Yuan Li, Siyuan Wu, Tianyi Zhou, Xiangliang Zhang, Lichao Sun

Sampling-Based Approximations to Minimum Bayes Risk Decoding for Neural Machine Translation (2022.emnlp-main)

Challenge:	Large Language Models (LLMs) have been recognized for their impressive capabilities in natural language processing (NLP).
Approach:	They propose a method to enhance the multilingual performance of Large Language Models by aggregating knowledge from diverse languages.
Outcome:	The proposed method reduces the performance disparity across languages and offers valuable insights for further exploration.

Copied to clipboard

Bryan Eikema, Wilker Aziz

Machine Translation

Challenge:	Existing methods to improve beam search quality are inadequate in many ways . a new approximation to the beam search curse has been proposed .
Approach:	They propose an approximation to minimum Bayes risk decoding that would solve the beam search curse.
Outcome:	The proposed approximation has no equivalent to the beam search curse.

KRLS: Improving End-to-End Response Generation in Task Oriented Dialog with Reinforced Keywords Learning (2023.emnlp-main)

Copied to clipboard

Xiao Yu, Qingyang Wu, Kun Qian, Zhou Yu

Student Guides Teacher: Weak-to-Strong Inference via Spectral Orthogonal Exploration (2026.acl-long)

Challenge:	Task-oriented dialogs (TOD) require a model to generate a response that optimizes for task-related metrics.
Approach:	They propose a faster generation procedure that samples from independent next-word distributions and introduce a fine-grained reward function to help the model focus on learning key information in a dialog.
Outcome:	The proposed algorithm achieves state-of-the-art performance on an offline task with 15% training time reduction compared to a standard RL algorithm using auto-regressive generation.

Copied to clipboard

Dayu Wang, Jiaye Yang, Weikang Li, Jiahui Liang, Yang Li, Deguo Xia, Jizhou Huang

The Hidden Strength of Disagreement: Unraveling the Consensus-Diversity Tradeoff in Adaptive Multi-Agent Systems (2025.emnlp-main)

Challenge:	Existing Large Language Models suffer from "Reasoning Collapse" on mathematical reasoning tasks where stochastic sampling produces lexical variations of the same erroneous logic rather than genuine semantic exploration.
Approach:	They propose a geometric inference framework that uses a spectral orthogonal probe to introduce semantically heterogeneous reasoning signals into the teacher's orthogonale complement of its dominant subspace.
Outcome:	The proposed framework improves accuracy and sampling efficiency over baseline methods on logic and code generation benchmarks.

Copied to clipboard

Zengqing Wu, Takayuki Ito

Low-Resource NLP

Challenge:	Conventional LLM-based MAS rely on explicit coordination, e.g., prompts or voting, risking premature homogenization.
Approach:	They propose to preserve partial diversity by combining in-context learning with explicit coordination to form consensus in dynamic environments.
Outcome:	The proposed model outperforms explicit consensus models on three scenarios showing that partial deviation from group norms boosts exploration, robustness, and performance.

ProCeedRL: Process Critic with Explorative Demonstration Reinforcement Learning for LLM Agentic Reasoning (2026.findings-acl)

Copied to clipboard

Jingyue Gao, Yanjiang Guo, Chen Xiaoshuai, Jianyu Chen

VisEscape: A Benchmark for Evaluating Exploration-driven Decision-making in Virtual Escape Rooms (2025.emnlp-main)

Challenge:	Large Language Models (LLMs) exhibit exceptional reasoning capabilities, driven by Reinforcement Learning with Verifiable Rewards (RLVR).
Approach:	They propose a method that uses a process-level critic to monitor interactions in real time, incorporating reflection-based demonstrations to guide agents in stopping the accumulation of errors.
Outcome:	The proposed approach exceeds the model’s saturated exploration performance and achieves superior performance on complex deep search and embodied tasks.

