Medical language models face critical barriers to real-world clinical reasoning applications. However, mainstream efforts, which fall short in task coverage, lack fine-grained supervision for intermediate reasoning steps, and rely on proprietary systems, are still far from a versatile, credible and efficient language model for clinical reasoning usage. To this end, we propose MedS3, a self-evolving framework that imparts robust reasoning capabilities to small, deployable models. Starting with 8,000 curated instances sampled via a curriculum strategy across five medical domains and 16 datasets, we use a small base policy model to conduct Monte Carlo Tree Search (MCTS) for constructing rule-verifiable reasoning trajectories. Self-explored reasoning trajectories ranked by node values are used to bootstrap the policy model via reinforcement fine-tuning and preference learning. Moreover, we introduce a soft dual process reward model that incorporates value dynamics: steps that degrade node value are penalized, enabling fine-grained identification of reasoning errors even when the final answer is correct. Experiments on eleven benchmarks show that MedS3 outperforms the previous state-of-the-art medical model by +6.45 accuracy points and surpasses 32B-scale general-purpose reasoning models by +8.57 points. Additional empirical analysis further demonstrates that MedS3 achieves robust and faithful reasoning behavior.

Hongcheng Liu, Yuhao Wang, Zhe Chen, Pingjie Wang, Zhiyuan Zhu, Yixuan Hou, Yanfeng Wang, Yu Wang. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Large language models hold promise for addressing medical challenges, such as medical diagnosis reasoning, research knowledge acquisition, clinical decision-making, and consumer health inquiry support. However, they often generate hallucinations due to limited medical knowledge. Incorporating external knowledge is therefore critical, which necessitates multi-source knowledge acquisition. We address this challenge by framing it as a source planning problem, which is to formulate context-appropriate queries tailored to the attributes of diverse sources. Existing approaches either overlook source planning or fail to achieve it effectively due to misalignment between the model’s expectation of the sources and their actual content. To bridge this gap, we present MedOmniKB, a repository comprising multigenre and multi-structured medical knowledge sources. Leveraging these sources, we propose the Source Planning Optimisation method, which enhances multi-source utilisation. Our approach involves enabling an expert model to explore and evaluate potential plans while training a smaller model to learn source alignment. Experimental results demonstrate that our method substantially improves multi-source planning performance, enabling the optimised small model to achieve state-of-the-art results in leveraging diverse medical knowledge sources.

Large language models (LLMs) are trained on extensive historical corpora, but their ability to understand time and maintain temporal awareness of time-evolving factual knowledge remains limited. Previous studies often neglect the critical aspect of utilizing knowledge from various sources. To address this gap, we introduce EvolveBench, a comprehensive benchmark that evaluates temporal competence along five key dimensions: Cognition, which examines the ability to recall and contextualize historical facts. Awareness, which tests LLMs’ awareness of temporal misalignment between external inputs and the temporal context of a query. Trustworthiness, which assesses whether models can identify and appropriately refuse queries based on invalid timestamps. Understanding, which focuses on interpreting both explicit dates and implicit historical markers. Finally, reasoning evaluates the capacity to analyze temporal relationships and draw accurate inferences. Evaluating 15 widely used LLMs on EvolveBench shows that GPT-4o achieves the highest average EM score of 79.36, while the open-source Llama3.1-70B demonstrates notable strength in handling temporally misaligned contexts with an average score of 72.47. Despite these advances, all models still struggle with handling temporal misaligned context. Our code and dataset are available at https://github.com/zzysjtuiwct/EvolveBench.

The reliability of large language models remains a critical challenge, particularly due to their susceptibility to hallucinations and factual inaccuracies during text generation.Existing solutions either underutilize models' selfcorrection with preemptive strategies or use costly post-hoc verification.To further explore the potential of real-time self-verification and correction, we present Dynamic Self-Verify Decoding (DSVD), a novel decoding framework that enhances generation reliability through real-time hallucination detection and efficient error correction.DSVD integrates two key components: (1) parallel self-verification architecture for continuous quality assessment, (2) dynamic rollback mechanism for targeted error recovery.Extensive experiments across five benchmarks demonstrate DSVD's effectiveness, achieving significant improvement in truthfulness (Quesetion-Answering) and factual accuracy (FActScore).Results show the DSVD can be further incorporated with existing faithful decoding methods to achieve stronger performance.Our work establishes that real-time self-verification during generation offers a viable path toward more faithful language models without sacrificing practical deployability.

