Despite the development of various AI systems to support learning in various domains, AI assistance for art appreciation education has not been extensively explored. Art appreciation, often perceived as an unfamiliar and challenging endeavor for most students, can be more accessible with a generative AI enabled conversation partner that provides tailored questions and encourages the audience to deeply appreciate artwork. This study explores the application of multimodal large language models (MLLMs) in art appreciation education, with a focus on developing LLaVA-Docent, a model designed to serve as a personal tutor for art appreciation. Our approach involved design and development research, focusing on iterative enhancement to design and develop the application to produce a functional MLLM-enabled chatbot along with a data design framework for art appreciation education. To that end, we established a virtual dialogue dataset that was generated by GPT-4, which was instrumental in training our MLLM, LLaVA-Docent. The performance of LLaVA-Docent was evaluated by benchmarking it against alternative settings and revealed its distinct strengths and weaknesses. Our findings highlight the efficacy of the MMLM-based personalized art appreciation chatbot and demonstrate its applicability for a novel approach in which art appreciation is taught and experienced.
This paper focuses on jailbreaking attacks against multi-modal large language models (MLLMs), seeking to elicit MLLMs to generate objectionable responses to harmful user queries. A maximum likelihood-based algorithm is proposed to find an \emph{image Jailbreaking Prompt} (imgJP), enabling jailbreaks against MLLMs across multiple unseen prompts and images (i.e., data-universal property). Our approach exhibits strong model-transferability, as the generated imgJP can be transferred to jailbreak various models, including MiniGPT-v2, LLaVA, InstructBLIP, and mPLUG-Owl2, in a black-box manner. Moreover, we reveal a connection between MLLM-jailbreaks and LLM-jailbreaks. As a result, we introduce a construction-based method to harness our approach for LLM-jailbreaks, demonstrating greater efficiency than current state-of-the-art methods. The code is available here. \textbf{Warning: some content generated by language models may be offensive to some readers.}
Knowledge-based visual question answering (VQA) requires world knowledge beyond the image for accurate answer. Recently, instead of extra knowledge bases, a large language model (LLM) like GPT-3 is activated as an implicit knowledge engine to jointly acquire and reason the necessary knowledge for answering by converting images into textual information (e.g., captions and answer candidates). However, such conversion may introduce irrelevant information, which causes the LLM to misinterpret images and ignore visual details crucial for accurate knowledge. We argue that multimodal large language model (MLLM) is a better implicit knowledge engine than the LLM for its superior capability of visual understanding. Despite this, how to activate the capacity of MLLM as the implicit knowledge engine has not been explored yet. Therefore, we propose GeReA, a generate-reason framework that prompts a MLLM like InstructBLIP with question relevant vision and language information to generate knowledge-relevant descriptions and reasons those descriptions for knowledge-based VQA. Specifically, the question-relevant image regions and question-specific manual prompts are encoded in the MLLM to generate the knowledge relevant descriptions, referred to as question-aware prompt captions. After that, the question-aware prompt captions, image-question pair, and similar samples are sent into the multi-modal reasoning model to learn a joint knowledge-image-question representation for answer prediction. GeReA unlocks the use of MLLM as the implicit knowledge engine, surpassing all previous state-of-the-art methods on OK-VQA and A-OKVQA datasets, with test accuracies of 66.5% and 63.3% respectively. Our code will be released at https://github.com/Upper9527/GeReA.
