Yixiao Ge

Alex Jinpeng Wang, Yixiao Ge, Rui Yan et al. · tencent-ai

Mainstream Video-Language Pre-training models \cite{actbert,clipbert,violet} consist of three parts, a video encoder, a text encoder, and a video-text fusion Transformer. They pursue better performance via utilizing heavier unimodal encoders or multimodal fusion Transformers, resulting in increased parameters with lower efficiency in downstream tasks. In this work, we for the first time introduce an end-to-end video-language model, namely \textit{all-in-one Transformer}, that embeds raw video and textual signals into joint representations using a unified backbone architecture. We argue that the unique temporal information of video data turns out to be a key barrier hindering the design of a modality-agnostic Transformer. To overcome the challenge, we introduce a novel and effective token rolling operation to encode temporal representations from video clips in a non-parametric manner. The careful design enables the representation learning of both video-text multimodal inputs and unimodal inputs using a unified backbone model. Our pre-trained all-in-one Transformer is transferred to various downstream video-text tasks after fine-tuning, including text-video retrieval, video-question answering, multiple choice and visual commonsense reasoning. State-of-the-art performances with the minimal model FLOPs on nine datasets demonstrate the superiority of our method compared to the competitive counterparts. The code and pretrained model have been released in https://github.com/showlab/all-in-one.

Yuxin Fang, Shusheng Yang, Shijie Wang et al. · tencent-ai

We present an approach to efficiently and effectively adapt a masked image modeling (MIM) pre-trained vanilla Vision Transformer (ViT) for object detection, which is based on our two novel observations: (i) A MIM pre-trained vanilla ViT encoder can work surprisingly well in the challenging object-level recognition scenario even with randomly sampled partial observations, e.g., only 25% $\sim$ 50% of the input embeddings. (ii) In order to construct multi-scale representations for object detection from single-scale ViT, a randomly initialized compact convolutional stem supplants the pre-trained large kernel patchify stem, and its intermediate features can naturally serve as the higher resolution inputs of a feature pyramid network without further upsampling or other manipulations. While the pre-trained ViT is only regarded as the 3$^{rd}$-stage of our detector's backbone instead of the whole feature extractor. This results in a ConvNet-ViT hybrid feature extractor. The proposed detector, named MIMDet, enables a MIM pre-trained vanilla ViT to outperform hierarchical Swin Transformer by 2.5 box AP and 2.6 mask AP on COCO, and achieves better results compared with the previous best adapted vanilla ViT detector using a more modest fine-tuning recipe while converging 2.8$\times$ faster. Code and pre-trained models are available at https://github.com/hustvl/MIMDet.

Zhuoyan Luo, Fengyuan Shi, Yixiao Ge et al. · tencent-ai

The Open-MAGVIT2 project produces an open-source replication of Google's MAGVIT-v2 tokenizer, a tokenizer with a super-large codebook (i.e., $2^{18}$ codes), and achieves the state-of-the-art reconstruction performance on ImageNet and UCF benchmarks. We also provide a tokenizer pre-trained on large-scale data, significantly outperforming Cosmos on zero-shot benchmarks (1.93 vs. 0.78 rFID on ImageNet original resolution). Furthermore, we explore its application in plain auto-regressive models to validate scalability properties, producing a family of auto-regressive image generation models ranging from 300M to 1.5B. To assist auto-regressive models in predicting with a super-large vocabulary, we factorize it into two sub-vocabulary of different sizes by asymmetric token factorization, and further introduce ``next sub-token prediction'' to enhance sub-token interaction for better generation quality. We release all models and codes to foster innovation and creativity in the field of auto-regressive visual generation.

Yunpeng Bai, Xintao Wang, Yan-pei Cao et al. · tencent-ai

This paper introduces DreamDiffusion, a novel method for generating high-quality images directly from brain electroencephalogram (EEG) signals, without the need to translate thoughts into text. DreamDiffusion leverages pre-trained text-to-image models and employs temporal masked signal modeling to pre-train the EEG encoder for effective and robust EEG representations. Additionally, the method further leverages the CLIP image encoder to provide extra supervision to better align EEG, text, and image embeddings with limited EEG-image pairs. Overall, the proposed method overcomes the challenges of using EEG signals for image generation, such as noise, limited information, and individual differences, and achieves promising results. Quantitative and qualitative results demonstrate the effectiveness of the proposed method as a significant step towards portable and low-cost ``thoughts-to-image'', with potential applications in neuroscience and computer vision. The code is available here \url{https://github.com/bbaaii/DreamDiffusion}.

Bohao Li, Yuying Ge, Yixiao Ge et al. · tencent-ai

Multimodal large language models (MLLMs), building upon the foundation of powerful large language models (LLMs), have recently demonstrated exceptional capabilities in generating not only texts but also images given interleaved multimodal inputs (acting like a combination of GPT-4V and DALL-E 3). However, existing MLLM benchmarks remain limited to assessing only models' comprehension ability of single image-text inputs, failing to keep up with the strides made in MLLMs. A comprehensive benchmark is imperative for investigating the progress and uncovering the limitations of current MLLMs. In this work, we categorize the capabilities of MLLMs into hierarchical levels from $L_0$ to $L_4$ based on the modalities they can accept and generate, and propose SEED-Bench-2, a comprehensive benchmark that evaluates the \textbf{hierarchical} capabilities of MLLMs. Specifically, SEED-Bench-2 comprises 24K multiple-choice questions with accurate human annotations, which spans 27 dimensions, including the evaluation of both text and image generation. Multiple-choice questions with groundtruth options derived from human annotation enables an objective and efficient assessment of model performance, eliminating the need for human or GPT intervention during evaluation. We further evaluate the performance of 23 prominent open-source MLLMs and summarize valuable observations. By revealing the limitations of existing MLLMs through extensive evaluations, we aim for SEED-Bench-2 to provide insights that will motivate future research towards the goal of General Artificial Intelligence. Dataset and evaluation code are available at \href{https://github.com/AILab-CVC/SEED-Bench}

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

Image BERT pre-training with masked image modeling (MIM) becomes a popular practice to cope with self-supervised representation learning. A seminal work, BEiT, casts MIM as a classification task with a visual vocabulary, tokenizing the continuous visual signals into discrete vision tokens using a pre-learned dVAE. Despite a feasible solution, the improper discretization hinders further improvements of image pre-training. Since image discretization has no ground-truth answers, we believe that the masked patch should not be assigned with a unique token id even if a better tokenizer can be obtained. In this work, we introduce an improved BERT-style image pre-training method, namely mc-BEiT, which performs MIM proxy tasks towards eased and refined multi-choice training objectives. Specifically, the multi-choice supervision for the masked image patches is formed by the soft probability vectors of the discrete token ids, which are predicted by the off-the-shelf image tokenizer and further refined by high-level inter-patch perceptions resorting to the observation that similar patches should share their choices. Extensive experiments on classification, segmentation, and detection tasks demonstrate the superiority of our method, e.g., the pre-trained ViT-B achieves 84.1% top-1 fine-tuning accuracy on ImageNet-1K classification, 49.2% AP^b and 44.0% AP^m of object detection and instance segmentation on COCO, 50.8% mIOU on ADE20K semantic segmentation, outperforming the competitive counterparts. The code will be available at https://github.com/lixiaotong97/mc-BEiT.

