Papers by Zhenwen Liang
Verified Critical Step Optimization for LLM Agents (2026.findings-acl)
Copied to clipboard
| Challenge: | Critical Step Optimization (CSO) focuses preference learning on verified critical steps where alternative actions demonstrably flip task outcomes from failure to success. |
| Approach: | They propose a method which focuses preference learning on verified critical steps where alternative actions demonstrably flip task outcomes from failure to success. |
| Outcome: | The proposed method outperforms the existing methods on GAIA-Text-103 and XBench-DeepSearch while requiring supervision at only 16% of trajectory steps. |
ArMATH: a Dataset for Solving Arabic Math Word Problems (2022.lrec-1)
Copied to clipboard
| Challenge: | This paper is the first to use deep learning methods to solve Arabic MWPs . it is also the first study to use transfer learning to solve MWp across different languages . |
| Approach: | They contribute to the first large-scale dataset for Arabic Math Word Problems . they use deep learning methods to solve Arabic MWPs and a transfer learning model to promote performance . |
| Outcome: | The proposed model improves Arabic MWP solvers by 3% over the existing model. |
UniMath: A Foundational and Multimodal Mathematical Reasoner (2023.emnlp-main)
Copied to clipboard
| Challenge: | Existing methods for interpreting and processing diverse mathematical modalities are limited . existing systems are limited in interpreting complex mathematical tasks and implementing them in a multimodal manner. |
| Approach: | They propose a multimodal mathematical reasoning system that utilizes a fine-tuned T5 model augmented with a variational autoencoder (VAE)-based image tokenizer. |
| Outcome: | The proposed model achieves state-of-the-art performance on SVAMP, GeoQA, and TableMWP datasets and is generalized on two additional datasets. |
SceMQA: A Scientific College Entrance Level Multimodal Question Answering Benchmark (2024.acl-short)
Copied to clipboard
Zhenwen Liang, Kehan Guo, Gang Liu, Taicheng Guo, Yujun Zhou, Tianyu Yang, Jiajun Jiao, Renjie Pi, Jipeng Zhang, Xiangliang Zhang
| Challenge: | SceMQA focuses on core science subjects including Mathematics, Physics, Chemistry, and Biology. |
| Approach: | They propose to use SceMQA to evaluate multimodal question answering at college entrance level. |
| Outcome: | The proposed model provides specific knowledge points for each problem and detailed explanations for each answer. |
MWP-BERT: Numeracy-Augmented Pre-training for Math Word Problem Solving (2022.findings-naacl)
Copied to clipboard
| Challenge: | Existing work on math word problem solvers replace real numbers with symbolic placeholders to focus on logic reasoning. |
| Approach: | They propose to inject numerical properties into symbolic placeholders with contextualized representation learning schema to solve number representation dilemma. |
| Outcome: | The proposed model can solve MWP problems on English and Chinese benchmarks. |
Too Correct to Learn: Reinforcement Learning on Saturated Reasoning Data (2026.acl-short)
Copied to clipboard
| Challenge: | Strong base models saturate benchmarks, resulting in weaker performance, a paradox . a new approach to Reinforcement Learning (RL) is needed to improve performance . |
| Approach: | They propose a method that uses constrained uniform top-k sampling to flatten the local optimization landscape by sampling uniformly from constrained high-confidence candidates. |
| Outcome: | Experiments show that the proposed approach prevents policy degeneration and boosts out-of-domain generalization. |
Your Reasoning Model is Secretly a Reward Model - Optimization-Free Verification from Experience (2026.acl-long)
Copied to clipboard
| Challenge: | Existing verifiers operate on the surface text or on confidence proxies derived from token probabilities, which can be brittle. |
| Approach: | They propose a training-free, non-parametric verifier that summarizes each reasoning trace by an activation delta and compares it to two class centroids computed from labeled experience. |
| Outcome: | The proposed model improves selection and reranking on large and less-calibrated models. |
Let GPT be a Math Tutor: Teaching Math Word Problem Solvers with Customized Exercise Generation (2023.emnlp-main)
Copied to clipboard
| Challenge: | Existing approaches for distilling large language models into smaller, more efficient student models are based on educational science principles such as knowledge tracing and personalized learning. |
| Approach: | They propose a method for distilling large language models into smaller, more efficient student models that are aligned with educational science principles such as knowledge tracing and personalized learning. |
| Outcome: | The proposed approach outperforms LLMs on three benchmarks while employing significantly fewer parameters. |
Data-Efficient Language Shaped Few-shot Image Classification (2021.findings-emnlp)
Copied to clipboard
| Challenge: | Existing studies have shown that language is helpful guider for image understanding by neural networks. |
| Approach: | They propose a language-shaped learning method that makes the best use of the few-shot images and the language available only in training. |
| Outcome: | The proposed method outperforms state-of-the-art methods on a few-shot dataset with limited training data. |
Save the Good Prefix: Precise Error Penalization via Process-Supervised RL to Enhance LLM Reasoning (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing reinforcement learning methods rely on sparse outcome rewards, which fail to credit correct intermediate steps in partially successful solutions. |