Instruction finetuning on a variety of image-text instruction data is the key to obtaining a versatile Multimodal Large Language Model (MLLM), and different configurations of the instruction data can lead to finetuned models with different capabilities. However, we have discovered that data conflicts are inevitable when mixing instruction data from distinct domains, which can result in performance drops for tasks of a specific domain. To address this issue, we propose to apply an efficient Mixture of Experts (MoE) design, which is a sparse Mixture of LoRA Experts (MoLE) for instruction finetuning MLLMs. Within the Transformer layers, we extend the popular Low-Rank Adaption (LoRA) method by creating a set of LoRA experts specifically for the MLP layer, and route each token to the top-1 expert based on a routing function, allowing adaptive choices for tokens from different domains. Since the LoRA experts are sparsely activated, the training and inference cost are kept roughly constant compared to the original LoRA method. By replacing the plain-LoRA of LLaVA-1.5 with our MoE design, our final model is named LLaVA-MoLE. Extensive experiments proved that LLaVA-MoLE effectively mitigates the data conflict issue when mixing multiple distinct instruction datasets with various configurations, and achieves consistent performance gains over the strong plain-LoRA baselines. Most importantly, on the mixed datasets, LLaVA-MoLE can even outperform the plain-LoRA baseline trained with twice the samples.
Existing text-to-image diffusion models primarily generate images from text prompts. However, the inherent conciseness of textual descriptions poses challenges in faithfully synthesizing images with intricate details, such as specific entities or scenes. This paper presents UNIMO-G, a simple multimodal conditional diffusion framework that operates on multimodal prompts with interleaved textual and visual inputs, which demonstrates a unified ability for both text-driven and subject-driven image generation. UNIMO-G comprises two core components: a Multimodal Large Language Model (MLLM) for encoding multimodal prompts, and a conditional denoising diffusion network for generating images based on the encoded multimodal input. We leverage a two-stage training strategy to effectively train the framework: firstly pre-training on large-scale text-image pairs to develop conditional image generation capabilities, and then instruction tuning with multimodal prompts to achieve unified image generation proficiency. A well-designed data processing pipeline involving language grounding and image segmentation is employed to construct multi-modal prompts. UNIMO-G excels in both text-to-image generation and zero-shot subject-driven synthesis, and is notably effective in generating high-fidelity images from complex multimodal prompts involving multiple image entities.
Multimodal large language models (MLLM) have achieved satisfactory results in many tasks. However, their performance in the task of ReID (ReID) has not been explored to date. This paper will investigate how to adapt them for the task of ReID. An intuitive idea is to fine-tune MLLM with ReID image-text datasets, and then use their visual encoder as a backbone for ReID. However, there still exist two apparent issues: (1) Designing instructions for ReID, MLLMs may overfit specific instructions, and designing a variety of instructions will lead to higher costs. (2) When fine-tuning the visual encoder of a MLLM, it is not trained synchronously with the ReID task. As a result, the effectiveness of the visual encoder fine-tuning cannot be directly reflected in the performance of the ReID task. To address these problems, this paper proposes MLLMReID: Multimodal Large Language Model-based ReID. Firstly, we proposed Common Instruction, a simple approach that leverages the essence ability of LLMs to continue writing, avoiding complex and diverse instruction design. Secondly, we propose a multi-task learning-based synchronization module to ensure that the visual encoder of the MLLM is trained synchronously with the ReID task. The experimental results demonstrate the superiority of our method.
Recently, the astonishing performance of large language models (LLMs) in natural language comprehension and generation tasks triggered lots of exploration of using them as central controllers to build agent systems. Multiple studies focus on bridging the LLMs to external tools to extend the application scenarios. However, the current LLMs' perceiving tool-use ability is limited to a single text query, which may result in ambiguity in understanding the users' real intentions. LLMs are expected to eliminate that by perceiving the visual- or auditory-grounded instructions' information. Therefore, in this paper, we propose MLLM-Tool, a system incorporating open-source LLMs and multi-modal encoders so that the learnt LLMs can be conscious of multi-modal input instruction and then select the function-matched tool correctly. To facilitate the evaluation of the model's capability, we collect a dataset featured by consisting of multi-modal input tools from HuggingFace. Another important feature of our dataset is that our dataset also contains multiple potential choices for the same instruction due to the existence of identical functions and synonymous functions, which provides more potential solutions for the same query. The experiments reveal that our MLLM-Tool is capable of recommending appropriate tools for multi-modal instructions. Codes and data are available at https://github.com/MLLM-Tool/MLLM-Tool.