Xiaotong Li, Zixuan Hu, Jun Liu et al. · tencent-ai

Though deep neural networks have achieved impressive success on various vision tasks, obvious performance degradation still exists when models are tested in out-of-distribution scenarios. In addressing this limitation, we ponder that the feature statistics (mean and standard deviation), which carry the domain characteristics of the training data, can be properly manipulated to improve the generalization ability of deep learning models. Existing methods commonly consider feature statistics as deterministic values measured from the learned features and do not explicitly model the uncertain statistics discrepancy caused by potential domain shifts during testing. In this paper, we improve the network generalization ability by modeling domain shifts with uncertainty (DSU), i.e., characterizing the feature statistics as uncertain distributions during training. Specifically, we hypothesize that the feature statistic, after considering the potential uncertainties, follows a multivariate Gaussian distribution. During inference, we propose an instance-wise adaptation strategy that can adaptively deal with the unforeseeable shift and further enhance the generalization ability of the trained model with negligible additional cost. We also conduct theoretical analysis on the aspects of generalization error bound and the implicit regularization effect, showing the efficacy of our method. Extensive experiments demonstrate that our method consistently improves the network generalization ability on multiple vision tasks, including image classification, semantic segmentation, instance retrieval, and pose estimation. Our methods are simple yet effective and can be readily integrated into networks without additional trainable parameters or loss constraints. Code will be released in https://github.com/lixiaotong97/DSU.

Chengyue Wu, Teng Wang, Yixiao Ge et al. · tencent-ai

Foundation models have achieved great advances in multi-task learning with a unified interface of unimodal and multimodal tasks. However, the potential of such multi-task learners has not been exploited during transfer learning. In this work, we present a universal parameter-efficient transfer learning method, termed Predict-Interpolate Tuning ($π$-Tuning), for vision, language, and vision-language tasks. It aggregates the parameters of lightweight task-specific experts learned from similar tasks to aid the target downstream task. The task similarities are predicted in a unified modality-independent space, yielding a scalable graph to demonstrate task relationships. $π$-Tuning has several appealing benefits. First, it flexibly explores both intra- and inter-modal transferability between similar tasks to improve the accuracy and robustness of transfer learning, especially in data-scarce scenarios. Second, it offers a systematical solution for transfer learning with multi-task prediction-and-then-interpolation, compatible with diverse types of parameter-efficient experts, such as prompt and adapter. Third, an extensive study of task-level mutual benefits on 14 unimodal and 6 multimodal datasets shows that $π$-Tuning surpasses fine-tuning and other parameter-efficient transfer learning methods both in full-shot and low-shot regimes. The task graph also enables an in-depth interpretable analysis of task transferability across modalities. The code will be available at https://github.com/TencentARC/pi-Tuning.

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.

Wenqi Shao, Xun Zhao, Yixiao Ge et al. · tencent-ai

This paper addresses an important problem of ranking the pre-trained deep neural networks and screening the most transferable ones for downstream tasks. It is challenging because the ground-truth model ranking for each task can only be generated by fine-tuning the pre-trained models on the target dataset, which is brute-force and computationally expensive. Recent advanced methods proposed several lightweight transferability metrics to predict the fine-tuning results. However, these approaches only capture static representations but neglect the fine-tuning dynamics. To this end, this paper proposes a new transferability metric, called \textbf{S}elf-challenging \textbf{F}isher \textbf{D}iscriminant \textbf{A}nalysis (\textbf{SFDA}), which has many appealing benefits that existing works do not have. First, SFDA can embed the static features into a Fisher space and refine them for better separability between classes. Second, SFDA uses a self-challenging mechanism to encourage different pre-trained models to differentiate on hard examples. Third, SFDA can easily select multiple pre-trained models for the model ensemble. Extensive experiments on $33$ pre-trained models of $11$ downstream tasks show that SFDA is efficient, effective, and robust when measuring the transferability of pre-trained models. For instance, compared with the state-of-the-art method NLEEP, SFDA demonstrates an average of $59.1$\% gain while bringing $22.5$x speedup in wall-clock time. The code will be available at \url{https://github.com/TencentARC/SFDA}.

Weixian Lei, Yixiao Ge, Kun Yi et al. · tencent-ai

Aiming to advance AI agents, large foundation models significantly improve reasoning and instruction execution, yet the current focus on vision and language neglects the potential of perceiving diverse modalities in open-world environments. However, the success of data-driven vision and language models is costly or even infeasible to be reproduced for rare modalities. In this paper, we present ViT-Lens-2 that facilitates efficient omni-modal representation learning by perceiving novel modalities with a pretrained ViT and aligning them to a pre-defined space. Specifically, the modality-specific lens is tuned to project any-modal signals to an intermediate embedding space, which are then processed by a strong ViT with pre-trained visual knowledge. The encoded representations are optimized toward aligning with the modal-independent space, pre-defined by off-the-shelf foundation models. ViT-Lens-2 provides a unified solution for representation learning of increasing modalities with two appealing advantages: (i) Unlocking the great potential of pretrained ViTs to novel modalities effectively with efficient data regime; (ii) Enabling emergent downstream capabilities through modality alignment and shared ViT parameters. We tailor ViT-Lens-2 to learn representations for 3D point cloud, depth, audio, tactile and EEG, and set new state-of-the-art results across various understanding tasks, such as zero-shot classification. By seamlessly integrating ViT-Lens-2 into Multimodal Foundation Models, we enable Any-modality to Text and Image Generation in a zero-shot manner. Code and models are available at https://github.com/TencentARC/ViT-Lens.

Jay Zhangjie Wu, Yixiao Ge, Xintao Wang et al. · tencent-ai

To replicate the success of text-to-image (T2I) generation, recent works employ large-scale video datasets to train a text-to-video (T2V) generator. Despite their promising results, such paradigm is computationally expensive. In this work, we propose a new T2V generation setting$\unicode{x2014}$One-Shot Video Tuning, where only one text-video pair is presented. Our model is built on state-of-the-art T2I diffusion models pre-trained on massive image data. We make two key observations: 1) T2I models can generate still images that represent verb terms; 2) extending T2I models to generate multiple images concurrently exhibits surprisingly good content consistency. To further learn continuous motion, we introduce Tune-A-Video, which involves a tailored spatio-temporal attention mechanism and an efficient one-shot tuning strategy. At inference, we employ DDIM inversion to provide structure guidance for sampling. Extensive qualitative and numerical experiments demonstrate the remarkable ability of our method across various applications.

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.

Xiaotong Li, Zixuan Hu, Yixiao Ge et al. · tencent-ai

Transfer learning has become crucial in computer vision tasks due to the vast availability of pre-trained deep learning models. However, selecting the optimal pre-trained model from a diverse pool for a specific downstream task remains a challenge. Existing methods for measuring the transferability of pre-trained models rely on statistical correlations between encoded static features and task labels, but they overlook the impact of underlying representation dynamics during fine-tuning, leading to unreliable results, especially for self-supervised models. In this paper, we present an insightful physics-inspired approach named PED to address these challenges. We reframe the challenge of model selection through the lens of potential energy and directly model the interaction forces that influence fine-tuning dynamics. By capturing the motion of dynamic representations to decline the potential energy within a force-driven physical model, we can acquire an enhanced and more stable observation for estimating transferability. The experimental results on 10 downstream tasks and 12 self-supervised models demonstrate that our approach can seamlessly integrate into existing ranking techniques and enhance their performances, revealing its effectiveness for the model selection task and its potential for understanding the mechanism in transfer learning. Code will be available at https://github.com/lixiaotong97/PED.

