Boosting MLLM Reasoning with Text-Debiased Hint-GRPO

MLLM reasoning has drawn widespread research for its excellent problem-solving capability. Current reasoning methods fall into two types: PRM, which supervises the intermediate reasoning steps, and ORM, which supervises the final results. Recently, DeepSeek-R1 has challenged the traditional view that PRM outperforms ORM, which demonstrates strong generalization performance using an ORM method (i.e., GRPO). However, current MLLM's GRPO algorithms still struggle to handle challenging and complex multimodal reasoning tasks (e.g., mathematical reasoning). In this work, we reveal two problems that impede the performance of GRPO on the MLLM: Low data utilization and Text-bias. Low data utilization refers to that GRPO cannot acquire positive rewards to update the MLLM on difficult samples, and text-bias is a phenomenon that the MLLM bypasses image condition and solely relies on text condition for generation after GRPO training. To tackle these problems, this work proposes Hint-GRPO that improves data utilization by adaptively providing hints for samples of varying difficulty, and text-bias calibration that mitigates text-bias by calibrating the token prediction logits with image condition in test-time. Experiment results on three base MLLMs across eleven datasets demonstrate that our proposed methods advance the reasoning capability of original MLLM by a large margin, exhibiting superior performance to existing MLLM reasoning methods. Our code is available at https://github.com/hqhQAQ/Hint-GRPO.

Less is More: Adaptive Program Repair with Bug Localization and Preference Learning

Automated Program Repair (APR) is a task to automatically generate patches for the buggy code. However, most research focuses on generating correct patches while ignoring the consistency between the fixed code and the original buggy code. How to conduct adaptive bug fixing and generate patches with minimal modifications have seldom been investigated. To bridge this gap, we first introduce a novel task, namely AdaPR (Adaptive Program Repair). We then propose a two-stage approach AdaPatcher (Adaptive Patch Generator) to enhance program repair while maintaining the consistency. In the first stage, we utilize a Bug Locator with self-debug learning to accurately pinpoint bug locations. In the second stage, we train a Program Modifier to ensure consistency between the post-modified fixed code and the pre-modified buggy code. The Program Modifier is enhanced with a location-aware repair learning strategy to generate patches based on identified buggy lines, a hybrid training strategy for selective reference and an adaptive preference learning to prioritize fewer changes. The experimental results show that our approach outperforms a set of baselines by a large margin, validating the effectiveness of our two-stage framework for the newly proposed AdaPR task.

Disentangled Knowledge Tracing for Alleviating Cognitive Bias

In the realm of Intelligent Tutoring System (ITS), the accurate assessment of students' knowledge states through Knowledge Tracing (KT) is crucial for personalized learning. However, due to data bias, $\textit{i.e.}$, the unbalanced distribution of question groups ($\textit{e.g.}$, concepts), conventional KT models are plagued by cognitive bias, which tends to result in cognitive underload for overperformers and cognitive overload for underperformers. More seriously, this bias is amplified with the exercise recommendations by ITS. After delving into the causal relations in the KT models, we identify the main cause as the confounder effect of students' historical correct rate distribution over question groups on the student representation and prediction score. Towards this end, we propose a Disentangled Knowledge Tracing (DisKT) model, which separately models students' familiar and unfamiliar abilities based on causal effects and eliminates the impact of the confounder in student representation within the model. Additionally, to shield the contradictory psychology ($\textit{e.g.}$, guessing and mistaking) in the students' biased data, DisKT introduces a contradiction attention mechanism. Furthermore, DisKT enhances the interpretability of the model predictions by integrating a variant of Item Response Theory. Experimental results on 11 benchmarks and 3 synthesized datasets with different bias strengths demonstrate that DisKT significantly alleviates cognitive bias and outperforms 16 baselines in evaluation accuracy.