Multimodal Large Language Models (MLLMs) have demonstrated proficiency in handling a variety of visual-language tasks. However, current MLLM benchmarks are predominantly designed to evaluate reasoning based on static information about a single image, and the ability of modern MLLMs to extrapolate from image sequences, which is essential for understanding our ever-changing world, has been less investigated. To address this challenge, this paper introduces Mementos, a new benchmark designed to assess MLLMs' sequential image reasoning abilities. Mementos features 4,761 diverse image sequences with varying lengths. We also employ a GPT-4 assisted method to evaluate MLLM reasoning performance. Through a careful evaluation of nine recent MLLMs on Mementos, including GPT-4V and Gemini, we find that they struggle to accurately describe dynamic information about given image sequences, often leading to hallucinations/misrepresentations of objects and their corresponding behaviors. Our quantitative analysis and case studies identify three key factors impacting MLLMs' sequential image reasoning: the correlation between object and behavioral hallucinations, the influence of cooccurring behaviors, and the compounding impact of behavioral hallucinations. Our dataset is available at https://github.com/umd-huang-lab/Mementos.
Leveraging multimodal data is an inherent requirement for comprehending geographic objects. However, due to the high heterogeneity in structure and semantics among various spatio-temporal modalities, the joint interpretation of multimodal spatio-temporal data has long been an extremely challenging problem. The primary challenge resides in striking a trade-off between the cohesion and autonomy of diverse modalities. This trade-off becomes progressively nonlinear as the number of modalities expands. Inspired by the human cognitive system and linguistic philosophy, where perceptual signals from the five senses converge into language, we introduce the Language as Reference Framework (LaRF), a fundamental principle for constructing a multimodal unified model. Building upon this, we propose AllSpark, a multimodal spatio-temporal general artificial intelligence model. Our model integrates ten different modalities into a unified framework. To achieve modal cohesion, AllSpark introduces a modal bridge and multimodal large language model (LLM) to map diverse modal features into the language feature space. To maintain modality autonomy, AllSpark uses modality-specific encoders to extract the tokens of various spatio-temporal modalities. Finally, observing a gap between the model's interpretability and downstream tasks, we designed modality-specific prompts and task heads, enhancing the model's generalization capability across specific tasks. Experiments indicate that the incorporation of language enables AllSpark to excel in few-shot classification tasks for RGB and point cloud modalities without additional training, surpassing baseline performance by up to 41.82\%. The source code is available at https://github.com/GeoX-Lab/AllSpark.
Robot manipulation relies on accurately predicting contact points and end-effector directions to ensure successful operation. However, learning-based robot manipulation, trained on a limited category within a simulator, often struggles to achieve generalizability, especially when confronted with extensive categories. Therefore, we introduce an innovative approach for robot manipulation that leverages the robust reasoning capabilities of Multimodal Large Language Models (MLLMs) to enhance the stability and generalization of manipulation. By fine-tuning the injected adapters, we preserve the inherent common sense and reasoning ability of the MLLMs while equipping them with the ability for manipulation. The fundamental insight lies in the introduced fine-tuning paradigm, encompassing object category understanding, affordance prior reasoning, and object-centric pose prediction to stimulate the reasoning ability of MLLM in manipulation. During inference, our approach utilizes an RGB image and text prompt to predict the end effector's pose in chain of thoughts. After the initial contact is established, an active impedance adaptation policy is introduced to plan the upcoming waypoints in a closed-loop manner. Moreover, in real world, we design a test-time adaptation (TTA) strategy for manipulation to enable the model better adapt to the current real-world scene configuration. Experiments in simulator and real-world show the promising performance of ManipLLM. More details and demonstrations can be found at https://sites.google.com/view/manipllm.