Teng Wang, Yixiao Ge, Feng Zheng et al. · tencent-ai

The state of the arts in vision-language pretraining (VLP) achieves exemplary performance but suffers from high training costs resulting from slow convergence and long training time, especially on large-scale web datasets. An essential obstacle to training efficiency lies in the entangled prediction rate (percentage of tokens for reconstruction) and corruption rate (percentage of corrupted tokens) in masked language modeling (MLM), that is, a proper corruption rate is achieved at the cost of a large portion of output tokens being excluded from prediction loss. To accelerate the convergence of VLP, we propose a new pretraining task, namely, free language modeling (FLM), that enables a 100% prediction rate with arbitrary corruption rates. FLM successfully frees the prediction rate from the tie-up with the corruption rate while allowing the corruption spans to be customized for each token to be predicted. FLM-trained models are encouraged to learn better and faster given the same GPU time by exploiting bidirectional contexts more flexibly. Extensive experiments show FLM could achieve an impressive 2.5x pretraining time reduction in comparison to the MLM-based methods, while keeping competitive performance on both vision-language understanding and generation tasks. Code will be public at https://github.com/TencentARC/FLM.

Binjie Zhang, Yixiao Ge, Xuyuan Xu et al. · tencent-ai

Visual foundation models like CLIP excel in learning feature representations from extensive datasets through self-supervised methods, demonstrating remarkable transfer learning and generalization capabilities. A growing number of applications based on visual foundation models are emerging, including innovative solutions such as BLIP-2. These applications employ pre-trained CLIP models as upstream feature extractors and train various downstream modules to accomplish diverse tasks. In situations involving system upgrades that require updating the upstream foundation model, it becomes essential to re-train all downstream modules to adapt to the new foundation model, which is inflexible and inefficient. In this paper, we introduce a parameter-efficient and task-agnostic adapter, dubbed TaCA, that facilitates compatibility across distinct foundation models while ensuring enhanced performance for the new models. TaCA allows downstream applications to seamlessly integrate better-performing foundation models without necessitating retraining. We conduct extensive experimental validation of TaCA using different scales of models with up to one billion parameters on various tasks such as video-text retrieval, video recognition, and visual question answering. The results consistently demonstrate the emergent ability of TaCA on hot-plugging upgrades for visual foundation models. Codes and models will be available at https://github.com/TencentARC/TaCA.

Ziyun Zeng, Yuying Ge, Xihui Liu et al. · tencent-ai

Pre-training on large-scale video data has become a common recipe for learning transferable spatiotemporal representations in recent years. Despite some progress, existing methods are mostly limited to highly curated datasets (e.g., K400) and exhibit unsatisfactory out-of-the-box representations. We argue that it is due to the fact that they only capture pixel-level knowledge rather than spatiotemporal semantics, which hinders further progress in video understanding. Inspired by the great success of image-text pre-training (e.g., CLIP), we take the first step to exploit language semantics to boost transferable spatiotemporal representation learning. We introduce a new pretext task, Turning to Video for Transcript Sorting (TVTS), which sorts shuffled ASR scripts by attending to learned video representations. We do not rely on descriptive captions and learn purely from video, i.e., leveraging the natural transcribed speech knowledge to provide noisy but useful semantics over time. Our method enforces the vision model to contextualize what is happening over time so that it can re-organize the narrative transcripts, and can seamlessly apply to large-scale uncurated video data in the real world. Our method demonstrates strong out-of-the-box spatiotemporal representations on diverse benchmarks, e.g., +13.6% gains over VideoMAE on SSV2 via linear probing. The code is available at https://github.com/TencentARC/TVTS.

Bohao Li, Rui Wang, Guangzhi Wang et al. · tencent-ai

Based on powerful Large Language Models (LLMs), recent generative Multimodal Large Language Models (MLLMs) have gained prominence as a pivotal research area, exhibiting remarkable capability for both comprehension and generation. In this work, we address the evaluation of generative comprehension in MLLMs as a preliminary step towards a comprehensive assessment of generative models, by introducing a benchmark named SEED-Bench. SEED-Bench consists of 19K multiple choice questions with accurate human annotations (x 6 larger than existing benchmarks), which spans 12 evaluation dimensions including the comprehension of both the image and video modality. We develop an advanced pipeline for generating multiple-choice questions that target specific evaluation dimensions, integrating both automatic filtering and manual verification processes. Multiple-choice questions with groundtruth options derived from human annotation enables an objective and efficient assessment of model performance, eliminating the need for human or GPT intervention during evaluation. We further evaluate the performance of 18 models across all 12 dimensions, covering both the spatial and temporal understanding. By revealing the limitations of existing MLLMs through evaluation results, we aim for SEED-Bench to provide insights for motivating future research. We will launch and consistently maintain a leaderboard to provide a platform for the community to assess and investigate model capability.

Guanyu Cai, Yixiao Ge, Binjie Zhang et al. · tencent-ai

Recent dominant methods for video-language pre-training (VLP) learn transferable representations from the raw pixels in an end-to-end manner to achieve advanced performance on downstream video-language retrieval. Despite the impressive results, VLP research becomes extremely expensive with the need for massive data and a long training time, preventing further explorations. In this work, we revitalize region features of sparsely sampled video clips to significantly reduce both spatial and temporal visual redundancy towards democratizing VLP research at the same time achieving state-of-the-art results. Specifically, to fully explore the potential of region features, we introduce a novel bidirectional region-word alignment regularization that properly optimizes the fine-grained relations between regions and certain words in sentences, eliminating the domain/modality disconnections between pre-extracted region features and text. Extensive results of downstream video-language retrieval tasks on four datasets demonstrate the superiority of our method on both effectiveness and efficiency, \textit{e.g.}, our method achieves competing results with 80\% fewer data and 85\% less pre-training time compared to the most efficient VLP method so far \cite{lei2021less}. The code will be available at \url{https://github.com/showlab/DemoVLP}.

Chen Li, Xutan Peng, Teng Wang et al. · tencent-ai

Art forms such as movies and television (TV) dramas are reflections of the real world, which have attracted much attention from the multimodal learning community recently. However, existing corpora in this domain share three limitations: (1) annotated in a scene-oriented fashion, they ignore the coherence within plots; (2) their text lacks empathy and seldom mentions situational context; (3) their video clips fail to cover long-form relationship due to short duration. To address these fundamental issues, using 1,106 TV drama episodes and 24,875 informative plot-focused sentences written by professionals, with the help of 449 human annotators, we constructed PTVD, the first plot-oriented multimodal dataset in the TV domain. It is also the first non-English dataset of its kind. Additionally, PTVD contains more than 26 million bullet screen comments (BSCs), powering large-scale pre-training. Next, aiming to open-source a strong baseline for follow-up works, we developed the multimodal algorithm that attacks different cinema/TV modelling problems with a unified architecture. Extensive experiments on three cognitive-inspired tasks yielded a number of novel observations (some of them being quite counter-intuition), further validating the value of PTVD in promoting multimodal research. The dataset and codes are released at \url{https://ptvd.github.io/}.

