INTERMEDIATE REASONING

RiT: Rubrics-in-Thinking Reinforcement Learning for Improved Reasoning in Large Language Models (2026.findings-acl)

Challenge:	Existing state-of-the-art language models do not make intermediate reasoning steps explicit . large pretrained language models such as BERT and RoBERTa have been successfully used in multi-hop reasoning problems .
Approach:	They propose to decompose multi-hop reasoning problems into several simple ones and use natural language to guide intermediate reasoning hops.
Outcome:	The proposed model can generate subgoals and perform inference in natural language at each reasoning step.

Copied to clipboard

Xiaobin Tian, Shuai Yuan, Muyun Ding, Haonan Chen, Xiaoxi Jiang

From Reasoning to Answer: Empirical, Attention-Based and Mechanistic Insights into Distilled DeepSeek R1 Models (2025.emnlp-main)

Challenge:	Large Reasoning Models benefit from generating intermediate reasoning steps alongside final answers.
Approach:	They propose a framework to introduce thinking-rubric supervision into intermediate reasoning.
Outcome:	The proposed framework outperforms outcome-only RL baselines on reasoning-intensive and open-ended tasks.

Copied to clipboard

Jue Zhang, Qingwei Lin, Saravan Rajmohan, Dongmei Zhang

Faithful Question Answering with Monte-Carlo Planning (2023.acl-long)

Challenge:	Large Reasoning Models generate explicit reasoning traces alongside final answers . the extent to which these traces influence answer generation remains unclear .
Approach:	They conduct empirical evaluation of Large Reasoning Models that include explicit reasoning . they also show that answer tokens attend substantially to reasoning tokens .
Outcome:	The results show that including explicit reasoning improves answer quality across domains . they also show that answer tokens attend substantially to reasoning tokens - the authors .

Copied to clipboard

Ruixin Hong, Hongming Zhang, Hong Zhao, Dong Yu, Changshui Zhang

Question Answering

Challenge:	Existing approaches to answer questions using large language models lack the ability to faithfully follow the intermediate reasoning steps from the known premises to the answer.
Approach:	They propose a faithful question-answering task that uses a Monte-Carlo planning algorithm to produce faithful reasoning steps from the known premises to the answer.
Outcome:	The proposed task can produce valid and faithful reasoning steps compared with large language models with a much smaller model size.

The Art of SOCRATIC QUESTIONING: Recursive Thinking with Large Language Models (2023.emnlp-main)

Copied to clipboard

Jingyuan Qi, Zhiyang Xu, Ying Shen, Minqian Liu, Di Jin, Qifan Wang, Lifu Huang

Noisy Exemplars Make Large Language Models More Robust: A Domain-Agnostic Behavioral Analysis (2023.emnlp-main)

Challenge:	Chain-of-Thought (CoT) prompting relies on the initial decisions, causing errors in early steps to accumulate and impact the final answers.
Approach:	They propose a divide-and-conquer style algorithm that leverages large language models to raise and answer sub-questions until collecting enough information to tackle the original one.
Outcome:	The proposed algorithm is more robust to errors and errors than CoT prompting and Tree-of-Thought prompting methods.

Copied to clipboard

Hongyi Zheng, Abulhair Saparov

Your Co-Workers Matter: Evaluating Collaborative Capabilities of Language Models in Blocks World (2024.findings-acl)

Challenge:	Existing studies on the robustness of LLMs with few-shot prompting techniques are limited.
Approach:	They propose to test the robustness of LLMs in multi-hop reasoning tasks via domain-agnostic perturbations.
Outcome:	The proposed model is more sensitive to certain perturbations such as replacing words with synonyms and more robust to few-shot prompting methods.

Copied to clipboard

Guande Wu, Chen Zhao, Claudio Silva, He He

Faithful-First Reasoning, Planning, and Acting for Multimodal LLMs (2026.findings-acl)

Challenge:	Existing studies on how large language model agents collaborate with humans in equal roles emphasize the importance of coordination and communication.
Approach:	They propose to use chain-of-thought prompts to evaluate different collaboration perspectives, from independent to more complex, dependent tasks.
Outcome:	The proposed model significantly improves the evaluation metric.

