Dian Li

Shusheng Yang, Yixiao Ge, Kun Yi et al. · tencent-ai

Both masked image modeling (MIM) and natural language supervision have facilitated the progress of transferable visual pre-training. In this work, we seek the synergy between two paradigms and study the emerging properties when MIM meets natural language supervision. To this end, we present a novel masked visual Reconstruction In Language semantic Space (RILS) pre-training framework, in which sentence representations, encoded by the text encoder, serve as prototypes to transform the vision-only signals into patch-sentence probabilities as semantically meaningful MIM reconstruction targets. The vision models can therefore capture useful components with structured information by predicting proper semantic of masked tokens. Better visual representations could, in turn, improve the text encoder via the image-text alignment objective, which is essential for the effective MIM target transformation. Extensive experimental results demonstrate that our method not only enjoys the best of previous MIM and CLIP but also achieves further improvements on various tasks due to their mutual benefits. RILS exhibits advanced transferability on downstream classification, detection, and segmentation, especially for low-shot regimes. Code will be made available at https://github.com/hustvl/RILS.

Chen Li, Yixiao Ge, Dian Li et al. · tencent-ai

Instruction tuning is a crucial supervised training phase in Large Language Models (LLMs), aiming to enhance the LLM's ability to generalize instruction execution and adapt to user preferences. With the increasing integration of multi-modal data into LLMs, there is growing interest in Vision-Language Instruction Tuning (VLIT), which presents more complex characteristics compared to pure text instruction tuning. In this paper, we systematically review the latest VLIT settings and corresponding datasets in multi-modal LLMs and provide insights into the intrinsic motivations behind their design. For the first time, we offer a detailed multi-perspective categorization for existing VLIT datasets and identify the characteristics that high-quality VLIT data should possess. By incorporating these characteristics as guiding principles into the existing VLIT data construction process, we conduct extensive experiments and verify their positive impact on the performance of tuned multi-modal LLMs. Furthermore, we discuss the current challenges and future research directions of VLIT, providing insights for the continuous development of this field. The code and dataset related to this paper have been open-sourced at https://github.com/palchenli/VL-Instruction-Tuning.

Kun Yi, Yixiao Ge, Xiaotong Li et al. · tencent-ai

Since the development of self-supervised visual representation learning from contrastive learning to masked image modeling (MIM), there is no significant difference in essence, that is, how to design proper pretext tasks for vision dictionary look-up. MIM recently dominates this line of research with state-of-the-art performance on vision Transformers (ViTs), where the core is to enhance the patch-level visual context capturing of the network via denoising auto-encoding mechanism. Rather than tailoring image tokenizers with extra training stages as in previous works, we unleash the great potential of contrastive learning on denoising auto-encoding and introduce a pure MIM method, ConMIM, to produce simple intra-image inter-patch contrastive constraints as the sole learning objectives for masked patch prediction. We further strengthen the denoising mechanism with asymmetric designs, including image perturbations and model progress rates, to improve the network pre-training. ConMIM-pretrained models with various scales achieve competitive results on downstream image classification, semantic segmentation, object detection, and instance segmentation tasks, e.g., on ImageNet-1K classification, we achieve 83.9% top-1 accuracy with ViT-Small and 85.3% with ViT-Base without extra data for pre-training.