| Approach: | They propose a process reward model that rewards correct steps only when they detect errors . they propose VPPO, which rewards the correct prefix and an erroneous suffix . |
| Outcome: | a new approach outperforms sparse-reward RL and prior PRM-guided baselines on Pass@1 and Pass@K . a process reward model (PRM) outperformed sparser-rebound RL on multiple reasoning benchmarks . |
Compositional Mathematical Encoding for Math Word Problems (2023.findings-acl)
Copied to clipboard
| Challenge: | Existing MWP encoders work in a unimodal setting and map problem description to latent representation, then for decoding. |
| Approach: | They propose a Compositional Math Word Problem Solver which maps problem description to latent representation and decodes it in an interactive way. |
| Outcome: | Extensive experiments show that the proposed model outperforms state-of-the-art models on public benchmarks. |
Analogical Math Word Problems Solving with Enhanced Problem-Solution Association (2022.emnlp-main)
Copied to clipboard
| Challenge: | Analogical reasoning has long been used in mathematical education, as it enables students to apply common relational structures of mathematical situations to solve new problems. |
| Approach: | They propose to leverage analogical MWPs to advance the solver’s generalization ability across different kinds of MWps. |
| Outcome: | The proposed model has a stronger generalization ability in solving difficult MWPs due to the analogical learning from easy MWPS. |
MinT: Boosting Generalization in Mathematical Reasoning via Multi-view Fine-tuning (2024.lrec-main)
Copied to clipboard
| Challenge: | Existing methods focus on specializing LMs in mathematical reasoning and rely on knowledge distillation. |
| Approach: | They propose a multi-view fine-tuning method that exploits existing mathematical problem datasets with diverse annotation styles. |
| Outcome: | The proposed method outperforms existing methods that rely heavily on LLM teachers . it grants models generalization ability across views and datasets, and the capability to learn from inaccurate or incomplete data. |
EconProver: Towards More Economical Test-Time Scaling for Automated Theorem Proving (2026.acl-long)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have recently advanced the field of Automated Theorem Proving (ATP) Existing cost analyses regulate only the number of sampling passes, ignoring the substantial disparities in sampling costs. |
| Approach: | They propose to integrate two complementary methods into a unified EconRL pipeline to increase pass rates under constrained sampling passes. |
| Outcome: | The proposed method reduces token usage and sample passes while maintaining the original performance. |
Dissecting Logical Reasoning in LLMs: A Fine-Grained Evaluation and Supervision Study (2025.findings-emnlp)
Copied to clipboard
Yujun Zhou, Jiayi Ye, Zipeng Ling, Yufei Han, Yue Huang, Haomin Zhuang, Zhenwen Liang, Kehan Guo, Taicheng Guo, Xiangqi Wang, Xiangliang Zhang
| Challenge: | Existing benchmarks that rely on final-answer accuracy fail to capture the quality of the reasoning process. |
| Approach: | They propose a fine-grained evaluation framework that assesses logical reasoning across three dimensions: overall accuracy, stepwise soundness, and representation-level probing. |
| Outcome: | The proposed framework assesses logical reasoning across three dimensions: overall accuracy, stepwise soundness, and representation-level probing. |
A Survey of Multimodal Mathematical Reasoning: From Perception, Alignment to Reasoning (2026.acl-long)
Copied to clipboard
Tianyu Yang, Sihong Wu, Yilun Zhao, Zhenwen Liang, Lisen Dai, Chen Zhao, Minhao Cheng, Arman Cohan, Xiangliang Zhang
| Challenge: | Multimodal mathematical Reasoning (MMR) has attracted increasing attention for its ability to solve mathematical problems involving both textual and visual modalities. |
| Approach: | They review the theoretical frameworks of multimodal reasoning and examine the challenges they face in visual math tasks. |
| Outcome: | The proposed models can solve problems involving both textual and visual modalities. |
Learn Beyond The Answer: Training Language Models with Reflection for Mathematical Reasoning (2024.emnlp-main)
Copied to clipboard
| Challenge: | Existing studies focus on *broadening* the training set with data augmentation techniques to maximize such benefits. |
| Approach: | They propose a method that embeds problem reflection into each training instance. |
| Outcome: | The proposed method enhances performance in standard and complex scenarios that require reflective thinking. |
SaSR-Net: Source-Aware Semantic Representation Network for Enhancing Audio-Visual Question Answering (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Existing AVQA methods often fail to link sound-producing objects in the video with the audio-visual information. |
| Approach: | They introduce a source-aware semantic representation network for AVQA . they use source-wise learnable tokens to capture and align audio-visual elements with the question . |
| Outcome: | The proposed model outperforms state-of-the-art models on the Music-AVQA and AVQA-Yang datasets. |
Defending Jailbreak Prompts via In-Context Adversarial Game (2024.emnlp-main)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) demonstrate remarkable capabilities across diverse applications, but concerns regarding their security persist. |
| Approach: | They propose an adversarial game that leverages agent learning to extend knowledge to defend against jailbreaks. |
| Outcome: | The proposed game shows that LLMs safeguarded by ICAG exhibit significantly reduced jailbreak success rates across various attack scenarios. |