Xiaohan Ding, Yiyuan Zhang, Yixiao Ge et al. · tencent-ai

Large-kernel convolutional neural networks (ConvNets) have recently received extensive research attention, but two unresolved and critical issues demand further investigation. 1) The architectures of existing large-kernel ConvNets largely follow the design principles of conventional ConvNets or transformers, while the architectural design for large-kernel ConvNets remains under-addressed. 2) As transformers have dominated multiple modalities, it remains to be investigated whether ConvNets also have a strong universal perception ability in domains beyond vision. In this paper, we contribute from two aspects. 1) We propose four architectural guidelines for designing large-kernel ConvNets, the core of which is to exploit the essential characteristics of large kernels that distinguish them from small kernels - they can see wide without going deep. Following such guidelines, our proposed large-kernel ConvNet shows leading performance in image recognition (ImageNet accuracy of 88.0%, ADE20K mIoU of 55.6%, and COCO box AP of 56.4%), demonstrating better performance and higher speed than the recent powerful competitors. 2) We discover large kernels are the key to unlocking the exceptional performance of ConvNets in domains where they were originally not proficient. With certain modality-related preprocessing approaches, the proposed model achieves state-of-the-art performance on time-series forecasting and audio recognition tasks even without modality-specific customization to the architecture. All the code and models are publicly available on GitHub and Huggingface.

Yuying Ge, Sijie Zhao, Ziyun Zeng et al. · tencent-ai

The great success of Large Language Models (LLMs) has expanded the potential of multimodality, contributing to the gradual evolution of General Artificial Intelligence (AGI). A true AGI agent should not only possess the capability to perform predefined multi-tasks but also exhibit emergent abilities in an open-world context. However, despite the considerable advancements made by recent multimodal LLMs, they still fall short in effectively unifying comprehension and generation tasks, let alone open-world emergent abilities. We contend that the key to overcoming the present impasse lies in enabling text and images to be represented and processed interchangeably within a unified autoregressive Transformer. To this end, we introduce SEED, an elaborate image tokenizer that empowers LLMs with the ability to SEE and Draw at the same time. We identify two crucial design principles: (1) Image tokens should be independent of 2D physical patch positions and instead be produced with a 1D causal dependency, exhibiting intrinsic interdependence that aligns with the left-to-right autoregressive prediction mechanism in LLMs. (2) Image tokens should capture high-level semantics consistent with the degree of semantic abstraction in words, and be optimized for both discriminativeness and reconstruction during the tokenizer training phase. With SEED tokens, LLM is able to perform scalable multimodal autoregression under its original training recipe, i.e., next-word prediction. SEED-LLaMA is therefore produced by large-scale pretraining and instruction tuning on the interleaved textual and visual data, demonstrating impressive performance on a broad range of multimodal comprehension and generation tasks. More importantly, SEED-LLaMA has exhibited compositional emergent abilities such as multi-turn in-context multimodal generation, acting like your AI assistant.

Yuying Ge, Yixiao Ge, Ziyun Zeng et al. · tencent-ai

We present SEED, an elaborate image tokenizer that empowers Large Language Models (LLMs) with the emergent ability to SEE and Draw at the same time. Research on image tokenizers has previously reached an impasse, as frameworks employing quantized visual tokens have lost prominence due to subpar performance and convergence in multimodal comprehension (compared to BLIP-2, etc.) or generation (compared to Stable Diffusion, etc.). Despite the limitations, we remain confident in its natural capacity to unify visual and textual representations, facilitating scalable multimodal training with LLM's original recipe. In this study, we identify two crucial principles for the architecture and training of SEED that effectively ease subsequent alignment with LLMs. (1) Image tokens should be independent of 2D physical patch positions and instead be produced with a 1D causal dependency, exhibiting intrinsic interdependence that aligns with the left-to-right autoregressive prediction mechanism in LLMs. (2) Image tokens should capture high-level semantics consistent with the degree of semantic abstraction in words, and be optimized for both discriminativeness and reconstruction during the tokenizer training phase. As a result, the off-the-shelf LLM is able to perform both image-to-text and text-to-image generation by incorporating our SEED through efficient LoRA tuning. Comprehensive multimodal pretraining and instruction tuning, which may yield improved results, are reserved for future investigation. This version of SEED was trained in 5.7 days using only 64 V100 GPUs and 5M publicly available image-text pairs. Our preliminary study emphasizes the great potential of discrete visual tokens in versatile multimodal LLMs and the importance of proper image tokenizers in broader research.

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.

Shuai Yang, Yuying Ge, Yang Li et al. · tencent-ai

With the remarkable advancements in image generation and open-form text generation, the creation of interleaved image-text content has become an increasingly intriguing field. Multimodal story generation, characterized by producing narrative texts and vivid images in an interleaved manner, has emerged as a valuable and practical task with broad applications. However, this task poses significant challenges, as it necessitates the comprehension of the complex interplay between texts and images, and the ability to generate long sequences of coherent, contextually relevant texts and visuals. In this work, we propose SEED-Story, a novel method that leverages a Multimodal Large Language Model (MLLM) to generate extended multimodal stories. Our model, built upon the powerful comprehension capability of MLLM, predicts text tokens as well as visual tokens, which are subsequently processed with an adapted visual de-tokenizer to produce images with consistent characters and styles. We further propose multimodal attention sink mechanism to enable the generation of stories with up to 25 sequences (only 10 for training) in a highly efficient autoregressive manner. Additionally, we present a large-scale and high-resolution dataset named StoryStream for training our model and quantitatively evaluating the task of multimodal story generation in various aspects.

Yuying Ge, Yixiao Ge, Xihui Liu et al. · tencent-ai

Dominant pre-training work for video-text retrieval mainly adopt the "dual-encoder" architectures to enable efficient retrieval, where two separate encoders are used to contrast global video and text representations, but ignore detailed local semantics. The recent success of image BERT pre-training with masked visual modeling that promotes the learning of local visual context, motivates a possible solution to address the above limitation. In this work, we for the first time investigate masked visual modeling in video-text pre-training with the "dual-encoder" architecture. We perform Masked visual modeling with Injected LanguagE Semantics (MILES) by employing an extra snapshot video encoder as an evolving "tokenizer" to produce reconstruction targets for masked video patch prediction. Given the corrupted video, the video encoder is trained to recover text-aligned features of the masked patches via reasoning with the visible regions along the spatial and temporal dimensions, which enhances the discriminativeness of local visual features and the fine-grained cross-modality alignment. Our method outperforms state-of-the-art methods for text-to-video retrieval on four datasets with both zero-shot and fine-tune evaluation protocols. Our approach also surpasses the baseline models significantly on zero-shot action recognition, which can be cast as video-to-text retrieval.