Copied to clipboard

Seungwon Lim, Sungwoong Kim, Jihwan Yu, Sungjae Lee, Jiwan Chung, Youngjae Yu

Cognitive Modeling

Challenge:	Existing studies on embodied agents have addressed the importance of exploration in environments where tasks and solutions are not predefined.
Approach:	They propose a virtual escape room that evaluates AI models in a dynamic environment . they propose to integrate memory management and reasoning into the simulation .
Outcome:	The proposed model improves in dynamic and exploration-driven environments by integrating memory management and reasoning.

Targeted Exploration via Unified Entropy Control for Reinforcement Learning (2026.findings-acl)

Copied to clipboard

Chen Wang, Lai Wei, Yanzhi Zhang, Chenyang Shao, Zedong Dan, Weiran Huang, Ge Lan, Yue Wang

Scattered Hypothesis Generation for Open-Ended Event Forecasting (2026.findings-acl)

Challenge:	Existing methods for group relative policy optimization suffer from entropy collapse . Existing exploration methods introduce additional bias or variance during exploration, making it difficult to maintain stability.
Approach:	They propose a framework that provides targeted mechanisms for exploration and stabilization.
Outcome:	The proposed framework expands search space on difficult prompts while preventing entropy growth uncontrollably.

Copied to clipboard

He Chang, Zhulin Tao, Lifang Yang, Xianglin Huang, Yunshan Ma

Challenge:	Existing methods for event forecasting focus on the most probable outcomes, neglecting the intrinsic uncertainty of real-world events.
Approach:	They propose a reinforcement learning framework that optimizes inclusiveness and diversity of the hypothesis by integrating validity-gated score into the overall objective.
Outcome:	The proposed framework outperforms baselines on two real-world benchmark datasets.

SPS: Steering Probability Squeezing for Better Exploration in Reinforcement Learning for Large Language Models (2026.findings-acl)

Copied to clipboard

Yifu Huo, Chenglong Wang, Ziming Zhu, Shunjie Xing, Peinan Feng, Tongran Liu, Qiaozhi He, Tian Hua Zhou, null Changxiaojia, JingBo Zhu, Zhengtao Yu, Tong Xiao

Leveraging Information Redundancy of Real-World Data through Distant Supervision (2024.lrec-main)

Challenge:	Reinforcement learning (RL) training typically improves single-sample success rates but limited exploration of diverse reasoning trajectories.
Approach:	They propose a training paradigm that interleaves conventional RL with inverse reinforcement learning (IRL) they propose 'Steering Probability Squeezing' to enhance exploration without external supervision .
Outcome:	The proposed training paradigm improves Pass@k and improves exploration of diverse reasoning trajectories without external supervision.

Copied to clipboard

Ariel Cohen, Alexandrine Lanson, Emmanuelle Kempf, Xavier Tannier

Challenge:	Existing methods for annotation of health care notes are promising but they are limited due to privacy regulations.
Approach:	They propose a text labeling method that leverages the redundancy of temporal information in a data lake to create a large programmatically annotated corpus and train transformer models using distant supervision.
Outcome:	The proposed method reduces expert annotation time, a scarce and expensive resource.

SEMA-RAG: A Self-Evolving Multi-Agent Retrieval-Augmented Generation Framework for Medical Reasoning (2026.findings-acl)

Copied to clipboard

Yongfeng Huang, Ruiying Chen, James Cheng

Selective Prompting Tuning for Personalized Conversations with LLMs (2024.findings-acl)

Challenge:	Standard RAG frameworks treat retrieval as a static, single-round auxiliary step . compressed workflow makes it difficult to form reliable evidence chains .
Approach:	They propose a framework that decouples tasks and allows for dynamic multi-round exploration . they propose retrieval-augmented generation (RAG) to mitigate hallucinations and knowledge obsolescence .
Outcome:	The proposed framework improves the strongest baseline by +6.46 accuracy points on average across five benchmarks and five LLM backbones.

Copied to clipboard

Qiushi Huang, Xubo Liu, Tom Ko, Bo Wu, Wenwu Wang, Yu Zhang, Lilian Tang

Low-Resource NLP

Challenge:	Personalization in conversational AI requires persona profiles and contextual understanding to create meaningful conversations.
Approach:	They propose a method that softly prompts LLMs for personalized conversations in a selective way.
Outcome:	The proposed approach improves response diversity by up to 90% on the CONVAI2 dataset.

