SemEval-2026 Task 12: Abductive Event Reasoning: Towards Real-World Event Causal Inference for Large Language Models

Understanding why real-world events occur is important for both natural language processing and practical decision-making, yet direct-cause inference remains underexplored in evidence-rich settings. To address this gap, we organized SemEval-2026 Task 12: Abductive Event Reasoning (AER).\footnote{The task data is available at https://github.com/sooo66/semeval2026-task12-dataset.git} The task asks systems to identify the most plausible direct cause of a target event from supporting evidence. We formulate AER as an evidence-grounded multiple-choice benchmark that captures key challenges of real-world causal reasoning, including distributed evidence, indirect background factors, and semantically related but non-causal distractors. The shared task attracted 122 participants and received 518 submissions. This paper presents the task formulation, dataset construction pipeline, evaluation setup, and system results. AER provides a focused benchmark for abductive reasoning over real-world events and highlights challenges for future work on causal reasoning and multi-document understanding.

GIP-RAG: An Evidence-Grounded Retrieval-Augmented Framework for Interpretable Gene Interaction and Pathway Impact Analysis

Understanding mechanistic relationships among genes and their impacts on biological pathways is essential for elucidating disease mechanisms and advancing precision medicine. Despite the availability of extensive molecular interaction and pathway data in public databases, integrating heterogeneous knowledge sources and enabling interpretable multi-step reasoning across biological networks remain challenging. We present GIP-RAG (Gene Interaction Prediction through Retrieval-Augmented Generation), a computational framework that combines biomedical knowledge graphs with large language models (LLMs) to infer and interpret gene interactions. The framework constructs a unified gene interaction knowledge graph by integrating curated data from KEGG, WikiPathways, SIGNOR, Pathway Commons, and PubChem. Given user-specified genes, a query-driven module retrieves relevant subgraphs, which are incorporated into structured prompts to guide LLM-based stepwise reasoning. This enables identification of direct and indirect regulatory relationships and generation of mechanistic explanations supported by biological evidence. Beyond pairwise interactions, GIP-RAG includes a pathway-level functional impact module that simulates propagation of gene perturbations through signaling networks and evaluates potential pathway state changes. Evaluation across diverse biological scenarios demonstrates that the framework generates consistent, interpretable, and evidence-supported insights into gene regulatory mechanisms. Overall, GIP-RAG provides a general and interpretable approach for integrating knowledge graphs with retrieval-augmented LLMs to support mechanistic reasoning in complex molecular systems.

Think While Watching: Online Streaming Segment-Level Memory for Multi-Turn Video Reasoning in Multimodal Large Language Models

Multimodal large language models (MLLMs) have shown strong performance on offline video understanding, but most are limited to offline inference or have weak online reasoning, making multi-turn interaction over continuously arriving video streams difficult. Existing streaming methods typically use an interleaved perception-generation paradigm, which prevents concurrent perception and generation and leads to early memory decay as streams grow, hurting long-range dependency modeling. We propose Think While Watching, a memory-anchored streaming video reasoning framework that preserves continuous segment-level memory during multi-turn interaction. We build a three-stage, multi-round chain-of-thought dataset and adopt a stage-matched training strategy, while enforcing strict causality through a segment-level streaming causal mask and streaming positional encoding. During inference, we introduce an efficient pipeline that overlaps watching and thinking and adaptively selects the best attention backend. Under both single-round and multi-round streaming input protocols, our method achieves strong results. Built on Qwen3-VL, it improves single-round accuracy by 2.6% on StreamingBench and by 3.79% on OVO-Bench. In the multi-round setting, it maintains performance while reducing output tokens by 56%. Code is available at: https://github.com/wl666hhh/Think_While_Watching/

MMR-Life: Piecing Together Real-life Scenes for Multimodal Multi-image Reasoning

Recent progress in the reasoning capabilities of multimodal large language models (MLLMs) has empowered them to address more complex tasks such as scientific analysis and mathematical reasoning. Despite their promise, MLLMs' reasoning abilities across different scenarios in real life remain largely unexplored and lack standardized benchmarks for evaluation. To address this gap, we introduce MMR-Life, a comprehensive benchmark designed to evaluate the diverse multimodal multi-image reasoning capabilities of MLLMs across real-life scenarios. MMR-Life consists of 2,646 multiple-choice questions based on 19,108 images primarily sourced from real-world contexts, comprehensively covering seven reasoning types: abductive, analogical, causal, deductive, inductive, spatial, and temporal. Unlike existing reasoning benchmarks, MMR-Life does not rely on domain-specific expertise but instead requires models to integrate information across multiple images and apply diverse reasoning abilities. The evaluation of 37 advanced models highlights the substantial challenge posed by MMR-Life. Even top models like GPT-5 achieve only 58% accuracy and display considerable variance in performance across reasoning types. Moreover, we analyze the reasoning paradigms of existing MLLMs, exploring how factors such as thinking length, reasoning method, and reasoning type affect their performance. In summary, MMR-Life establishes a comprehensive foundation for evaluating, analyzing, and improving the next generation of multimodal reasoning systems.

WideSeek: Advancing Wide Research via Multi-Agent Scaling

Search intelligence is evolving from Deep Research to Wide Research, a paradigm essential for retrieving and synthesizing comprehensive information under complex constraints in parallel. However, progress in this field is impeded by the lack of dedicated benchmarks and optimization methodologies for search breadth. To address these challenges, we take a deep dive into Wide Research from two perspectives: Data Pipeline and Agent Optimization. First, we produce WideSeekBench, a General Broad Information Seeking (GBIS) benchmark constructed via a rigorous multi-phase data pipeline to ensure diversity across the target information volume, logical constraints, and domains. Second, we introduce WideSeek, a dynamic hierarchical multi-agent architecture that can autonomously fork parallel sub-agents based on task requirements. Furthermore, we design a unified training framework that linearizes multi-agent trajectories and optimizes the system using end-to-end RL. Experimental results demonstrate the effectiveness of WideSeek and multi-agent RL, highlighting that scaling the number of agents is a promising direction for advancing the Wide Research paradigm.