Knowledge is Power: Harnessing Large Language Models for Enhanced Cognitive Diagnosis

Cognitive Diagnosis Models (CDMs) are designed to assess students' cognitive states by analyzing their performance across a series of exercises. However, existing CDMs often struggle with diagnosing infrequent students and exercises due to a lack of rich prior knowledge. With the advancement in large language models (LLMs), which possess extensive domain knowledge, their integration into cognitive diagnosis presents a promising opportunity. Despite this potential, integrating LLMs with CDMs poses significant challenges. LLMs are not well-suited for capturing the fine-grained collaborative interactions between students and exercises, and the disparity between the semantic space of LLMs and the behavioral space of CDMs hinders effective integration. To address these issues, we propose a novel Knowledge-enhanced Cognitive Diagnosis (KCD) framework, which is a model-agnostic framework utilizing LLMs to enhance CDMs and compatible with various CDM architectures. The KCD framework operates in two stages: LLM Diagnosis and Cognitive Level Alignment. In the LLM Diagnosis stage, both students and exercises are diagnosed to achieve comprehensive and detailed modeling. In the Cognitive Level Alignment stage, we bridge the gap between the CDMs' behavioral space and the LLMs' semantic space using contrastive learning and mask-reconstruction approaches. Experiments on several real-world datasets demonstrate the effectiveness of our proposed framework.

sDREAMER: Self-distilled Mixture-of-Modality-Experts Transformer for Automatic Sleep Staging

Automatic sleep staging based on electroencephalography (EEG) and electromyography (EMG) signals is an important aspect of sleep-related research. Current sleep staging methods suffer from two major drawbacks. First, there are limited information interactions between modalities in the existing methods. Second, current methods do not develop unified models that can handle different sources of input. To address these issues, we propose a novel sleep stage scoring model sDREAMER, which emphasizes cross-modality interaction and per-channel performance. Specifically, we develop a mixture-of-modality-expert (MoME) model with three pathways for EEG, EMG, and mixed signals with partially shared weights. We further propose a self-distillation training scheme for further information interaction across modalities. Our model is trained with multi-channel inputs and can make classifications on either single-channel or multi-channel inputs. Experiments demonstrate that our model outperforms the existing transformer-based sleep scoring methods for multi-channel inference. For single-channel inference, our model also outperforms the transformer-based models trained with single-channel signals.

Revisiting Applicable and Comprehensive Knowledge Tracing in Large-Scale Data

Knowledge Tracing (KT) is a fundamental component of Intelligent Tutoring Systems (ITS), enabling the modeling of students' knowledge states to predict future performance. The introduction of Deep Knowledge Tracing (DKT), the first deep learning-based KT (DLKT) model, has brought significant advantages in terms of applicability and comprehensiveness. However, recent DLKT models, such as Attentive Knowledge Tracing (AKT), have often prioritized predictive performance at the expense of these benefits. While deep sequential models like DKT have shown potential, they face challenges related to parallel computing, storage decision modification, and limited storage capacity. To address these limitations, we propose DKT2, a novel KT model that leverages the recently developed xLSTM architecture. DKT2 enhances input representation using the Rasch model and incorporates Item Response Theory (IRT) for interpretability, allowing for the decomposition of learned knowledge into familiar and unfamiliar knowledge. By integrating this knowledge with predicted questions, DKT2 generates comprehensive knowledge states. Extensive experiments conducted across three large-scale datasets demonstrate that DKT2 consistently outperforms 17 baseline models in various prediction tasks, underscoring its potential for real-world educational applications. This work bridges the gap between theoretical advancements and practical implementation in KT.Our code and datasets will be available at https://github.com/codebase-2025/DKT2.