Jiajun Fei, Dian Li, Zhidong Deng et al. · tsinghua

Multi-modal large language models (MLLMs) have demonstrated considerable potential across various downstream tasks that require cross-domain knowledge. MLLMs capable of processing videos, known as Video-MLLMs, have attracted broad interest in video-language understanding. However, videos, especially long videos, contain more visual tokens than images, making them difficult for LLMs to process. Existing works either downsample visual features or extend the LLM context size, risking the loss of high-resolution information or slowing down inference speed. To address these limitations, we apply cross-attention layers in the intermediate projector between the visual encoder and the large language model (LLM). As the naive cross-attention mechanism is insensitive to temporal order, we further introduce causal cross-attention masks (CCAMs) within the cross-attention layers. This Video-MLLM, named Video-CCAM, is trained in a straightforward two-stage fashion: feature alignment and visual instruction tuning. We develop several Video-CCAM models based on LLMs of different sizes (4B, 9B, and 14B). Video-CCAM proves to be a robust Video-MLLM and shows outstanding performance from short videos to long ones. Among standard video benchmarks like MVBench and VideoChatGPT-QA, Video-CCAM shows outstanding performances (1st/2nd/3rd in MVBench and TGIF-QA, 2nd/3rd/4th in MSVD-QA, MSRVTT-QA, and ActivityNet-QA). In benchmarks encompassing long videos, Video-CCAM models can be directly adapted to long video understanding and still achieve exceptional scores despite being trained solely with images and 16-frame videos. Using 96 frames (6$\times$ the training number of frames), Video-CCAM models rank 1st/2nd/3rd in VideoVista and 1st/2nd/4th in MLVU among all open-source Video-MLLMs, respectively. The code is publicly available in \url{https://github.com/QQ-MM/Video-CCAM}.

Xiaohan Wang, Dian Li, Yilin Zhao et al.

Utilizing tools with Large Language Models (LLMs) is essential for grounding AI agents in real-world applications. The prevailing approach involves few-shot prompting with demonstrations or fine-tuning with expert annotations. However, mere in-context demonstrations may fail to cover sufficient knowledge for complex tools and tasks. Training on solution paths is also hindered by the high cost of expert annotations and generalizing to new tools. A core challenge of generalizable tool use lies in understanding the "meta", or fundamental natures of tools that are transferable across tasks, such as causality and constraints. In this paper, we present MetaTool, a novel tool learning methodology designed to generalize across any reusable toolset. Our approach incorporates a self-supervised augmentation technique derived from a series of meta-tasks. This involves predicting masked elements in the tool execution process. The self-supervised procedure enables scalable generation of high-quality QA data, which is handy for supervising tool understanding. By incorporating meta-task data into task-oriented training, our method significantly enhances the performance of open-source LLMs, achieving results comparable to ChatGPT in both tool-based planning and chatting scenarios. Through large-scale instruction tuning, the MetaTool model demonstrates impressive zero-shot generalizability on new tasks.

Xiaohan Wang, Mingze Yin, Yilin Zhao et al.

Advanced Large Multimodal Models (LMMs) have demonstrated impressive performance in K-12 reasoning tasks, exhibiting great promise as intelligent tutors. Realizing this potential requires models to navigate real-world examinations effectively, yet most existing benchmarks fail to capture the complexity of authentic testing environments. Specifically, most datasets are static, prone to data contamination, and are often confined to restricted modalities, disciplines, and evaluation criteria. To address these issues, we introduce LiveK12Bench, a dynamic, holistic, multi-disciplinary benchmark designed to evaluate the reasoning abilities of LMMs in realistic examination scenarios. LiveK12Bench comprises 2K+ verified questions spanning Mathematics, Physics, Chemistry, and Biology, sourced from the latest real-world exam papers and designed to grow over time. Our framework features several core innovations: 1) featuring an automated pipeline that continuously ingests and parses the latest examination papers to mitigate data leakage; and 2) proposing a novel `Mock Exam' evaluation scheme, which assesses the ability to complete end-to-end exams autonomously with accurate and efficient reasoning paths. Extensive experiments on 12 LMMs reveal that advanced models suffer substantial performance degradation under exam-realistic constraints: GPT-5's score drops from 79 to 53 (out of 100) when process rigor and efficiency are jointly evaluated. Our findings expose critical vulnerabilities, such as sensitivity to complex visual layouts, highlighting the gap between idealized reasoning capabilities and true educational readiness. Both code and dataset are publicly available.

Yongxian Wei, Yilin Zhao, Li Shen et al.