Copied to clipboard

Junxian Li, Xinyue Xu, Sai Ma, Di Zhang, Sichao Li

Explainable Question Answering based on Semantic Graph by Global Differentiable Learning and Dynamic Adaptive Reasoning (2022.emnlp-main)

Challenge:	Existing efforts to improve task accuracy or enrich COT generation are lacking in multimodal large language models.
Approach:	They propose a Faithful-First Reasoning, Planning, and Acting framework that evaluates faithfulness of intermediate reasoning and uses it to plan and execute faithfulness-aware actions during inference.
Outcome:	The proposed framework improves perceptual faithfulness by up to 24% over prompt-based and tool-augmented reasoning frameworks without degrading task accuracy.

Copied to clipboard

Jianguo Mao, Wenbin Jiang, Xiangdong Wang, Hong Liu, Yu Xia, Yajuan Lyu, QiaoQiao She

Explainability & Interpretability in NLP

Challenge:	Existing models for multi-hop Question Answering have improved the implicit reasoning ability, but the black box nature of pure neural networks has hindered the construction of explainable intelligent systems.
Approach:	They propose a global differentiation strategy to explore optimal reasoning paths from latent probability space and a Dynamic Adaptive Reasoner to enhance generalization of unseen questions.
Outcome:	The proposed method achieves 17% improvements in F1-score against BreakRC and shows better interpretability.

MirageBackdoor: A Stealthy Attack that Induces Think-Well-Answer-Wrong Reasoning (2026.acl-long)

Copied to clipboard

Yizhe Zeng, Wei Zhang, Yunpeng Li, Juxin Xiao, Xiao Wang, Yuling Liu

Robustness in NLP

Challenge:	Existing CoT backdoor attacks manipulate intermediate reasoning steps to steer the model toward incorrect answers, but these corrupted reasoning traces are readily detected by prevalent process-monitoring defenses.
Approach:	They propose a backdoor attack that exploits the model's post-output space to preserve clean CoTs while selectively steering the final answer toward a specific target.
Outcome:	Experiments show that MirageBD achieves over 90% success rate across four datasets and five models with a poison ratio of only 5%.

Offloaded Reasoning: Efficient Inference for Large Language Models via Modular Reasoning and Refinement (2025.findings-emnlp)

Copied to clipboard

Ishan Jindal, Jayant Taneja, Badrinath Chandana, Vikas Kapur, Sachin Dev Sharma

Bi-Chainer: Automated Large Language Models Reasoning with Bidirectional Chaining (2024.findings-acl)

Challenge:	Large language models (LLMs) demonstrate strong reasoning capabilities but are expensive to run at inference time, limiting their practical deployment.
Approach:	They propose Offloaded Reasoning, a modular strategy where a lightweight model generates intermediate reasoning traces that are then used by a larger model to produce the final answer.
Outcome:	The proposed approach achieves faster inferences than full large-model reasoning with minimal accuracy loss while recovering or exceeding full accuracy at substantially lower cost.

Copied to clipboard

Shuqi Liu, Bowei He, Linqi Song

Capability Decomposition for Unified Information Extraction via Hierarchical Mixture-of-Experts (2026.acl-long)

Challenge:	Existing unidirectional chaining methods suffer from low prediction accuracy and efficiency.
Approach:	They propose a bidirectional chaining method which dynamically switches to depth-first reasoning in the opposite reasoning direction when it encounters multiple branching options within the current direction.
Outcome:	The proposed method achieves sizable accuracy boots over unidirectional chaining frameworks on four challenging logical reasoning datasets.

Copied to clipboard

Jing Zhou, Peng Wang, Wenjun Ke, Jiajun Liu, Yao He

Unleashing Spatial Reasoning in Multimodal Large Language Models via Textual Representation Guided Reasoning (2026.acl-long)

Challenge:	Existing methods for IE tasks suffer from inconsistent schema representation and implicitly intermediate reasoning . UC-UIE adopts a low-rank adapted hierarchical Mixture-of-Experts adapter for UIE tasks .
Approach:	They propose a framework that decomposes IE reasoning into three universal capabilities . UC-UIE adopts a low-rank Adaptation adapter to fine-tune LLMs for IE tasks .
Outcome:	The proposed framework outperforms full-parameter tuning methods with 1.24% trainable parameters and outperformed existing methods in generalization and interpretability.