ThinkQE: Query Expansion via an Evolving Thinking Process (2025.findings-emnlp)

Copied to clipboard

Yibin Lei, Tao Shen, Andrew Yates

Semantic-Aware Action Space Compression via LLM-DRL Synergy for Efficient Task-oriented Dialogue Policy Exploration (2025.findings-emnlp)

Challenge:	LLM-based methods often generate narrowly focused expansions that overlook these desiderata.
Approach:	They propose a test-time query expansion framework that promotes exploration and result diversity . ThinkQE encourages deeper and comprehensive semantic exploration and a corpus-interaction strategy that iteratively refines expansions .
Outcome:	The proposed framework outperforms prior approaches on diverse web search benchmarks.

Copied to clipboard

Yangyang Zhao, Ben Niu, Yuxuan Tan, Shihan Wang, Libo Qin

FOSSIL: Harnessing Feedback on Suboptimal Samples for Data-Efficient Generalisation with Imitation Learning for Embodied Vision-and-Language Tasks (2025.findings-emnlp)

Challenge:	Pre-trained large language models (LLMs) with world knowledge and semantic understanding are promising for task-oriented dialogue systems.
Approach:	a framework that synergizes pre-trained large language models with DRL is proposed . a lightweight action pruning mechanism is employed to eliminate implausible actions .
Outcome:	a new framework synergizes pre-trained large language models with DRL to guide decision-making . the proposed framework eliminates semantically implausible or low-potential actions from multi-turn dialogue context .

Copied to clipboard

Sabrina McCallum, Amit Parekh, Alessandro Suglia

Step-GRPO: Internalizing Dynamic Early Exit for Efficient Reasoning (2026.acl-long)

Challenge:	Current approaches to embodied AI tend to learn policies from expert demonstrations, but without a mechanism to evaluate the quality of demonstrated actions, they are limited to learning from optimal behaviour or risk replicating errors and inefficiencies.
Approach:	They propose to embed language feedback into a Transformer-based policy and optionally complement the traditional next action prediction objective with auxiliary self-supervised learning objectives for feedback prediction.
Outcome:	The proposed method improves agents’ compositional generalisation abilities and robustness on a range of embodied Vision-and-Language tasks in a custom babyAI-XGen environment.

Copied to clipboard

Benteng Chen, Weida Wang, Shufei Zhang, Mingbao Lin, Min Zhang

Green & Sustainable NLP

Challenge:	Large reasoning models that use long chain-of-thought excel at problem-solving but waste computational resources.
Approach:	They propose a framework that internalizes dynamic early-exit capabilities directly into the model.
Outcome:	The proposed framework reduces token consumption by 32.0% on a Qwen3-8B model compared to the vanilla model .

EvoCoT: Overcoming the Exploration Bottleneck in Reinforcement Learning for LLMs (2026.findings-acl)

Copied to clipboard

Huanyu Liu, Jia Li, Yihong Dong, Chang Yu, Taozhi Chen, Lecheng Wang, Yongding Tao, Bin Gu, Ge Li

Breaking Consensus Bias: Unsupervised Reinforcement Learning for Machine Translation (2026.findings-acl)

Challenge:	Existing approaches to reinforcement learning with verifiable reward (RLVR) are limited by difficulty or lack of exploration.
Approach:	They propose a self-evolving curriculum learning framework based on chain-of-thought reasoning optimization that constrains exploration space by self-generating and verifying CoT trajectories.
Outcome:	The proposed framework enables LLMs to solve previously unsolved problems without external supervision and is compatible with various RL fine-tuning methods.

Copied to clipboard

Shuting Jiang, Ran Song, Siqi Zhang, Yuxin Huang, Shengxiang Gao, Zhengtao Yu

Machine Translation

Challenge:	Existing RL approaches for MT face fixed references or the production of homogeneous references leading to mode collapse in unsupervised settings.
Approach:	They propose an Entropy-Driven Unsupervised RL framework for machine translation that leverages entropy for supervision construction and self-evolution.
Outcome:	The proposed framework outperforms supervised and unsupervised baselines in multiple language pairs.