Cheng Cheng, Lin Song, Ruoyi Xue et al. · tencent-ai

The contrastive vision-language pre-training, known as CLIP, demonstrates remarkable potential in perceiving open-world visual concepts, enabling effective zero-shot image recognition. Nevertheless, few-shot learning methods based on CLIP typically require offline fine-tuning of the parameters on few-shot samples, resulting in longer inference time and the risk of over-fitting in certain domains. To tackle these challenges, we propose the Meta-Adapter, a lightweight residual-style adapter, to refine the CLIP features guided by the few-shot samples in an online manner. With a few training samples, our method can enable effective few-shot learning capabilities and generalize to unseen data or tasks without additional fine-tuning, achieving competitive performance and high efficiency. Without bells and whistles, our approach outperforms the state-of-the-art online few-shot learning method by an average of 3.6\% on eight image classification datasets with higher inference speed. Furthermore, our model is simple and flexible, serving as a plug-and-play module directly applicable to downstream tasks. Without further fine-tuning, Meta-Adapter obtains notable performance improvements in open-vocabulary object detection and segmentation tasks.

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

The recent progress in Large Language Models (LLM) has spurred various advancements in image-language conversation agents, while how to build a proficient video-based dialogue system is still under exploration. Considering the extensive scale of LLM and visual backbone, minimal GPU memory is left for facilitating effective temporal modeling, which is crucial for comprehending and providing feedback on videos. To this end, we propose Branching Temporal Adapter (BT-Adapter), a novel method for extending image-language pretrained models into the video domain. Specifically, BT-Adapter serves as a plug-and-use temporal modeling branch alongside the pretrained visual encoder, which is tuned while keeping the backbone frozen. Just pretrained once, BT-Adapter can be seamlessly integrated into all image conversation models using this version of CLIP, enabling video conversations without the need for video instructions. Besides, we develop a unique asymmetric token masking strategy inside the branch with tailor-made training tasks for BT-Adapter, facilitating faster convergence and better results. Thanks to BT-Adapter, we are able to empower existing multimodal dialogue models with strong video understanding capabilities without incurring excessive GPU costs. Without bells and whistles, BT-Adapter achieves (1) state-of-the-art zero-shot results on various video tasks using thousands of fewer GPU hours. (2) better performance than current video chatbots without any video instruction tuning. (3) state-of-the-art results of video chatting using video instruction tuning, outperforming previous SOTAs by a large margin.

Yuchao Gu, Xintao Wang, Yixiao Ge et al. · tencent-ai

Vector-Quantized (VQ-based) generative models usually consist of two basic components, i.e., VQ tokenizers and generative transformers. Prior research focuses on improving the reconstruction fidelity of VQ tokenizers but rarely examines how the improvement in reconstruction affects the generation ability of generative transformers. In this paper, we surprisingly find that improving the reconstruction fidelity of VQ tokenizers does not necessarily improve the generation. Instead, learning to compress semantic features within VQ tokenizers significantly improves generative transformers' ability to capture textures and structures. We thus highlight two competing objectives of VQ tokenizers for image synthesis: semantic compression and details preservation. Different from previous work that only pursues better details preservation, we propose Semantic-Quantized GAN (SeQ-GAN) with two learning phases to balance the two objectives. In the first phase, we propose a semantic-enhanced perceptual loss for better semantic compression. In the second phase, we fix the encoder and codebook, but enhance and finetune the decoder to achieve better details preservation. The proposed SeQ-GAN greatly improves VQ-based generative models and surpasses the GAN and Diffusion Models on both unconditional and conditional image generation. Our SeQ-GAN (364M) achieves Frechet Inception Distance (FID) of 6.25 and Inception Score (IS) of 140.9 on 256x256 ImageNet generation, a remarkable improvement over VIT-VQGAN (714M), which obtains 11.2 FID and 97.2 IS.

Rui Yang, Lin Song, Yixiao Ge et al. · tencent-ai

Box-supervised instance segmentation has gained much attention as it requires only simple box annotations instead of costly mask or polygon annotations. However, existing box-supervised instance segmentation models mainly focus on mask-based frameworks. We propose a new end-to-end training technique, termed BoxSnake, to achieve effective polygonal instance segmentation using only box annotations for the first time. Our method consists of two loss functions: (1) a point-based unary loss that constrains the bounding box of predicted polygons to achieve coarse-grained segmentation; and (2) a distance-aware pairwise loss that encourages the predicted polygons to fit the object boundaries. Compared with the mask-based weakly-supervised methods, BoxSnake further reduces the performance gap between the predicted segmentation and the bounding box, and shows significant superiority on the Cityscapes dataset. The code has been available publicly.

Binjie Zhang, Yixiao Ge, Yantao Shen et al. · tencent-ai

Conventional model upgrades for visual search systems require offline refresh of gallery features by feeding gallery images into new models (dubbed as "backfill"), which is time-consuming and expensive, especially in large-scale applications. The task of backward-compatible representation learning is therefore introduced to support backfill-free model upgrades, where the new query features are interoperable with the old gallery features. Despite the success, previous works only investigated a close-set training scenario (i.e., the new training set shares the same classes as the old one), and are limited by more realistic and challenging open-set scenarios. To this end, we first introduce a new problem of universal backward-compatible representation learning, covering all possible data split in model upgrades. We further propose a simple yet effective method, dubbed as Universal Backward-Compatible Training (UniBCT) with a novel structural prototype refinement algorithm, to learn compatible representations in all kinds of model upgrading benchmarks in a unified manner. Comprehensive experiments on the large-scale face recognition datasets MS1Mv3 and IJB-C fully demonstrate the effectiveness of our method.

Sijie Zhao, Yixiao Ge, Zhongang Qi et al. · tencent-ai

Stickers have become a ubiquitous part of modern-day communication, conveying complex emotions through visual imagery. To facilitate the development of more powerful algorithms for analyzing stickers, we propose a large-scale Chinese sticker dataset, namely Sticker820K, which consists of 820k image-text pairs. Each sticker has rich and high-quality textual annotations, including descriptions, optical characters, emotional labels, and style classifications. Although vision-language tasks in the domain of natural images have been well studied, directly applying the those models, such as CLIP, to sticker data is not an optimal solution due to the discrepant nature between natural and emotive image data. Therefore, we propose StickerCLIP as a benchmark model on the Sticker820K dataset. For the text-to-image retrieval task, our StickerCLIP demonstrates strong superiority over the CLIP, which achieves an absolute gain of 66.0\% in mean recall on the Sticker820K test set. Additionally, we endeavor to extend the recently popularized LLM by means of prompt tuning, integrating its ability for sticker retrieval and allowing users to retrieve stickers through instructions. We validate the feasibility of this method, demonstrating the immense potential of prompt tuning in expanding LLM abilities while not affecting the quality of upstream tasks.

Shupeng Su, Binjie Zhang, Yixiao Ge et al. · tencent-ai

The task of privacy-preserving model upgrades in image retrieval desires to reap the benefits of rapidly evolving new models without accessing the raw gallery images. A pioneering work introduced backward-compatible training, where the new model can be directly deployed in a backfill-free manner, i.e., the new query can be directly compared to the old gallery features. Despite a possible solution, its improvement in sequential model upgrades is gradually limited by the fixed and under-quality old gallery embeddings. To this end, we propose a new model upgrade paradigm, termed Bidirectional Compatible Training (BiCT), which will upgrade the old gallery embeddings by forward-compatible training towards the embedding space of the backward-compatible new model. We conduct comprehensive experiments to verify the prominent improvement by BiCT and interestingly observe that the inconspicuous loss weight of backward compatibility actually plays an essential role for both backward and forward retrieval performance. To summarize, we introduce a new and valuable problem named privacy-preserving model upgrades, with a proper solution BiCT. Several intriguing insights are further proposed to get the most out of our method.