Copied to clipboard

Jiacheng Hua, Yishu Yin, Yuhang Wu, Tai Wang, Yifei Huang, Miao Liu

Semantic Text Processing

Challenge:	Existing Multimodal Large Language Models struggle with 3D spatial reasoning as they fail to construct structured abstractions of the 3D environment depicted in video inputs.
Approach:	They propose a prompting method that induces MLLMs to generate 3D representations as reasoning traces for more accurate spatial question answering.
Outcome:	Extensive experiments on VSI-Bench and OST-Bech show that TRACE improves over prior prompting strategies across a diverse range of MLLM backbones.

FinChain: A Symbolic Benchmark for Verifiable Chain-of-Thought Financial Reasoning (2026.acl-long)

Copied to clipboard

Zhuohan Xie, Daniil Orel, Rushil Thareja, Dhruv Sahnan, Hachem Madmoun, Fan Zhang, Debopriyo Banerjee, Georgi Nenkov Georgiev, Xueqing Peng, Lingfei Qian, Jimin Huang, Jinyan Su, Aaryamonvikram Singh, Rui Xing, Rania Elbadry, Chen Xu, Haonan Li, Fajri Koto, Ivan Koychev, Tanmoy Chakraborty, Yuxia Wang, Salem Lahlou, Veselin Stoyanov, Sophia Ananiadou, Preslav Nakov

Empowering Multi-step Reasoning across Languages via Program-Aided Language Models (2024.emnlp-main)

Challenge:	Existing benchmarks emphasize final numerical answers while neglecting intermediate reasoning steps.
Approach:	They propose a symbolic benchmark for verifiable Chain-of-Thought evaluation in finance . FINCHAIN spans 58 topics across 12 financial domains and three difficulty levels .
Outcome:	The proposed benchmark aims to bridge symbolic reasoning and factual verification.

Copied to clipboard

Leonardo Ranaldi, Giulia Pucci, Barry Haddow, Alexandra Birch

FAC2E: Better Understanding Large Language Model Capabilities by Dissociating Language and Cognition (2024.emnlp-main)

Challenge:	In-context learning methods elicit Large Language Models to solve tasks using provided demonstrations without parameter updates.
Approach:	They propose a method for aligning reasoning programs across languages using a double-step cross-lingual prompting mechanism.
Outcome:	The proposed method outperforms existing prompting methods and reduces interaction time.

Copied to clipboard

Xiaoqiang Wang, Lingfei Wu, Tengfei Ma, Bang Liu

Semantic Text Processing

Challenge:	Large language models (LLMs) are evaluated by overall performance on various text understanding and generation tasks.
Approach:	They propose a framework for Fine-grAined and Cognition-grounded LLMs’ Capability Evaluation that dissociates the language-related capabilities from cognition-related ones.
Outcome:	The proposed framework dissociates the language-related capabilities from cognition-related ones and breaks down the process of applying a specific capability into three sub-steps: recalling relevant knowledge, utilizing knowledge, and solving problems.

RFS-Guard: Detecting Reasoning Hallucinations via Cross-Phase Routing Focus in Large Reasoning Models (2026.acl-long)

Copied to clipboard

Zihang Liu, Fang Zhouhua, Hui Liu, Zhiwei Liu, Yong Li, Haishuai Wang

Beyond Ranking: Fine-Grained Diagnostics and Self-Improvement for MLLMs (2026.acl-long)

Challenge:	Large reasoning models (LRMs) generate intermediate reasoning traces before the final answer, yet they remain vulnerable to reasoning hallucinations such as subtle arithmetic errors.
Approach:	They propose a Routing Focus Score (RFS) that measures how strongly cross-step attention routing aligns with semantic proximity derived from hidden-state cosine similarity.
Outcome:	The proposed framework detects and localizes hallucinations without external tools or repeated sampling.

Copied to clipboard

Mingze Xu, Zijing Zhao, Qiming Peng, Houwen Peng, Han Hu, Zhanhui Kang, Yuxing Han

GRACE: Discriminator-Guided Chain-of-Thought Reasoning (2023.findings-emnlp)

Challenge:	Current paradigms rely on holistic scoring and static leaderboards to disentangle fine-grained competencies.
Approach:	They propose a framework to shift the focus from ranking to fine-grained diagnosis.
Outcome:	The proposed framework surpasses the strongest baseline by 7.92%.