In recent years, multimodal large language models (MLLMs) such as GPT-4V have demonstrated remarkable advancements, excelling in a variety of vision-language tasks. Despite their prowess, the closed-source nature and computational demands of such models limit their accessibility and applicability. This study introduces TinyGPT-V, a novel open-source MLLM, designed for efficient training and inference across various vision-language tasks, including image captioning (IC) and visual question answering (VQA). Leveraging a compact yet powerful architecture, TinyGPT-V integrates the Phi-2 language model with pre-trained vision encoders, utilizing a unique mapping module for visual and linguistic information fusion. With a training regimen optimized for small backbones and employing a diverse dataset amalgam, TinyGPT-V requires significantly lower computational resources 24GB for training and as little as 8GB for inference without compromising on performance. Our experiments demonstrate that TinyGPT-V, with its language model 2.8 billion parameters, achieves comparable results in VQA and image inference tasks to its larger counterparts while being uniquely suited for deployment on resource-constrained devices through innovative quantization techniques. This work not only paves the way for more accessible and efficient MLLMs but also underscores the potential of smaller, optimized models in bridging the gap between high performance and computational efficiency in real-world applications. Additionally, this paper introduces a new approach to multimodal large language models using smaller backbones. Our code and training weights are available in the supplementary material.
Scalable Vector Graphics (SVGs) are vital for modern image rendering due to their scalability and versatility. Previous SVG generation methods have focused on curve-based vectorization, lacking semantic understanding, often producing artifacts, and struggling with SVG primitives beyond path curves. To address these issues, we introduce StarVector, a multimodal large language model for SVG generation. It performs image vectorization by understanding image semantics and using SVG primitives for compact, precise outputs. Unlike traditional methods, StarVector works directly in the SVG code space, leveraging visual understanding to apply accurate SVG primitives. To train StarVector, we create SVG-Stack, a diverse dataset of 2M samples that enables generalization across vectorization tasks and precise use of primitives like ellipses, polygons, and text. We address challenges in SVG evaluation, showing that pixel-based metrics like MSE fail to capture the unique qualities of vector graphics. We introduce SVG-Bench, a benchmark across 10 datasets, and 3 tasks: Image-to-SVG, Text-to-SVG generation, and diagram generation. Using this setup, StarVector achieves state-of-the-art performance, producing more compact and semantically rich SVGs.
This paper presents Audio-Visual LLM, a Multimodal Large Language Model that takes both visual and auditory inputs for holistic video understanding. A key design is the modality-augmented training, which involves the integration of modality-specific tokens engineered to activate the appropriate visual and/or auditory encoder selectively. This mechanism is pivotal in enabling end-to-end joint training with video data at different modalities, including visual-only, audio-only, and audio-visual formats. Moreover, we introduce a high-quality video instruction dataset, derived from GPT-4. This dataset allows Audio-Visual LLM to adeptly process a variety of task-oriented video instructions, ranging from multi-turn conversations and audio-visual narratives to complex reasoning tasks. Extensive experiments demonstrate that Audio-Visual LLM impressively achieves strong zero-shot results across a range of video understanding tasks. For example, Audio-Visual LLM achieves an accuracy of 53.7% on MSRVTT-QA, outperforming non-LLM-based InterVideo by 6.6% and LLM-based Valley by 4.4%, respectively. Additionally, our Audio-Visual LLM also achieves competitive performance on audio tasks (e.g., AudioCaps).