Ouroboros-Diffusion: Exploring Consistent Content Generation in Tuning-free Long Video Diffusion

The first-in-first-out (FIFO) video diffusion, built on a pre-trained text-to-video model, has recently emerged as an effective approach for tuning-free long video generation. This technique maintains a queue of video frames with progressively increasing noise, continuously producing clean frames at the queue's head while Gaussian noise is enqueued at the tail. However, FIFO-Diffusion often struggles to keep long-range temporal consistency in the generated videos due to the lack of correspondence modeling across frames. In this paper, we propose Ouroboros-Diffusion, a novel video denoising framework designed to enhance structural and content (subject) consistency, enabling the generation of consistent videos of arbitrary length. Specifically, we introduce a new latent sampling technique at the queue tail to improve structural consistency, ensuring perceptually smooth transitions among frames. To enhance subject consistency, we devise a Subject-Aware Cross-Frame Attention (SACFA) mechanism, which aligns subjects across frames within short segments to achieve better visual coherence. Furthermore, we introduce self-recurrent guidance. This technique leverages information from all previous cleaner frames at the front of the queue to guide the denoising of noisier frames at the end, fostering rich and contextual global information interaction. Extensive experiments of long video generation on the VBench benchmark demonstrate the superiority of our Ouroboros-Diffusion, particularly in terms of subject consistency, motion smoothness, and temporal consistency.

Large Language Models for Bioinformatics

With the rapid advancements in large language model (LLM) technology and the emergence of bioinformatics-specific language models (BioLMs), there is a growing need for a comprehensive analysis of the current landscape, computational characteristics, and diverse applications. This survey aims to address this need by providing a thorough review of BioLMs, focusing on their evolution, classification, and distinguishing features, alongside a detailed examination of training methodologies, datasets, and evaluation frameworks. We explore the wide-ranging applications of BioLMs in critical areas such as disease diagnosis, drug discovery, and vaccine development, highlighting their impact and transformative potential in bioinformatics. We identify key challenges and limitations inherent in BioLMs, including data privacy and security concerns, interpretability issues, biases in training data and model outputs, and domain adaptation complexities. Finally, we highlight emerging trends and future directions, offering valuable insights to guide researchers and clinicians toward advancing BioLMs for increasingly sophisticated biological and clinical applications.

Fine-Grained Guidance for Retrievers: Leveraging LLMs' Feedback in Retrieval-Augmented Generation

Retrieval-Augmented Generation (RAG) has proven to be an effective method for mitigating hallucination issues inherent in large language models (LLMs). Previous approaches typically train retrievers based on semantic similarity, lacking optimization for RAG. More recent works have proposed aligning retrievers with the preference signals of LLMs. However, these preference signals are often difficult for dense retrievers, which typically have weaker language capabilities, to understand and learn effectively. Drawing inspiration from pedagogical theories like Guided Discovery Learning, we propose a novel framework, FiGRet (Fine-grained Guidance for Retrievers), which leverages the language capabilities of LLMs to construct examples from a more granular, information-centric perspective to guide the learning of retrievers. Specifically, our method utilizes LLMs to construct easy-to-understand examples from samples where the retriever performs poorly, focusing on three learning objectives highly relevant to the RAG scenario: relevance, comprehensiveness, and purity. These examples serve as scaffolding to ultimately align the retriever with the LLM's preferences. Furthermore, we employ a dual curriculum learning strategy and leverage the reciprocal feedback between LLM and retriever to further enhance the performance of the RAG system. A series of experiments demonstrate that our proposed framework enhances the performance of RAG systems equipped with different retrievers and is applicable to various LLMs.

Classifier-guided Gradient Modulation for Enhanced Multimodal Learning

Multimodal learning has developed very fast in recent years. However, during the multimodal training process, the model tends to rely on only one modality based on which it could learn faster, thus leading to inadequate use of other modalities. Existing methods to balance the training process always have some limitations on the loss functions, optimizers and the number of modalities and only consider modulating the magnitude of the gradients while ignoring the directions of the gradients. To solve these problems, in this paper, we present a novel method to balance multimodal learning with Classifier-Guided Gradient Modulation (CGGM), considering both the magnitude and directions of the gradients. We conduct extensive experiments on four multimodal datasets: UPMC-Food 101, CMU-MOSI, IEMOCAP and BraTS 2021, covering classification, regression and segmentation tasks. The results show that CGGM outperforms all the baselines and other state-of-the-art methods consistently, demonstrating its effectiveness and versatility. Our code is available at https://github.com/zrguo/CGGM.