Copied to clipboard

Muhammad Khalifa, Lajanugen Logeswaran, Moontae Lee, Honglak Lee, Lu Wang

Whiteboard-of-Thought: Thinking Step-by-Step Across Modalities (2024.emnlp-main)

Challenge:	Existing language models (LMs) can assign a high likelihood to incorrect steps . Existing models (LLMs), however, struggle with complex multi-step reasoning.
Approach:	They propose a stepwise decoding approach that steers the decoding process towards producing correct reasoning steps.
Outcome:	The proposed approach outperforms existing methods on math and symbolic reasoning tasks.

Copied to clipboard

Sachit Menon, Richard Zemel, Carl Vondrick

AgentV-RL: Scaling Reward Modeling with Agentic Verifier (2026.findings-acl)

Challenge:	Large language models have shown promising results in arithmetic and symbolic reasoning by expressing intermediate reasoning in text as a chain of thought, yet struggle to extend this capability to answer text queries that are easily solved by visual reasoning.
Approach:	They propose a method to unlock the visual reasoning capabilities of multimodal large language models by using a metaphorical ‘whiteboard’ to draw out reasoning steps as images and return these images back to the model for further processing.
Outcome:	The proposed method shows that it can be used on four difficult tasks that involve visual and spatial reasoning with no demonstrations or specialized modules.

Copied to clipboard

Jiazheng Zhang, Ziche Fu, Zhiheng Xi, Wenqing Jing, Mingxu Chai, Wei He, Guoqiang Zhang, Chenghao Fan, Chenxin An, Wenxiang Chen, Zhicheng Liu, Haojie Pan, Dingwei Zhu, Tao Gui, Qi Zhang, Xuanjing Huang

FLARE: Faithful Logic-Aided Reasoning and Exploration (2025.emnlp-main)

Challenge:	Existing approaches to improve LLM reasoning are limited in complex domains and lack external grounding makes verifiers unreliable on computation-intensive tasks.
Approach:	They propose a framework that transforms reward modeling into a multi-turn, tool-augmented deliberative process.
Outcome:	The proposed framework surpasses state-of-the-art ORMs by 25.2% under parallel and sequential TTS.

Copied to clipboard

Erik Arakelyan, Pasquale Minervini, Patrick Lewis, Pat Verga, Isabelle Augenstein

M3Hop-CoT: Misogynous Meme Identification with Multimodal Multi-hop Chain-of-Thought (2024.emnlp-main)

Challenge:	Modern Question Answering (QA) and Reasoning approaches with Large Language Models (LLMs) use Chain-of-Thought (CoT) prompting but struggle with ambiguous tasks.
Approach:	They propose a method that uses large language models to plan solutions and formalize queries without external solvers to generate outputs faithful to their intermediate reasoning chains.
Outcome:	The proposed method achieves SOTA results on 7 out of 9 diverse reasoning benchmarks and 3 out of 3 logic inference benchmarks while enabling measurement of reasoning faithfulness.

Copied to clipboard

Gitanjali Kumari, Kirtan Jain, Asif Ekbal

Look and Think: Efficient Multimodal Reasoning via Modality-Decoupled Compression (2026.findings-acl)

Challenge:	Recent studies have shown that Large Language Models (LLMs) neglect cultural diversity and key aspects like emotion and contextual knowledge hidden in the visual modalities.
Approach:	They propose a framework for misogynous meme identification using a multimodal multimodal prompting principle and a CLIP-based classifier.
Outcome:	The proposed framework performs well on the SemEval-2022 task 5 dataset, and is generalizable across different datasets.

Copied to clipboard

Xidi Cai, Junhao Zheng, Jingye Li, Boyuan Li, Shaowei Zhang, Qianli Ma

CMedCalc-Bench: A Fine-Grained Benchmark for Chinese Medical Calculations in LLM (2025.emnlp-main)

Challenge:	Multimodal large language models have strong performance on visual question answering benchmarks . however, their inference efficiency is severely constrained by the rapidly growing context .
Approach:	They propose a modality-decoupled compression method that enables efficient multimodal inference . they propose to evict visual tokens whenever visual grounding is unnecessary .
Outcome:	The proposed method reduces the average context length by up to 57% while maintaining comparable performance to the standard MLLM baseline.

Copied to clipboard

Yunyan Zhang, Zhihong Zhu, Xian Wu

SmartAD: Capacity-Aligned Agent Distillation for Small Language Models (2026.findings-acl)

Challenge:	Existing medical NLP benchmarks focus on qualitative reasoning and textual comprehension, but lack of fine-grained evaluation of intermediate reasoning.
Approach:	They propose a Chinese medical calculation benchmark that disentangles clinical entity extraction from numerical computation.
Outcome:	The proposed framework disentangles clinical entity extraction from numerical computation, enabling systematic diagnosis of model deficiencies.