NEAT: Neuron-Based Early Exit for Large Reasoning Models

Large Reasoning Models (LRMs) often suffer from \emph{overthinking}, a phenomenon in which redundant reasoning steps are generated after a correct solution has already been reached. Existing early reasoning exit methods primarily rely on output-level heuristics or trained probing models to skip redundant reasoning steps, thereby mitigating overthinking. However, these approaches typically require additional rollout computation or externally labeled datasets. In this paper, we propose \textbf{NEAT}, a \textbf{N}euron-based \textbf{E}arly re\textbf{A}soning exi\textbf{T} framework that monitors neuron-level activation dynamics to enable training-free early exits, without introducing additional test-time computation. NEAT identifies exit-associated neurons and tracks their activation patterns during reasoning to dynamically trigger early exit or suppress reflection, thereby reducing unnecessary reasoning while preserving solution quality. Experiments on four reasoning benchmarks across six models with different scales and architectures show that, for each model, NEAT achieves an average token reduction of 22\% to 28\% when averaged over the four benchmarks, while maintaining accuracy.

A Dynamic Parametric Simulator for Fetal Heart Sounds

Research on fetal phonocardiogram (fPCG) is challenged by the limited number of abdominal recordings, substantial maternal interference, and marked transmissioninduced signal attenuation that complicate reproducible benchmarking. We present a reproducible dynamic parametric simulator that generates long abdominal fPCG sequences by combining cycle-level fetal S1/S2 event synthesis with a convolutional transmission module and configurable interference and background noise. Model parameters are calibrated cyclewise from real abdominal recordings to capture beat-to-beat variability and to define data-driven admissible ranges for controllable synthesis. The generated signals are validated against real recordings in terms of envelope-based temporal structure and frequency-domain characteristics. The simulator is released as open software to support rapid, reproducible evaluation of fPCG processing methods under controlled acquisition conditions.

Learning How to Remember: A Meta-Cognitive Management Method for Structured and Transferable Agent Memory

Large language model (LLM) agents increasingly rely on accumulated memory to solve long-horizon decision-making tasks. However, most existing approaches store memory in fixed representations and reuse it at a single or implicit level of abstraction, which limits generalization and often leads to negative transfer when distribution shift. This paper proposes the Meta-Cognitive Memory Abstraction method (MCMA), which treats memory abstraction as a learnable cognitive skill rather than a fixed design choice. MCMA decouples task execution from memory management by combining a frozen task model with a learned memory copilot. The memory copilot is trained using direct preference optimization, it determines how memories should be structured, abstracted, and reused. Memories are further organized into a hierarchy of abstraction levels, enabling selective reuse based on task similarity. When no memory is transferable, MCMA transfers the ability to abstract and manage memory by transferring the memory copilot. Experiments on ALFWorld, ScienceWorld, and BabyAI demonstrate substantial improvements in performance, out-of-distribution generalization, and cross-task transfer over several baselines.

Bottom-up Policy Optimization: Your Language Model Policy Secretly Contains Internal Policies

Existing reinforcement learning (RL) approaches treat large language models (LLMs) as a single unified policy, overlooking their internal mechanisms. Understanding how policy evolves across layers and modules is therefore crucial for enabling more targeted optimization and raveling out complex reasoning mechanisms. In this paper, we decompose the language model policy by leveraging the intrinsic split of the Transformer residual stream and the equivalence between the composition of hidden states with the unembedding matrix and the resulting samplable policy. This decomposition reveals Internal Layer Policies, corresponding to contributions from individual layers, and Internal Modular Policies, which align with the self-attention and feed-forward network (FFN) components within each layer. By analyzing the entropy of internal policy, we find that: (a) Early layers keep high entropy for exploration, top layers converge to near-zero entropy for refinement, with convergence patterns varying across model series. (b) LLama's prediction space rapidly converges in the final layer, whereas Qwen-series models, especially Qwen3, exhibit a more human-like, progressively structured reasoning pattern. Motivated by these findings, we propose Bottom-up Policy Optimization (BuPO), a novel RL paradigm that directly optimizes the internal layer policy during early training. By aligning training objective at lower layer, BuPO reconstructs foundational reasoning capabilities and achieves superior performance. Extensive experiments on complex reasoning benchmarks demonstrates the effectiveness of our method. Our code is available at https://github.com/Trae1ounG/BuPO.

OPBO: Order-Preserving Bayesian Optimization

Bayesian optimization is an effective method for solving expensive black-box optimization problems. Most existing methods use Gaussian processes (GP) as the surrogate model for approximating the black-box objective function, it is well-known that it can fail in high-dimensional space (e.g., dimension over 500). We argue that the reliance of GP on precise numerical fitting is fundamentally ill-suited in high-dimensional space, where it leads to prohibitive computational complexity. In order to address this, we propose a simple order-preserving Bayesian optimization (OPBO) method, where the surrogate model preserves the order, instead of the value, of the black-box objective function. Then we can use a simple but effective OP neural network (NN) to replace GP as the surrogate model. Moreover, instead of searching for the best solution from the acquisition model, we select good-enough solutions in the ordinal set to reduce computational cost. The experimental results show that for high-dimensional (over 500) black-box optimization problems, the proposed OPBO significantly outperforms traditional BO methods based on regression NN and GP. The source code is available at https://github.com/pengwei222/OPBO.