Can LLMs Reason About Program Semantics? A Comprehensive Evaluation of LLMs on Formal Specification Inference (2025.acl-long)

Copied to clipboard

Thanh Le-Cong, Bach Le, Toby Murray

ProUIE: A Macro-to-Micro Progressive Learning Method for LLM-based Universal Information Extraction (2026.findings-acl)

Challenge:	Large Language Models (LLMs) are increasingly being used to automate programming tasks.
Approach:	They propose a benchmark to evaluate LLMs' reasoning abilities on program semantics.
Outcome:	The proposed benchmark shows that LLMs perform well with simple control flows but struggle with more complex structures, especially loops, even with advanced prompting.

Copied to clipboard

Wenda Liu, Song Zhigang, Shuai Nie, Guangyao Liu, Lisung Chen, Binyu Yang, Yaran Chen, Peng Zhou, Hongzhen Wang, Yuchen Liu, Wenyue Hu, Jiaming Xu, Runyu Shi, Ying Huang

Consolidation or Adaptation? PRISM: Disentangling SFT and RL Data via Gradient Concentration (2026.acl-long)

Challenge:	ProUIE improves universal information extraction (UIE) without external information . many LLM-based methods rely on extra schema cues, external resources or complex alignment and verification pipelines .
Approach:	They propose a Macro-to-Micro progressive learning approach that improves UIE without external information.
Outcome:	ProUIE outperforms instruction-tuned baselines on average for NER and RE while using a smaller backbone.

Copied to clipboard

Yang Zhao, Yangou Ouyang, Xiao Ding, Hepeng Wang, Bibo Cai, Kai Xiong, Jinglong Gao, Zhouhao Sun, Li Du, Bing Qin, Ting Liu

SPARK: Strategic Policy-Aware Exploration via Dynamic Branching for Long-Horizon Agentic Learning (2026.acl-long)

Challenge:	Existing data arbitration strategies for large language model training rely on surface-level heuristics that fail to diagnose intrinsic learning needs.
Approach:	They propose a framework that arbitrates data based on its degree of cognitive conflict with the model's existing knowledge.
Outcome:	Extensive experiments on WebShop and ALFWorld show that PRISM outperforms state-of-the-art hybrid methods while reducing computational costs by up to 3.22 .

Copied to clipboard

Jinyang Wu, Shuo Yang, Yuhao Shen, Shuai Zhang, Zhengqi Wen, Jianhua Tao

Challenge:	Existing methods for training large language models waste computation budget on trivial steps while failing to guarantee sample quality.
Approach:	They propose a framework that selectively branches at critical decision states for resource-efficient exploration.
Outcome:	The proposed framework activates adaptive branching exploration at critical decision states to probe promising trajectories, thereby achieving precise resource allocation that prioritizes sampling quality over blind coverage.

Low-probability Tokens Sustain Exploration in Reinforcement Learning with Verifiable Reward (2026.findings-acl)

Copied to clipboard

Guanhua Huang, Tingqiang Xu, Mingze Wang, Qi Yi, Xue Gong, Siheng Li, Ruibin Xiong, Kejiao Li, Yuhao Jiang, Bo Zhou

Diffuse Thinking: Exploring Diffusion Language Models as Efficient Thought Proposers for Reasoning (2026.acl-long)

Challenge:	Recent studies show that RLVR training is slow and results plateau as policy entropy collapses . low-probability regularization (Lp-Reg) reduces the number of low-quality exploratory tokens induced by RL training .
Approach:	They propose a method to reduce RLVR over-penalization by eliminating low-probability exploratory tokens . they propose 'Low-provability Regularization' to reduce the gradual elimination of low-quality exploratory entropy tokens.
Outcome:	The proposed method eliminates low-probability exploratory tokens and prevents suppression of potentially valuable low-property candidates.