Data synthesis for training large reasoning models offers a scalable alternative to limited, human-curated datasets, enabling the creation of high-quality data. However, existing approaches face several challenges: (i) indiscriminate generation that ignores the solver's ability and yields low-value problems, or reliance on complex data pipelines to balance problem difficulty; and (ii) a lack of reasoning in problem generation, leading to shallow problem variants. In this paper, we develop a problem generator that reasons explicitly to plan problem directions before synthesis and adapts difficulty to the solver's ability. Specifically, we construct related problem pairs and augment them with intermediate problem-design CoT produced by a reasoning model. These data bootstrap problem-design strategies from the generator. Then, we treat the solver's feedback on synthetic problems as a reward signal, enabling the generator to calibrate difficulty and produce complementary problems near the edge of the solver's competence. Extensive experiments on 10 mathematical and general reasoning benchmarks show that our method achieves an average improvement of 2.5% and generalizes to both language and vision-language models. Moreover, a solver trained on the synthesized data provides improved rewards for continued generator training, enabling co-evolution and yielding a further 0.7% performance gain. Our code will be made publicly available here.

Jingyun Wang, Dian Li, Xiaohan Wang et al.

Plane Geometry Problem Solving (PGPS) is a multimodal reasoning task that aims to solve a plane geometric problem based on a geometric diagram and problem textual descriptions. Although Large Language Models (LLMs) possess strong reasoning skills, their direct application to PGPS is hindered by their inability to process visual diagrams. Existing works typically fine-tune Multimodal LLMs (MLLMs) end-to-end on large-scale PGPS data to enhance visual understanding and reasoning simultaneously. However, such joint optimization may compromise base LLMs' inherent reasoning capability. In this work, we observe that LLM itself is potentially a powerful PGPS solver when appropriately formulating visual information as textual descriptions. We propose to train a MLLM Interpreter to generate geometric descriptions for the visual diagram, and an off-the-shelf LLM is utilized to perform reasoning. Specifically, we choose Conditional Declaration Language (CDL) as the geometric description as its conciseness eases the MLLM Interpreter training. The MLLM Interpreter is fine-tuned via CoT (Chain-of-Thought)-augmented SFT followed by GRPO to generate CDL. Instead of using a conventional solution-based reward that compares the reasoning result with the ground-truth answer, we design CDL matching rewards to facilitate more effective GRPO training, which provides more direct and denser guidance for CDL generation. To support training, we construct a new dataset, Formalgeo7k-Rec-CoT, by manually reviewing Formalgeo7k v2 and incorporating CoT annotations. Extensive experiments on Formalgeo7k-Rec-CoT, Unigeo, and MathVista show our method (finetuned on only 5.5k data) performs favorably against leading open-source and closed-source MLLMs.

Youze Xue, Dian Li, Gang Liu

With the rapid advancement of multi-modal large language models (MLLMs) in recent years, the foundational Contrastive Language-Image Pretraining (CLIP) framework has been successfully extended to MLLMs, enabling more powerful and universal multi-modal embeddings for a wide range of retrieval tasks. Despite these developments, the core contrastive learning paradigm remains largely unchanged from CLIP-style models to MLLMs. Within this framework, the effective mining of hard negative samples continues to be a critical factor for enhancing performance. Prior works have introduced both offline and online strategies for hard negative mining to improve the efficiency of contrastive learning. While these approaches have led to improved multi-modal embeddings, the specific contribution of each hard negative sample to the learning process has not been thoroughly investigated. In this work, we conduct a detailed analysis of the gradients of the info-NCE loss with respect to the query, positive, and negative samples, elucidating the role of hard negatives in updating model parameters. Building upon this analysis, we propose to explicitly amplify the gradients associated with hard negative samples, thereby encouraging the model to learn more discriminative embeddings. Our multi-modal embedding model, trained with the proposed Explicit Gradient Amplifier and based on the LLaVA-OneVision-7B architecture, achieves state-of-the-art performance on the MMEB benchmark compared to previous methods utilizing the same MLLM backbone. Furthermore, when integrated with our self-developed MLLM, QQMM, our approach attains the top rank on the MMEB leaderboard. Code and models are released on https://github.com/QQ-MM/QQMM-embed.

Weidong Wang, Dian Li, Yujian Kang

In software engineering (SE) tasks, the naming approach is so important that it attracts many scholars from all over the world to study how to improve the quality of method names. To accurately recommend method names, we employ a novel framework to handle this problem. In our expeirments, nearly 8 million Java methods are collected from open source organizations as our evaluation dataset. In the first-phase recommendation, we introduce a fast and simple classifier based on the fast text neural network for reccomending potential method category. In the second-phase recomendation, we employ both two Long Short Term Memory Networks to reccomend consitent method names from each classification. Evaluation results prove that the proposed approach significantly outperforms state-of-the-art approach.