Yukang Gan, Yixiao Ge, Chang Zhou et al. · tencent-ai

Large-scale embedding-based retrieval (EBR) is the cornerstone of search-related industrial applications. Given a user query, the system of EBR aims to identify relevant information from a large corpus of documents that may be tens or hundreds of billions in size. The storage and computation turn out to be expensive and inefficient with massive documents and high concurrent queries, making it difficult to further scale up. To tackle the challenge, we propose a binary embedding-based retrieval (BEBR) engine equipped with a recurrent binarization algorithm that enables customized bits per dimension. Specifically, we compress the full-precision query and document embeddings, formulated as float vectors in general, into a composition of multiple binary vectors using a lightweight transformation model with residual multilayer perception (MLP) blocks. We can therefore tailor the number of bits for different applications to trade off accuracy loss and cost savings. Importantly, we enable task-agnostic efficient training of the binarization model using a new embedding-to-embedding strategy. We also exploit the compatible training of binary embeddings so that the BEBR engine can support indexing among multiple embedding versions within a unified system. To further realize efficient search, we propose Symmetric Distance Calculation (SDC) to achieve lower response time than Hamming codes. We successfully employed the introduced BEBR to Tencent products, including Sogou, Tencent Video, QQ World, etc. The binarization algorithm can be seamlessly generalized to various tasks with multiple modalities. Extensive experiments on offline benchmarks and online A/B tests demonstrate the efficiency and effectiveness of our method, significantly saving 30%~50% index costs with almost no loss of accuracy at the system level.

Binjie Zhang, Shupeng Su, Yixiao Ge et al. · tencent-ai

The traditional model upgrading paradigm for retrieval requires recomputing all gallery embeddings before deploying the new model (dubbed as "backfilling"), which is quite expensive and time-consuming considering billions of instances in industrial applications. BCT presents the first step towards backward-compatible model upgrades to get rid of backfilling. It is workable but leaves the new model in a dilemma between new feature discriminativeness and new-to-old compatibility due to the undifferentiated compatibility constraints. In this work, we propose Darwinian Model Upgrades (DMU), which disentangle the inheritance and variation in the model evolving with selective backward compatibility and forward adaptation, respectively. The old-to-new heritable knowledge is measured by old feature discriminativeness, and the gallery features, especially those of poor quality, are evolved in a lightweight manner to become more adaptive in the new latent space. We demonstrate the superiority of DMU through comprehensive experiments on large-scale landmark retrieval and face recognition benchmarks. DMU effectively alleviates the new-to-new degradation and improves new-to-old compatibility, rendering a more proper model upgrading paradigm in large-scale retrieval systems.

Jun Zhang, Teng Wang, Yuying Ge et al.

This paper does not introduce a novel method but instead establishes a straightforward, incremental, yet essential baseline for video temporal grounding (VTG), a core capability in video understanding. While multimodal large language models (MLLMs) excel at various video understanding tasks, the recipes for optimizing them for VTG remain under-explored. In this paper, we present TimeLens, a systematic investigation into building MLLMs with strong VTG ability, along two primary dimensions: data quality and algorithmic design. We first expose critical quality issues in existing VTG benchmarks and introduce TimeLens-Bench, comprising meticulously re-annotated versions of three popular benchmarks with strict quality criteria. Our analysis reveals dramatic model re-rankings compared to legacy benchmarks, confirming the unreliability of prior evaluation standards. We also address noisy training data through an automated re-annotation pipeline, yielding TimeLens-100K, a large-scale, high-quality training dataset. Building on our data foundation, we conduct in-depth explorations of algorithmic design principles, yielding a series of meaningful insights and effective yet efficient practices. These include interleaved textual encoding for time representation, a thinking-free reinforcement learning with verifiable rewards (RLVR) approach as the training paradigm, and carefully designed recipes for RLVR training. These efforts culminate in TimeLens models, a family of MLLMs with state-of-the-art VTG performance among open-source models and even surpass proprietary models such as GPT-5 and Gemini-2.5-Flash. All codes, data, and models will be released to facilitate future research.

Yuying Ge, Sijie Zhao, Jinguo Zhu et al. · tencent-ai

The rapid evolution of multimodal foundation model has demonstrated significant progresses in vision-language understanding and generation, e.g., our previous work SEED-LLaMA. However, there remains a gap between its capability and the real-world applicability, primarily due to the model's limited capacity to effectively respond to various user instructions and interact with diverse visual data. In this work, we focus on bridging this gap through integrating two enhanced features: (1) comprehending images of arbitrary sizes and ratios, and (2) enabling multi-granularity image generation. We present a unified and versatile foundation model, namely, SEED-X, which is able to model multi-granularity visual semantics for comprehension and generation tasks. Besides the competitive results on public benchmarks, SEED-X demonstrates its effectiveness in handling real-world applications across various domains after instruction tuning. We hope that our work will inspire future research into what can be achieved by versatile multimodal foundation models in real-world applications. The models, codes, and datasets are released in https://github.com/AILab-CVC/SEED-X.

Ruyang Liu, Chen Li, Haoran Tang et al. · tencent-ai

Large Language Models (LLMs) have showcased impressive capabilities in text comprehension and generation, prompting research efforts towards video LLMs to facilitate human-AI interaction at the video level. However, how to effectively encode and understand videos in video-based dialogue systems remains to be solved. In this paper, we investigate a straightforward yet unexplored question: Can we feed all spatial-temporal tokens into the LLM, thus delegating the task of video sequence modeling to the LLMs? Surprisingly, this simple approach yields significant improvements in video understanding. Based upon this, we propose ST-LLM, an effective video-LLM baseline with Spatial-Temporal sequence modeling inside LLM. Furthermore, to address the overhead and stability issues introduced by uncompressed video tokens within LLMs, we develop a dynamic masking strategy with tailor-made training objectives. For particularly long videos, we have also designed a global-local input module to balance efficiency and effectiveness. Consequently, we harness LLM for proficient spatial-temporal modeling, while upholding efficiency and stability. Extensive experimental results attest to the effectiveness of our method. Through a more concise model and training pipeline, ST-LLM establishes a new state-of-the-art result on VideoChatGPT-Bench and MVBench. Codes have been available at https://github.com/TencentARC/ST-LLM.

Bohao Li, Yuying Ge, Yi Chen et al. · tencent-ai

Comprehending text-rich visual content is paramount for the practical application of Multimodal Large Language Models (MLLMs), since text-rich scenarios are ubiquitous in the real world, which are characterized by the presence of extensive texts embedded within images. Recently, the advent of MLLMs with impressive versatility has raised the bar for what we can expect from MLLMs. However, their proficiency in text-rich scenarios has yet to be comprehensively and objectively assessed, since current MLLM benchmarks primarily focus on evaluating general visual comprehension. In this work, we introduce SEED-Bench-2-Plus, a benchmark specifically designed for evaluating \textbf{text-rich visual comprehension} of MLLMs. Our benchmark comprises 2.3K multiple-choice questions with precise human annotations, spanning three broad categories: Charts, Maps, and Webs, each of which covers a wide spectrum of text-rich scenarios in the real world. These categories, due to their inherent complexity and diversity, effectively simulate real-world text-rich environments. We further conduct a thorough evaluation involving 34 prominent MLLMs (including GPT-4V, Gemini-Pro-Vision and Claude-3-Opus) and emphasize the current limitations of MLLMs in text-rich visual comprehension. We hope that our work can serve as a valuable addition to existing MLLM benchmarks, providing insightful observations and inspiring further research in the area of text-rich visual comprehension with MLLMs. The dataset and evaluation code can be accessed at https://github.com/AILab-CVC/SEED-Bench.