Copied to clipboard

Chenyang Shao, Sijian Ren, Fengli Xu, Yong Li

VLN-MME: Diagnosing MLLMs as Language-guided Visual Navigation Agents (2026.acl-long)

Challenge:	Large language models (LLMs) have demonstrated strong capabilities in complex reasoning tasks, yet their autoregressive generation paradigm makes it computationally prohibitive to explore diverse reasoning paths.
Approach:	They propose a framework that combines diffusion-based generation with autoregressive evaluation to efficiently generate diverse intermediate reasoning thoughts and employ LLMs as evaluators to assess and select candidates based on their plausibility and correctness.
Outcome:	The proposed framework improves inference efficiency while maintaining competitive or superior reasoning accuracy.

Copied to clipboard

Xunyi Zhao, Gengze Zhou, Qi Wu

Multimodality

Challenge:	Multimodal Large Language Models have demonstrated remarkable capabilities across vision-language tasks, but their performance as embodied agents needs further exploration.
Approach:	They propose a framework to evaluate multimodal large language models as zero-shot agents . they find that enhancing prevalent agents with Chain-of-Thought reasoning and self-reflection leads to an unexpected performance decrease.
Outcome:	The proposed framework enables comparisons and component-level ablations across diverse MLLM architectures, agent designs, and navigation tasks.

Learning from Mistakes: Negative Reasoning Samples Enhance Out-of-Domain Generalization (2026.acl-long)

Copied to clipboard

Tian Xueyun, MingHua Ma, Bingbing Xu, Nuoyan Lyu, Wei Li, Heng Dong, Zheng Chu, Yuanzhuo Wang, Huawei Shen

LaMDAgent: An Autonomous Framework for Post-Training Pipeline Optimization via LLM Agents (2025.emnlp-main)

Challenge:	Recent studies show that supervised fine-tuning (SFT) is a common approach for reasoning in large language models.
Approach:	They propose to use supervised fine-tuning (SFT) on chain-of-thought trajectories demonstrations . they find that incorporating negative traxories yields substantial OOD generalization gains .
Outcome:	The proposed scheme yields 5.51% OOD gain over positive-only training.

Copied to clipboard

Taro Yano, Yoichi Ishibashi, Masafumi Oyamada

Challenge:	Existing approaches to optimize large language models rely on manual design or focus on optimizing individual components.
Approach:	They propose a LaMDAgent framework that constructs and optimizes end-to-end post-training pipelines by exploring various model improving methods, objects, and their applied orderings based on task-based feedback.
Outcome:	The proposed framework achieves a 9.0-point gain in tool-use accuracy without degrading instruction-following, and reduces computational costs.

XAI-Attack: Utilizing Explainable AI to Find Incorrectly Learned Patterns for Black-Box Adversarial Example Creation (2024.lrec-main)

Copied to clipboard

Markus Bayer, Markus Neiczer, Maximilian Samsinger, Björn Buchhold, Christian Reuter

PACE: Prefix-Protected and Difficulty-Aware Compression for Efficient Reasoning (2026.findings-acl)

Challenge:	Adversarial examples can be used to trick machine learning models into making erroneous predictions, causing poorer insights and lower confidence in the information gathered.
Approach:	They propose a textual adversarial example method that identifies falsely learned word indicators by leveraging explainable AI methods as importance functions on incorrectly predicted instances.
Outcome:	The proposed method outperforms existing examples and training methods and shows baseline improvements of up to 23 percentage points on adversarial tasks.

Copied to clipboard

Ruixiang Feng, Yuntao Wen, Silin Zhou, Ke Shi, Yifan Wang, Ran Le, Zhenwei An, Zongchao Chen, Chen Yang, Guangyue Peng, Yiming Jia, Dongsheng Wang, Tao Zhang, Lisi Chen, Yang Song, Shen Gao, Shuo Shang

CreativeBench: Benchmarking and Enhancing Machine Creativity via Self-Evolving Challenges (2026.findings-acl)

Challenge:	Existing LRMs often suffer from "overthinking" and excessively long reasoning traces . a dual-level framework for length compression of LRM is proposed .
Approach:	They propose a framework for prefix-protected and difficulty-aware compression under hierarchical supervision.
Outcome:	The proposed framework reduces token usage while improving accuracy on math benchmarks.