Rui Qian, Yuxi Li, Huabin Liu et al.

The crux of self-supervised video representation learning is to build general features from unlabeled videos. However, most recent works have mainly focused on high-level semantics and neglected lower-level representations and their temporal relationship which are crucial for general video understanding. To address these challenges, this paper proposes a multi-level feature optimization framework to improve the generalization and temporal modeling ability of learned video representations. Concretely, high-level features obtained from naive and prototypical contrastive learning are utilized to build distribution graphs, guiding the process of low-level and mid-level feature learning. We also devise a simple temporal modeling module from multi-level features to enhance motion pattern learning. Experiments demonstrate that multi-level feature optimization with the graph constraint and temporal modeling can greatly improve the representation ability in video understanding. Code is available at https://github.com/shvdiwnkozbw/Video-Representation-via-Multi-level-Optimization.

Lexiang Hu, Youze Xue, Dian Li et al.

Multimodal embeddings serve as a bridge for aligning vision and language, with the two primary implementations -- CLIP-based and MLLM-based embedding models -- both limited to capturing only global semantic information. Although numerous studies have focused on fine-grained understanding, we observe that complex scenarios currently targeted by MLLM embeddings often involve a hybrid perceptual pattern of both global and fine-grained elements, thus necessitating a compatible fusion mechanism. In this paper, we propose Adaptive Global and Fine-grained perceptual Fusion for MLLM Embeddings (AGFF-Embed), a method that prompts the MLLM to generate multiple embeddings focusing on different dimensions of semantic information, which are then adaptively and smoothly aggregated. Furthermore, we adapt AGFF-Embed with the Explicit Gradient Amplification (EGA) technique to achieve in-batch hard negatives enhancement without requiring fine-grained editing of the dataset. Evaluation on the MMEB and MMVP-VLM benchmarks shows that AGFF-Embed comprehensively achieves state-of-the-art performance in both general and fine-grained understanding compared to other multimodal embedding models.

Han Wang, Yilin Zhao, Dian Li et al.

Humor is previously regarded as a gift exclusive to humans for the following reasons. Humor is a culturally nuanced aspect of human language, presenting challenges for its understanding and generation. Humor generation necessitates a multi-hop reasoning process, with each hop founded on proper rationales. Although many studies, such as those related to GPT-o1, focus on logical reasoning with reflection and correction, they still fall short in humor generation. Due to the sparsity of the knowledge graph in creative thinking, it is arduous to achieve multi-hop reasoning. Consequently, in this paper, we propose a more robust framework for addressing the humor reasoning task, named LoL. LoL aims to inject external information to mitigate the sparsity of the knowledge graph, thereby enabling multi-hop reasoning. In the first stage of LoL, we put forward an automatic instruction-evolution method to incorporate the deeper and broader thinking processes underlying humor. Judgment-oriented instructions are devised to enhance the model's judgment capability, dynamically supplementing and updating the sparse knowledge graph. Subsequently, through reinforcement learning, the reasoning logic for each online-generated response is extracted using GPT-4o. In this process, external knowledge is re-introduced to aid the model in logical reasoning and the learning of human preferences. Finally, experimental results indicate that the combination of these two processes can enhance both the model's judgment ability and its generative capacity. These findings deepen our comprehension of the creative capabilities of large language models (LLMs) and offer approaches to boost LLMs' creative abilities for cross-domain innovative applications.

Fanrui Zhang, Dian Li, Qiang Zhang et al.