Multi-modal large language models (MLLMs) have been shown to efficiently integrate natural language with visual information to handle multi-modal tasks. However, MLLMs still face a fundamental limitation of hallucinations, where they tend to generate erroneous or fabricated information. In this paper, we address hallucinations in MLLMs from a novel perspective of representation learning. We first analyzed the representation distribution of textual and visual tokens in MLLM, revealing two important findings: 1) there is a significant gap between textual and visual representations, indicating unsatisfactory cross-modal representation alignment; 2) representations of texts that contain and do not contain hallucinations are entangled, making it challenging to distinguish them. These two observations inspire us with a simple yet effective method to mitigate hallucinations. Specifically, we introduce contrastive learning into MLLMs and use text with hallucination as hard negative examples, naturally bringing representations of non-hallucinative text and visual samples closer while pushing way representations of non-hallucinating and hallucinative text. We evaluate our method quantitatively and qualitatively, showing its effectiveness in reducing hallucination occurrences and improving performance across multiple benchmarks. On the MMhal-Bench benchmark, our method obtains a 34.66% /29.5% improvement over the baseline MiniGPT-4/LLaVA. Our code is available on https://github.com/X-PLUG/mPLUG-HalOwl/tree/main/hacl.
Early weakly supervised video grounding (WSVG) methods often struggle with incomplete boundary detection due to the absence of temporal boundary annotations. To bridge the gap between video-level and boundary-level annotation, explicit-supervision methods, i.e., generating pseudo-temporal boundaries for training, have achieved great success. However, data augmentations in these methods might disrupt critical temporal information, yielding poor pseudo boundaries. In this paper, we propose a new perspective that maintains the integrity of the original temporal content while introducing more valuable information for expanding the incomplete boundaries. To this end, we propose EtC (Expand then Clarify), first use the additional information to expand the initial incomplete pseudo boundaries, and subsequently refine these expanded ones to achieve precise boundaries. Motivated by video continuity, i.e., visual similarity across adjacent frames, we use powerful multimodal large language models (MLLMs) to annotate each frame within initial pseudo boundaries, yielding more comprehensive descriptions for expanded boundaries. To further clarify the noise of expanded boundaries, we combine mutual learning with a tailored proposal-level contrastive objective to use a learnable approach to harmonize a balance between incomplete yet clean (initial) and comprehensive yet noisy (expanded) boundaries for more precise ones. Experiments demonstrate the superiority of our method on two challenging WSVG datasets.
Multimodal Large Language Models (MLLMs) have shown success in various general image processing tasks, yet their application in medical imaging is nascent, lacking tailored models. This study investigates the potential of MLLMs in improving the understanding and generation of Chest X-Rays (CXRs). We introduce MedXChat, a unified framework facilitating seamless interactions between medical assistants and users for diverse CXR tasks, including text report generation, visual question-answering (VQA), and Text-to-CXR generation. Our MLLMs using natural language as the input breaks task boundaries, maximally simplifying medical professional training by allowing diverse tasks within a single environment. For CXR understanding, we leverage powerful off-the-shelf visual encoders (e.g., ViT) and LLMs (e.g., mPLUG-Owl) to convert medical imagery into language-like features, and subsequently fine-tune our large pre-trained models for medical applications using a visual adapter network and a delta-tuning approach. For CXR generation, we introduce an innovative synthesis approach that utilizes instruction-following capabilities within the Stable Diffusion (SD) architecture. This technique integrates smoothly with the existing model framework, requiring no extra parameters, thereby maintaining the SD's generative strength while also bestowing upon it the capacity to render fine-grained medical images with high fidelity. Through comprehensive experiments, our model demonstrates exceptional cross-task adaptability, displaying adeptness across all three defined tasks. Our MedXChat model and the instruction dataset utilized in this research will be made publicly available to encourage further exploration in the field.
This work proposes TimeChat, a time-sensitive multimodal large language model specifically designed for long video understanding. Our model incorporates two key architectural contributions: (1) a timestamp-aware frame encoder that binds visual content with the timestamp of each frame, and (2) a sliding video Q-Former that produces a video token sequence of varying lengths to accommodate videos of various durations. Additionally, we construct an instruction-tuning dataset, encompassing 6 tasks and a total of 125K instances, to further enhance TimeChat's instruction-following performance. Experiment results across various video understanding tasks, such as dense captioning, temporal grounding, and highlight detection, demonstrate TimeChat's strong zero-shot temporal localization and reasoning capabilities. For example, it achieves +9.2 F1 score and +2.8 CIDEr on YouCook2, +5.8 HIT@1 on QVHighlights, and +27.5 R@1 (IoU=0.5) on Charades-STA, compared to state-of-the-art video large language models, holding the potential to serve as a versatile video assistant for long-form video comprehension tasks and satisfy realistic user requirements.