Weixian Lei, Yixiao Ge, Jianfeng Zhang et al.

Though the success of CLIP-based training recipes in vision-language models, their scalability to more modalities (e.g., 3D, audio, etc.) is limited to large-scale data, which is expensive or even inapplicable for rare modalities. In this paper, we present ViT-Lens that facilitates efficient omni-modal representation learning by perceiving novel modalities with a pretrained ViT and aligning to a pre-defined space. Specifically, the modality-specific lens is tuned to project multimodal signals to the shared embedding space, which are then processed by a strong ViT that carries pre-trained image knowledge. The encoded multimodal representations are optimized toward aligning with the modal-independent space, pre-defined by off-the-shelf foundation models. A well-trained lens with a ViT backbone has the potential to serve as one of these foundation models, supervising the learning of subsequent modalities. ViT-Lens provides a unified solution for representation learning of increasing modalities with two appealing benefits: (i) Exploiting the pretrained ViT across tasks and domains effectively with efficient data regime; (ii) Emergent downstream capabilities of novel modalities are demonstrated due to the modality alignment space. We evaluate ViT-Lens in the context of 3D as an initial verification. In zero-shot 3D classification, ViT-Lens achieves substantial improvements over previous state-of-the-art, showing 52.0% accuracy on Objaverse-LVIS, 87.4% on ModelNet40, and 60.6% on ScanObjectNN. Furthermore, we enable zero-shot 3D question-answering by simply integrating the trained 3D lens into the InstructBLIP model without any adaptation. We will release the results of ViT-Lens on more modalities in the near future.

Yuying Ge, Sijie Zhao, Chen Li et al. · tencent-ai

In this technical report, we introduce SEED-Data-Edit: a unique hybrid dataset for instruction-guided image editing, which aims to facilitate image manipulation using open-form language. SEED-Data-Edit is composed of three distinct types of data: (1) High-quality editing data produced by an automated pipeline, ensuring a substantial volume of diverse image editing pairs. (2) Real-world scenario data collected from the internet, which captures the intricacies of user intentions for promoting the practical application of image editing in the real world. (3) High-precision multi-turn editing data annotated by humans, which involves multiple rounds of edits for simulating iterative editing processes. The combination of these diverse data sources makes SEED-Data-Edit a comprehensive and versatile dataset for training language-guided image editing model. We fine-tune a pretrained Multimodal Large Language Model (MLLM) that unifies comprehension and generation with SEED-Data-Edit. The instruction tuned model demonstrates promising results, indicating the potential and effectiveness of SEED-Data-Edit in advancing the field of instructional image editing. The datasets are released in https://huggingface.co/datasets/AILab-CVC/SEED-Data-Edit.

Haokun Lin, Teng Wang, Yixiao Ge et al.

Pioneering token-based works such as Chameleon and Emu3 have established a foundation for multimodal unification but face challenges of high training computational overhead and limited comprehension performance due to a lack of high-level semantics. In this paper, we introduce TokLIP, a visual tokenizer that enhances comprehension by semanticizing vector-quantized (VQ) tokens and incorporating CLIP-level semantics while enabling end-to-end multimodal autoregressive training with standard VQ tokens. TokLIP integrates a low-level discrete VQ tokenizer with a ViT-based token encoder to capture high-level continuous semantics. Unlike previous approaches (e.g., VILA-U) that discretize high-level features, TokLIP disentangles training objectives for comprehension and generation, allowing the direct application of advanced VQ tokenizers without the need for tailored quantization operations. Our empirical results demonstrate that TokLIP achieves exceptional data efficiency, empowering visual tokens with high-level semantic understanding while enhancing low-level generative capacity, making it well-suited for autoregressive Transformers in both comprehension and generation tasks. The code and models are available at https://github.com/TencentARC/TokLIP.

Junhao Cheng, Yuying Ge, Teng Wang et al.

Recent advances in CoT reasoning and RL post-training have been reported to enhance video reasoning capabilities of MLLMs. This progress naturally raises a question: can these models perform complex video reasoning in a manner comparable to human experts? However, existing video benchmarks primarily evaluate visual perception and grounding abilities, with questions that can be answered based on explicit prompts or isolated visual cues. Such benchmarks do not fully capture the intricacies of real-world reasoning, where humans must actively search for, integrate, and analyze multiple clues before reaching a conclusion. To address this issue, we present Video-Holmes, a benchmark inspired by the reasoning process of Sherlock Holmes, designed to evaluate the complex video reasoning capabilities of MLLMs. Video-Holmes consists of 1,837 questions derived from 270 manually annotated suspense short films, which spans seven carefully designed tasks. Each task is constructed by first identifying key events and causal relationships within films, and then designing questions that require models to actively locate and connect multiple relevant visual clues scattered across different video segments. Our comprehensive evaluation of state-of-the-art MLLMs reveals that, while these models generally excel at visual perception, they encounter substantial difficulties with integrating information and often miss critical clues. For example, the best-performing model, Gemini-2.5-Pro, achieves an accuracy of only 45%, with most models scoring below 40%. We aim that Video-Holmes can serve as a "Holmes-test" for multimodal reasoning, motivating models to reason more like humans and emphasizing the ongoing challenges in this field. The benchmark is released in https://github.com/TencentARC/Video-Holmes.

Lu Qiu, Yi Chen, Yuying Ge et al. · tencent-ai

The advent of Multimodal Large Language Models, leveraging the power of Large Language Models, has recently demonstrated superior multimodal understanding and reasoning abilities, heralding a new era for artificial general intelligence. However, achieving AGI necessitates more than just comprehension and reasoning. A crucial capability required is effective planning in diverse scenarios, which involves making reasonable decisions based on complex environments to solve real-world problems. Despite its importance, the planning abilities of current MLLMs in varied scenarios remain underexplored. In this paper, we introduce EgoPlan-Bench2, a rigorous and comprehensive benchmark designed to assess the planning capabilities of MLLMs across a wide range of real-world scenarios. EgoPlan-Bench2 encompasses everyday tasks spanning 4 major domains and 24 detailed scenarios, closely aligned with human daily life. EgoPlan-Bench2 is constructed through a semi-automatic process utilizing egocentric videos, complemented by manual verification. Grounded in a first-person perspective, it mirrors the way humans approach problem-solving in everyday life. We evaluate 21 competitive MLLMs and provide an in-depth analysis of their limitations, revealing that they face significant challenges in real-world planning. To further improve the planning proficiency of current MLLMs, we propose a training-free approach using multimodal Chain-of-Thought (CoT) prompting through investigating the effectiveness of various multimodal prompts in complex planning. Our approach enhances the performance of GPT-4V by 10.24 on EgoPlan-Bench2 without additional training. Our work not only sheds light on the current limitations of MLLMs in planning, but also provides insights for future enhancements in this critical area. We have made data and code available at https://qiulu66.github.io/egoplanbench2/.