The rapid spread of multimodal misinformation on social media has raised growing concerns, while research on video misinformation detection remains limited due to the lack of large-scale, diverse datasets. Existing methods often overfit to rigid templates and lack deep reasoning over deceptive content. To address these challenges, we introduce FakeVV, a large-scale benchmark comprising over 100,000 video-text pairs with fine-grained, interpretable annotations. In addition, we further propose Fact-R1, a novel framework that integrates deep reasoning with collaborative rule-based reinforcement learning. Fact-R1 is trained through a three-stage process: (1) misinformation long-Chain-of-Thought (CoT) instruction tuning, (2) preference alignment via Direct Preference Optimization (DPO), and (3) Group Relative Policy Optimization (GRPO) using a novel verifiable reward function. This enables Fact-R1 to exhibit emergent reasoning behaviors comparable to those observed in advanced text-based reinforcement learning systems, but in the more complex multimodal misinformation setting. Our work establishes a new paradigm for misinformation detection, bridging large-scale video understanding, reasoning-guided alignment, and interpretable verification.

Zexian Yang, Dian Li, Dayan Wu et al.

Despite significant advancements in multimodal reasoning tasks, existing Large Vision-Language Models (LVLMs) are prone to producing visually ungrounded responses when interpreting associated images. In contrast, when humans embark on learning new knowledge, they often rely on a set of fundamental pre-study principles: reviewing outlines to grasp core concepts, summarizing key points to guide their focus and enhance understanding. However, such preparatory actions are notably absent in the current instruction tuning processes. This paper presents Re-Critic, an easily scalable rationale-augmented framework designed to incorporate fundamental rules and chain-of-thought (CoT) as a bridge to enhance reasoning abilities. Specifically, Re-Critic develops a visual rationale synthesizer that scalably augments raw instructions with rationale explanation. To probe more contextually grounded responses, Re-Critic employs an in-context self-critic mechanism to select response pairs for preference tuning. Experiments demonstrate that models fine-tuned with our rationale-augmented dataset yield gains that extend beyond hallucination-specific tasks to broader multimodal reasoning tasks.

Rui Qian, Weiyao Lin, John See et al.

This paper focuses on self-supervised video representation learning. Most existing approaches follow the contrastive learning pipeline to construct positive and negative pairs by sampling different clips. However, this formulation tends to bias to static background and have difficulty establishing global temporal structures. The major reason is that the positive pairs, i.e., different clips sampled from the same video, have limited temporal receptive field, and usually share similar background but differ in motions. To address these problems, we propose a framework to jointly utilize local clips and global videos to learn from detailed region-level correspondence as well as general long-term temporal relations. Based on a set of controllable augmentations, we achieve accurate appearance and motion pattern alignment through soft spatio-temporal region contrast. Our formulation is able to avoid the low-level redundancy shortcut by mutual information minimization to improve the generalization. We also introduce local-global temporal order dependency to further bridge the gap between clip-level and video-level representations for robust temporal modeling. Extensive experiments demonstrate that our framework is superior on three video benchmarks in action recognition and video retrieval, capturing more accurate temporal dynamics.

Weidong Wang, Yujian Kang, Dian Li

A large-scale evaluation for current naming approaches substantiates that such approaches are accurate. However, it is less known about which categories of method names work well via such naming approaches and how's the performance of naming approaches. To point out the superiority of the current naming approach, in this paper, we conduct an empirical study on such approaches in a new dataset. Moreover, we analyze the successful naming approaches above and find that: (1) around 60% of the accepted recommendation names are made on prefixes within get, set, is, and test. (2) A large portion (19.3%) of method names successfully recommended could be derived from the given method bodies. The comparisons also demonstrate the superior performance of the empirical study.

Yuying Ge, Yixiao Ge, Xihui Liu et al.

Pre-training a model to learn transferable video-text representation for retrieval has attracted a lot of attention in recent years. Previous dominant works mainly adopt two separate encoders for efficient retrieval, but ignore local associations between videos and texts. Another line of research uses a joint encoder to interact video with texts, but results in low efficiency since each text-video pair needs to be fed into the model. In this work, we enable fine-grained video-text interactions while maintaining high efficiency for retrieval via a novel pretext task, dubbed as Multiple Choice Questions (MCQ), where a parametric module BridgeFormer is trained to answer the "questions" constructed by the text features via resorting to the video features. Specifically, we exploit the rich semantics of text (i.e., nouns and verbs) to build questions, with which the video encoder can be trained to capture more regional content and temporal dynamics. In the form of questions and answers, the semantic associations between local video-text features can be properly established. BridgeFormer is able to be removed for downstream retrieval, rendering an efficient and flexible model with only two encoders. Our method outperforms state-of-the-art methods on the popular text-to-video retrieval task in five datasets with different experimental setups (i.e., zero-shot and fine-tune), including HowTo100M (one million videos). We further conduct zero-shot action recognition, which can be cast as video-to-text retrieval, and our approach also significantly surpasses its counterparts. As an additional benefit, our method achieves competitive results with much shorter pre-training videos on single-modality downstream tasks, e.g., action recognition with linear evaluation.