A big convergence of language, multimodal perception, action, and world modeling is a key step toward artificial general intelligence. In this work, we introduce Kosmos-1, a Multimodal Large Language Model (MLLM) that can perceive general modalities, learn in context (i.e., few-shot), and follow instructions (i.e., zero-shot). Specifically, we train Kosmos-1 from scratch on web-scale multimodal corpora, including arbitrarily interleaved text and images, image-caption pairs, and text data. We evaluate various settings, including zero-shot, few-shot, and multimodal chain-of-thought prompting, on a wide range of tasks without any gradient updates or finetuning. Experimental results show that Kosmos-1 achieves impressive performance on (i) language understanding, generation, and even OCR-free NLP (directly fed with document images), (ii) perception-language tasks, including multimodal dialogue, image captioning, visual question answering, and (iii) vision tasks, such as image recognition with descriptions (specifying classification via text instructions). We also show that MLLMs can benefit from cross-modal transfer, i.e., transfer knowledge from language to multimodal, and from multimodal to language. In addition, we introduce a dataset of Raven IQ test, which diagnoses the nonverbal reasoning capability of MLLMs.
Recently, the strong text creation ability of Large Language Models(LLMs) has given rise to many tools for assisting paper reading or even writing. However, the weak diagram analysis abilities of LLMs or Multimodal LLMs greatly limit their application scenarios, especially for scientific academic paper writing. In this work, towards a more versatile copilot for academic paper writing, we mainly focus on strengthening the multi-modal diagram analysis ability of Multimodal LLMs. By parsing Latex source files of high-quality papers, we carefully build a multi-modal diagram understanding dataset M-Paper. By aligning diagrams in the paper with related paragraphs, we construct professional diagram analysis samples for training and evaluation. M-Paper is the first dataset to support joint comprehension of multiple scientific diagrams, including figures and tables in the format of images or Latex codes. Besides, to better align the copilot with the user's intention, we introduce the `outline' as the control signal, which could be directly given by the user or revised based on auto-generated ones. Comprehensive experiments with a state-of-the-art Mumtimodal LLM demonstrate that training on our dataset shows stronger scientific diagram understanding performance, including diagram captioning, diagram analysis, and outline recommendation. The dataset, code, and model are available at https://github.com/X-PLUG/mPLUG-DocOwl/tree/main/PaperOwl.
We present CoDi-2, a versatile and interactive Multimodal Large Language Model (MLLM) that can follow complex multimodal interleaved instructions, conduct in-context learning (ICL), reason, chat, edit, etc., in an any-to-any input-output modality paradigm. By aligning modalities with language for both encoding and generation, CoDi-2 empowers Large Language Models (LLMs) to not only understand complex modality-interleaved instructions and in-context examples, but also autoregressively generate grounded and coherent multimodal outputs in the continuous feature space. To train CoDi-2, we build a large-scale generation dataset encompassing in-context multimodal instructions across text, vision, and audio. CoDi-2 demonstrates a wide range of zero-shot capabilities for multimodal generation, such as in-context learning, reasoning, and compositionality of any-to-any modality generation through multi-round interactive conversation. CoDi-2 surpasses previous domain-specific models on tasks such as subject-driven image generation, vision transformation, and audio editing. CoDi-2 signifies a substantial breakthrough in developing a comprehensive multimodal foundation model adept at interpreting in-context language-vision-audio interleaved instructions and producing multimodal outputs.