Copied to clipboard

Zihan Wang, Lam Nguyen, Zhengyang Zhao, Mengyue Yang, Chengwei Qin, Yujiu Yang, Linyi Yang

Challenge:	Increasing saturation of web data limits further scaling of model intelligence.
Approach:	They propose a benchmark to evaluate machine creativity in code generation that combines combinatorial and exploratory creativity through reverse engineering and self-play.
Outcome:	The proposed benchmark targets combinatorial and exploratory creativity through reverse engineering and self-play.

SeDev: Structured Semantic Exploration for LLM-Driven Code Generation (2026.acl-long)

Copied to clipboard

Ronghui Yang, Jie Liu, Jiajie Zeng, Jiexin Wang, Jiuchuan Jiang, Bo An, Yi Cai, Mengchen Zhao

Programming Languages in NLP

Challenge:	Large Language Models (LLMs) have shown remarkable capabilities in automating code generation, but they suffer from insufficient exploration of the vast solution space.
Approach:	They propose a large-scale LLM-driven code generation framework that efficiently finds high-quality solutions in only a few iterations.
Outcome:	The proposed framework outperforms baselines while maintaining reasonable time and computational costs.

Beyond High-Entropy Exploration: Correctness-Aware Low-Entropy Segment-Based Advantage Shaping for Reasoning LLMs (2026.findings-acl)

Copied to clipboard

Xinzhu Chen, Xuesheng Li, Zhongxiang Sun, Weijie Yu

GraphMind: LLMs as Dynamic Knowledge Builders for Sequential Decision-Making (2026.findings-acl)

Challenge:	Recent work studies RLVR through token entropy, arguing that high-entropies drive exploration and should receive stronger updates.
Approach:	They propose a correctness-aware reinforcement framework that performs fine-grained advantage modulation over low-entropy segments.
Outcome:	The proposed framework improves accuracy over strong RL baselines across three backbones and six math benchmarks while maintaining high-entropy exploration.

Copied to clipboard

Sunguk Shin, Hayeong Lee, Jun Ho Seo, Jinho Lee, Myunsoo Kim, Minsuk Chang, Byung-Jun Lee

SQL-Trail: Multi-Turn Reinforcement Learning with Interleaved Feedback for Text-to-SQL (2026.acl-long)

Challenge:	Large language models (LLMs) have demonstrated remarkable performance in natural language understanding and generation, establishing themselves as foundational tools across a wide range of domains.
Approach:	They propose an LLM agent architecture that integrates a knowledge graph as a graph-based memory module and integrates it into the agent to generate efficient plans.
Outcome:	The proposed architecture improves the performance and efficiency of the LLM in navigation tasks designed to present long-horizon and partially observable challenges.

Copied to clipboard

Harper Hua, Zhen Han, Zhengyuan Shen, Meng-Chieh Lee, Sheng Guan, Qi Zhu, Sullam Jeoung, Yueyan Chen, Yunfei Bai, Shuai Wang, Vassilis N. Ioannidis, Huzefa Rangwala

Programming Languages in NLP

Challenge:	Recent large language models (LLMs) have significantly improved Text-to-SQL generation, but a gap remains between AI systems and human experts on challenging benchmarks such as BIRD-Sql.
Approach:	They propose a multi-turn reinforcement learning agentic framework for Text-to-SQL that uses execution feedback to iteratively refine its predictions.
Outcome:	The proposed framework outperforms proprietary systems on 7B and 14B models by **5% on average, underscoring the effectiveness of interactive, agentic workflows for robust Text-to-SQL generation.

Robust Tool Use via Fission-GRPO: Learning to Recover from Execution Errors (2026.acl-long)

Copied to clipboard

Zhiwei Zhang, Fei Zhao, Rui Wang, Zezhong Wang, Bin Liang, Jiakang Wang, Yao Hu, Shaosheng Cao, Kam-Fai Wong

Robustness in NLP

Challenge:	Large language models (LLMs) can call tools effectively, but they remain brittle in multi-turn execution.
Approach:	They propose a framework that converts execution errors into on-policy corrective supervision within the RL training loop.
Outcome:	The proposed framework improves the error recovery rate of Qwen3-8B by 5.7% absolute and overall accuracy by 4.0% on BFCL v4 Multi-Turn.