Lu Qi, Jason Kuen, Zhe Lin et al.

To improve instance-level detection/segmentation performance, existing self-supervised and semi-supervised methods extract either task-unrelated or task-specific training signals from unlabeled data. We show that these two approaches, at the two extreme ends of the task-specificity spectrum, are suboptimal for the task performance. Utilizing too little task-specific training signals causes underfitting to the ground-truth labels of downstream tasks, while the opposite causes overfitting to the ground-truth labels. To this end, we propose a novel Class-Agnostic Semi-Supervised Learning (CA-SSL) framework to achieve a more favorable task-specificity balance in extracting training signals from unlabeled data. CA-SSL has three training stages that act on either ground-truth labels (labeled data) or pseudo labels (unlabeled data). This decoupling strategy avoids the complicated scheme in traditional SSL methods that balances the contributions from both data types. Especially, we introduce a warmup training stage to achieve a more optimal balance in task specificity by ignoring class information in the pseudo labels, while preserving localization training signals. As a result, our warmup model can better avoid underfitting/overfitting when fine-tuned on the ground-truth labels in detection and segmentation tasks. Using 3.6M unlabeled data, we achieve a significant performance gain of 4.7% over ImageNet-pretrained baseline on FCOS object detection. In addition, our warmup model demonstrates excellent transferability to other detection and segmentation frameworks.

Mingkang Tang, Zhanyu Wang, Zhenhua Liu et al.

Video captioning is a challenging task since it requires generating sentences describing various diverse and complex videos. Existing video captioning models lack adequate visual representation due to the neglect of the existence of gaps between videos and texts. To bridge this gap, in this paper, we propose a CLIP4Caption framework that improves video captioning based on a CLIP-enhanced video-text matching network (VTM). This framework is taking full advantage of the information from both vision and language and enforcing the model to learn strongly text-correlated video features for text generation. Besides, unlike most existing models using LSTM or GRU as the sentence decoder, we adopt a Transformer structured decoder network to effectively learn the long-range visual and language dependency. Additionally, we introduce a novel ensemble strategy for captioning tasks. Experimental results demonstrate the effectiveness of our method on two datasets: 1) on MSR-VTT dataset, our method achieved a new state-of-the-art result with a significant gain of up to 10% in CIDEr; 2) on the private test data, our method ranking 2nd place in the ACM MM multimedia grand challenge 2021: Pre-training for Video Understanding Challenge. It is noted that our model is only trained on the MSR-VTT dataset.

Mingkang Tang, Zhanyu Wang, Zhaoyang Zeng et al.

This report describes our solution to the VALUE Challenge 2021 in the captioning task. Our solution, named CLIP4Caption++, is built on X-Linear/X-Transformer, which is an advanced model with encoder-decoder architecture. We make the following improvements on the proposed CLIP4Caption++: We employ an advanced encoder-decoder model architecture X-Transformer as our main framework and make the following improvements: 1) we utilize three strong pre-trained CLIP models to extract the text-related appearance visual features. 2) we adopt the TSN sampling strategy for data enhancement. 3) we involve the video subtitle information to provide richer semantic information. 3) we introduce the subtitle information, which fuses with the visual features as guidance. 4) we design word-level and sentence-level ensemble strategies. Our proposed method achieves 86.5, 148.4, 64.5 CIDEr scores on VATEX, YC2C, and TVC datasets, respectively, which shows the superior performance of our proposed CLIP4Caption++ on all three datasets.