When performing reasoning tasks with userspecific requirements, such as strict output formats, large language models (LLMs) often prioritize reasoning over adherence to detailed instructions.Fine-tuning LLMs on supervised datasets to address this is impractical due to high computational costs and limited parameter access.To tackle this, we propose DICE, a lightweight framework that guides small language models (SLMs) to refine LLMs' outputs through chain-of-thought (CoT) correction.DICE decouples the process by first prompting LLMs to generate natural language responses, then using trained SLMs to analyze and refine these outputs to meet structured output specifications.This framework preserves LLMs' broad knowledge and reasoning capabilities while ensuring the outputs conform to user demands.Specifically, DICE first constructs structured CoT adaptation datasets via a two-stage method and subsequently applies a dual-tuning strategy to fine-tune SLMs for generating structured outputs in an analyze-thenanswer pattern. 1 Experiments demonstrate that DICE improves the average format accuracy and content correctness of LLM outputs by 35.4% and 29.4%, respectively, achieving stateof-the-art (SOTA) performance over other competitive baselines.

Clinical diagnosis is a highly specialized discipline requiring both domain expertise and strict adherence to rigorous guidelines. While current AI-driven medical research predominantly focuses on knowledge graphs or natural text pretraining paradigms to incorporate medical knowledge, these approaches primarily rely on implicitly encoded knowledge within model parameters, neglecting task-specific knowledge required by diverse downstream tasks. To address this limitation, we propose **R**etrieval-**A**ugmented **D**iagnosis (RAD), a novel framework that explicitly injects external knowledge into multimodal models directly on downstream tasks. Specifically, RAD operates through three key mechanisms: retrieval and refinement of disease-centered knowledge from multiple medical sources, a guideline-enhanced contrastive loss that constrains the latent distance between multi-modal features and guideline knowledge, and the dual transformer decoder that employs guidelines as queries to steer cross-modal fusion, aligning the models with clinical diagnostic workflows from guideline acquisition to feature extraction and decision-making. Moreover, recognizing the lack of quantitative evaluation of interpretability for multimodal diagnostic models, we introduce a set of criteria to assess the interpretability from both image and text perspectives. Extensive evaluations across four datasets with different anatomies demonstrate RAD's generalizability, achieving state-of-the-art performance. Furthermore, RAD enables the model to concentrate more precisely on abnormal regions and critical indicators, ensuring evidence-based, trustworthy diagnosis. Our code is available at https://github.com/tdlhl/RAD.

Zhe Chen, Heyang Liu, Wenyi Yu, Guangzhi Sun, Hongcheng Liu, Ji Wu, Chao Zhang, Yu Wang, Yanfeng Wang. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2024.

Generating dialogue grounded in videos requires a high level of understanding and reasoning about the visual scenes in the videos. However, existing large visual-language models are not effective due to their latent features and decoder-only structure, especially with respect to spatio-temporal relationship reasoning. In this paper, we propose a novel approach named MSG-BART, which enhances the integration of video information by incorporating a multi-granularity spatio-temporal scene graph into an encoder-decoder pre-trained language model. Specifically, we integrate the global and local scene graph into the encoder and decoder, respectively, to improve both overall perception and target reasoning capability. To further improve the information selection capability, we propose a multi-pointer network to facilitate selection between text and video. Extensive experiments are conducted on three video-grounded dialogue benchmarks, which show the significant superiority of the proposed MSG-BART compared to a range of state-of-the-art approaches.

In recent years, Audio Visual Scene-Aware Dialog (AVSD) has been an active research task in the multimodal dialogue community and has also been a core part of the Dialog System Technology Challenge (DSTC). This task is an extension of conventional visual question answering, where video-relevant answers must be generated taking into account multimodal contextual information from previous dialogue rounds. Despite recent advances in the AVSD task, there are still two major challenges in developing such a system: how to model the multimodal contextual information of multiple rounds of dialogues and how to integrate audio-visual information into the generation of textual responses. To tackle these two challenges, in this paper we propose a novel model, named DialogMCF, which constructs a multimodal context flow model to generate responses that are relevant to video scenes. This proposed context flow modeling can track the dynamics of the topic information across multiple rounds of dialogue history. To achieve an effective fusion of multimodal information, we propose an audio-visual memory network with cross-modality aligned features to model long multimodal dialogue context, and thus enhance the flow modeling. Furthermore, we make attempts to improve the performance of the proposed DialogMCF model with manual descriptions and explore the incorporation of temporal reasoning information. Extensive experiments on the DSTC AVSD datasets show that, compared to a range of baseline methods, the proposed method can yield state-of-art dialogue generation performance on most metrics when integrating the video descriptions.