Copied to clipboard

Guokai Tang, Feng Zhao

MavenCoder: Competitive Code Generation via Model Adaptive Planning Strategies and Multi-Perspective Verification Enhancement (2026.acl-long)

Challenge:	Large language models (LLMs) show strong reasoning and decision-making ability, but their high inference cost motivates transferring agentic skills to small language models.
Approach:	They propose a capacity-aligned agent distillation framework that trains SLMs on full reason–act–observe trajectories from a tool-using teacher.
Outcome:	The proposed framework outperforms all baselines on multi-hop QA and math benchmarks with 1.5B and 3B models.

Copied to clipboard

ZhenChun Xu, Yi Cai, Dajun Zheng, li Yuan, Mengchen Zhao, Qixiang Wang, Jiexin Wang

Programming Languages in NLP

Challenge:	Recent advances in large language models (LLMs) have significantly enhanced automated program synthesis.
Approach:	They propose a model-adaptive and verification–enhanced framework for competition-level code generation that leverages adaptive assessment aligned with the model’s capabilities to select planning strategies while providing timely feedback and correction via multi-perspective verification.
Outcome:	The proposed framework outperforms existing state-of-the-art approaches on livecodebench, humanEval+, MBPP+, and codecontests, and achieves pass@1 results exceeding 3%–40%.

Theorem Prover as a Judge for Synthetic Data Generation (2025.acl-long)

Copied to clipboard

Joshua Ong Jun Leang, Giwon Hong, Wenda Li, Shay B Cohen

VET: Verifiable Execution Tracing for Reliable Text-to-SQL Generation (2026.findings-acl)

Challenge:	Recent studies show that large language models are increasingly capable of tackling mathematical problems.
Approach:	They propose an approach that iteratively refines theorem prover formalisation to mitigate errors.
Outcome:	The proposed method increases execution rate on the Lean prover from 60% to 87%, while human annotation is replaced with theorem prover feedback.

Copied to clipboard

Dongyu Wang, Jingyu Li, Lan Zhang, null Ganggang.yu, Liang Huang

Programming Languages in NLP

Challenge:	Existing methods for text-to-SQL generation are prone to hallucinations and grounding . authors present a novel reasoning paradigm that transforms text- to-Sql from unverifiable textual rationales into step-wise executable semantics.
Approach:	They propose a reasoning paradigm that transforms text-to-SQL from unverifiable textual rationales into step-wise executable semantics.
Outcome:	The proposed reasoning paradigm transforms text-to-SQL from unverifiable textual rationales into step-wise executable semantics.

Interpretable Traces, Unexpected Outcomes: Investigating the Disconnect in Trace-Based Knowledge Distillation (2026.acl-long)

Copied to clipboard

Siddhant Bhambri, Upasana Biswas, Subbarao Kambhampati

Reasoning in the Dark: Interleaved Vision-Text Reasoning in Latent Space (2026.findings-acl)

Challenge:	Recent advances in reasoning-oriented Large Language Models have been driven by the introduction of Chain-of-Thought (CoT) traces.
Approach:	They propose to use CoT traces to guide model inference and serve as supervision signals for Knowledge Distillation to improve smaller models.
Outcome:	The proposed model is based on a rule-based problem decomposition method and is valid for both semantic correctness and interpretability to the end user.

Copied to clipboard

Chao Chen, Zhixin Ma, Yongqi Li, Yupeng Hu, Yinwei Wei, Wenjie Li, Liqiang Nie

Parallel Test-Time Scaling for Latent Reasoning Models (2026.acl-long)

Challenge:	Existing multimodal reasoning methods depend on explicit reasoning steps that require labor-intensive vision-text annotations and inherently introduce significant inference latency.
Approach:	They propose a method that integrates visual and visual information into the reasoning process to improve the performance of multimodal LLMs.
Outcome:	The proposed method achieves an average performance increase of 5.45% while achieving a speed increase of over 5 times compared to existing methods.

Copied to clipboard

Runyang You, Yongqi Li, Meng Liu, Wenjie Wang, Liqiang Nie, Wenjie Li