From log 𝜋 to 𝜋: Taming Divergence in Soft Clipping via Bilateral Decoupled Decay of Probability Gradient Weight (2026.acl-long)

Copied to clipboard

Xiaoliang Fu, Jiaye Lin, Yangyi Fang, Chaowen Hu, Cong Qin, Zekai Shao, Binbin Zheng, Lu Pan, Ke Zeng

Rhombus: Incentivizing Coordination in Parallel Thinking through Reinforcement Learning (2026.findings-acl)

Challenge:	Standard algorithms for Large Language Models (LLMs) enforce stability via "hard clipping" but relying on log-probability gradient yields divergent weights as probabilities vanish, destabilizing LLM training.
Approach:	They propose a decoupled gradient policy optimization that uses a decay mechanism to decouple the probability of a boundary token.
Outcome:	The proposed algorithm outperforms baselines on various mathematical benchmarks.

Copied to clipboard

Ziyuan Nan, Qi Yi, Di Huang, Yutong Wu, Guanhua Huang, Xue Gong, Kejiao Li, Yuhao Jiang, Chenchen Zhang, Zenan Xu, Xing Hu, Bo Zhou

AdaFuse: Adaptive Ensemble Decoding for Large Language Models (2026.acl-long)

Challenge:	Parallel thinking is a promising avenue for scaling test-time compute in Large Language Models . however, coordinating the exploration and aggregation stages remains challenging .
Approach:	They propose a parallel thinking framework that explicitly incentivizes coordination between components via end-to-end reinforcement learning.
Outcome:	The proposed framework improves accuracy by 6.0% over long chain-of-thought baselines while reducing wall-clock latency by 39.4% under matched token budgets.

Copied to clipboard

Chengming Cui, Tianxin Wei, Ziyi Chen, Ruizhong Qiu, Zhichen Zeng, Zhining Liu, Xuying Ning, Duo Zhou, Jingrui He

Rewarding the Rare: Uniqueness-Aware RL for Creative Problem Solving in LLMs (2026.findings-acl)

Challenge:	Existing ensemble approaches to large language models lack flexibility for mid-generation adaptation.
Approach:	They propose an adaptive ensemble decoding framework that dynamically selects semantically appropriate fusion units during generation.
Outcome:	The proposed framework outperforms existing ensemble frameworks on open-domain QA, arithmetic reasoning, and machine translation tasks.

Copied to clipboard

Zhiyuan Hu, Yucheng Wang, Yufei He, Jiaying Wu, Yilun Zhao, See-Kiong Ng, Cynthia Breazeal, Anh Tuan Luu, Hae Won Park, Bryan Hooi

AgentGym2: Benchmarking Large Language Model Agents in De-Idealized Real-World Environments (2026.acl-long)

Challenge:	Reinforcement learning (RL) is a paradigm for post-training large language models, but it suffers from exploration collapse . a new study finds that RL fails to reward correct solutions that exhibit rare high-level strategies .
Approach:	They propose a method that rewards correct solutions that exhibit rare high-level strategies by clustering rollouts according to their high- level solution strategies.
Outcome:	The proposed approach improves pass@k across large sampling budgets and increases area under the pass@K curve (AUC@K) without sacrificing pass@1.

Copied to clipboard

Zhiheng Xi, Dingwen Yang, Jiaqi Liu, Jixuan Huang, Honglin Guo, Baodai Huang, Tinggang Chen, Qi Zhang, Zhonghang Lu, Chenyu Liu, Jiajun Sun, Jiazheng Zhang, Dingwei Zhu, Xin Guo, Junzhe Wang, Zhihao Zhang, Yuming Yang, Junjie Ye, Minghe Gao, Dongrui Liu, Jiaming Ji, Guohao Li, Tao Gui, Qi Zhang, Xuanjing Huang