Ruyang Liu, Haoran Tang, Haibo Liu et al. · tencent-ai

The past year has witnessed the significant advancement of video-based large language models. However, the challenge of developing a unified model for both short and long video understanding remains unresolved. Most existing video LLMs cannot handle hour-long videos, while methods custom for long videos tend to be ineffective for shorter videos and images. In this paper, we identify the key issue as the redundant content in videos. To address this, we propose a novel pooling strategy that simultaneously achieves token compression and instruction-aware visual feature aggregation. Our model is termed Prompt-guided Pooling LLaVA, or PPLLaVA for short. Specifically, PPLLaVA consists of three core components: the CLIP-based visual-prompt alignment that extracts visual information relevant to the user's instructions, the prompt-guided pooling that compresses the visual sequence to arbitrary scales using convolution-style pooling, and the clip context extension designed for lengthy prompt common in visual dialogue. Moreover, our codebase also integrates the most advanced video Direct Preference Optimization (DPO) and visual interleave training. Extensive experiments have validated the performance of our model. With superior throughput and only 1024 visual context, PPLLaVA achieves better results on image benchmarks as a video LLM, while achieving state-of-the-art performance across various video benchmarks, excelling in tasks ranging from caption generation to multiple-choice questions, and handling video lengths from seconds to hours. Codes have been available at https://github.com/farewellthree/PPLLaVA.

Fengyuan Shi, Zhuoyan Luo, Yixiao Ge et al. · tencent-ai

Existing vector quantization (VQ) methods struggle with scalability, largely attributed to the instability of the codebook that undergoes partial updates during training. The codebook is prone to collapse as utilization decreases, due to the progressively widening distribution gap between non-activated codes and visual features. To solve the problem, we propose Index Backpropagation Quantization (IBQ), a new VQ method for the joint optimization of all codebook embeddings and the visual encoder. Applying a straight-through estimator on the one-hot categorical distribution between the encoded feature and codebook, all codes are differentiable and maintain a consistent latent space with the visual encoder. IBQ enables scalable training of visual tokenizers and, for the first time, achieves a large-scale codebook ($2^{18}$) with high dimension ($256$) and high utilization. Experiments on the standard ImageNet benchmark demonstrate the scalability and superiority of IBQ, achieving competitive results on reconstruction and the application of autoregressive visual generation. The code and models are available at https://github.com/TencentARC/SEED-Voken.

Yicheng Xiao, Lin Song, Rui Yang et al. · tsinghua

With the advancement of language models, unified multimodal understanding and generation have made significant strides, with model architectures evolving from separated components to unified single-model frameworks. This paper explores an efficient training paradigm to build a single transformer for unified multimodal understanding and generation. Specifically, we propose a multimodal warmup strategy utilizing prior knowledge to extend capabilities. To address cross-modal compatibility challenges, we introduce feature pre-scaling and multimodal AdaLN techniques. Integrating the proposed technologies, we present the HaploOmni, a new single multimodal transformer. With limited training costs, HaploOmni achieves competitive performance across multiple image and video understanding and generation benchmarks over advanced unified models. All codes will be made public at https://github.com/Tencent/HaploVLM.

Junhao Cheng, Yuying Ge, Yixiao Ge et al.

Recent advancements in image and video synthesis have opened up new promise in generative games. One particularly intriguing application is transforming characters from anime films into interactive, playable entities. This allows players to immerse themselves in the dynamic anime world as their favorite characters for life simulation through language instructions. Such games are defined as infinite game since they eliminate predetermined boundaries and fixed gameplay rules, where players can interact with the game world through open-ended language and experience ever-evolving storylines and environments. Recently, a pioneering approach for infinite anime life simulation employs large language models (LLMs) to translate multi-turn text dialogues into language instructions for image generation. However, it neglects historical visual context, leading to inconsistent gameplay. Furthermore, it only generates static images, failing to incorporate the dynamics necessary for an engaging gaming experience. In this work, we propose AnimeGamer, which is built upon Multimodal Large Language Models (MLLMs) to generate each game state, including dynamic animation shots that depict character movements and updates to character states, as illustrated in Figure 1. We introduce novel action-aware multimodal representations to represent animation shots, which can be decoded into high-quality video clips using a video diffusion model. By taking historical animation shot representations as context and predicting subsequent representations, AnimeGamer can generate games with contextual consistency and satisfactory dynamics. Extensive evaluations using both automated metrics and human evaluations demonstrate that AnimeGamer outperforms existing methods in various aspects of the gaming experience. Codes and checkpoints are available at https://github.com/TencentARC/AnimeGamer.

Tianheng Cheng, Lin Song, Yixiao Ge et al. · tencent-ai

The You Only Look Once (YOLO) series of detectors have established themselves as efficient and practical tools. However, their reliance on predefined and trained object categories limits their applicability in open scenarios. Addressing this limitation, we introduce YOLO-World, an innovative approach that enhances YOLO with open-vocabulary detection capabilities through vision-language modeling and pre-training on large-scale datasets. Specifically, we propose a new Re-parameterizable Vision-Language Path Aggregation Network (RepVL-PAN) and region-text contrastive loss to facilitate the interaction between visual and linguistic information. Our method excels in detecting a wide range of objects in a zero-shot manner with high efficiency. On the challenging LVIS dataset, YOLO-World achieves 35.4 AP with 52.0 FPS on V100, which outperforms many state-of-the-art methods in terms of both accuracy and speed. Furthermore, the fine-tuned YOLO-World achieves remarkable performance on several downstream tasks, including object detection and open-vocabulary instance segmentation.

Chengyue Wu, Yixiao Ge, Qiushan Guo et al.

The remarkable progress of Multi-modal Large Language Models (MLLMs) has attracted significant attention due to their superior performance in visual contexts. However, their capabilities in turning visual figure to executable code, have not been evaluated thoroughly. To address this, we introduce Plot2Code, a comprehensive visual coding benchmark designed for a fair and in-depth assessment of MLLMs. We carefully collect 132 manually selected high-quality matplotlib plots across six plot types from publicly available matplotlib galleries. For each plot, we carefully offer its source code, and an descriptive instruction summarized by GPT-4. This approach enables Plot2Code to extensively evaluate MLLMs' code capabilities across various input modalities. Furthermore, we propose three automatic evaluation metrics, including code pass rate, text-match ratio, and GPT-4V overall rating, for a fine-grained assessment of the output code and rendered images. Instead of simply judging pass or fail, we employ GPT-4V to make an overall judgement between the generated and reference images, which has been shown to be consistent with human evaluation. The evaluation results, which include analyses of 14 MLLMs such as the proprietary GPT-4V, Gemini-Pro, and the open-sourced Mini-Gemini, highlight the substantial challenges presented by Plot2Code. With Plot2Code, we reveal that most existing MLLMs struggle with visual coding for text-dense plots, heavily relying on textual instruction. We hope that the evaluation results from Plot2Code on visual coding will guide the future development of MLLMs. All data involved with Plot2Code are available at https://huggingface.co/datasets/TencentARC/Plot2Code.