LongLLaVA: Scaling Multi-modal LLMs to 1000 Images Efficiently via a Hybrid Architecture

• URL: http://arxiv.org/abs/2409.02889v3
• Date: Mon, 22 Sep 2025 19:16:45 GMT
• Title: LongLLaVA: Scaling Multi-modal LLMs to 1000 Images Efficiently via a Hybrid Architecture
• Authors: Xidong Wang, Dingjie Song, Shunian Chen, Junyin Chen, Zhenyang Cai, Chen Zhang, Lichao Sun, Benyou Wang,
• Abstract summary: LongLLaVA achieves competitive results across various benchmarks while maintaining high throughput and low memory consumption.<n> Notably, it can process nearly one thousand images on a single A100 80GB GPU, underscoring its potential for a wide range of multi-modal applications.
• Score: 42.12601112927263
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Expanding the long-context capabilities of Multi-modal Large Language Models~(MLLMs) is critical for advancing video understanding and high-resolution image analysis. Achieving this requires systematic improvements in model architecture, data construction, and training strategies, particularly to address challenges such as performance degradation with increasing image counts and high computational costs. In this paper, we propose a hybrid architecture that integrates Mamba and Transformer blocks, introduce data construction methods that capture both temporal and spatial dependencies, and employ a progressive training strategy. Our released model, LongLLaVA (\textbf{Long}-Context \textbf{L}arge \textbf{L}anguage \textbf{a}nd \textbf{V}ision \textbf{A}ssistant), demonstrates an effective balance between efficiency and performance. LongLLaVA achieves competitive results across various benchmarks while maintaining high throughput and low memory consumption. Notably, it can process nearly one thousand images on a single A100 80GB GPU, underscoring its potential for a wide range of multi-modal applications.

Related papers

• MiniCPM-SALA: Hybridizing Sparse and Linear Attention for Efficient Long-Context Modeling [80.48332380100915]
  MiniCPM-SALA is a hybrid model that integrates the high-fidelity long-context modeling of sparse attention with the global efficiency of linear attention.<n>On a single NVIDIA A6000D GPU, the model achieves up to 3.5x the inference speed of the full-attention model at the sequence length of 256K tokens.
  arXiv  Detail & Related papers  (2026-02-12T09:37:05Z)
• Co-Training Vision Language Models for Remote Sensing Multi-task Learning [68.15604397741753]
  Vision language models (VLMs) have achieved promising results in RS image understanding, grounding, and ultra-high-resolution (UHR) image reasoning.<n>We present RSCoVLM, a simple yet flexible VLM baseline for RS MTL.<n>We propose a unified dynamic-resolution strategy to address the diverse image scales inherent in RS imagery.
  arXiv  Detail & Related papers  (2025-11-26T10:55:07Z)
• Growing Visual Generative Capacity for Pre-Trained MLLMs [60.826355079902505]
  Bridge is a pure autoregressive unified MLLM that augments pre-trained visual understanding models with generative ability.<n>We propose a semantic-to-pixel discrete representation that integrates compact semantic tokens with fine-grained pixel tokens.
  arXiv  Detail & Related papers  (2025-10-02T00:40:02Z)
• Efficient Multi-modal Long Context Learning for Training-free Adaptation [96.21248144937627]
  This paper introduces Efficient Multi-Modal Long Context Learning (EMLoC)<n>It embeds demonstration examples directly into the model input.<n>It condenses long-context multimodal inputs into compact, task-specific memory representations.
  arXiv  Detail & Related papers  (2025-05-26T10:49:44Z)
• Understanding and Optimizing Multi-Stage AI Inference Pipelines [11.254219071373319]
  HERMES is a Heterogeneous Multi-stage LLM inference Execution Simulator. HERMES supports heterogeneous clients executing multiple models concurrently unlike prior frameworks. We explore the impact of reasoning stages on end-to-end latency, optimal strategies for hybrid pipelines, and the architectural implications of remote KV cache retrieval.
  arXiv  Detail & Related papers  (2025-04-14T00:29:49Z)
• From 128K to 4M: Efficient Training of Ultra-Long Context Large Language Models [54.44375226381814]
  Long-context capabilities are essential for a wide range of applications, including document and video understanding, in-context learning, and inference-time scaling. We introduce a efficient training recipe for building ultra-long context LLMs from aligned instruct model, pushing the boundaries of context lengths from 128K to 1M, 2M, and 4M tokens. Our approach achieves state-of-the-art performance across a diverse set of long-context benchmarks.
  arXiv  Detail & Related papers  (2025-04-08T16:58:58Z)
• SmolVLM: Redefining small and efficient multimodal models [8.849350918179752]
  We introduce SmolVLM, a series of compact multimodal models specifically engineered for resource-efficient inference.<n>We identify key design choices that yield substantial performance gains on image and video tasks with minimal memory footprints.<n>Our results emphasize that strategic architectural optimizations, aggressive yet efficient tokenization, and carefully curated training data significantly enhance multimodal performance.
  arXiv  Detail & Related papers  (2025-04-07T17:58:57Z)
• OmniMamba: Efficient and Unified Multimodal Understanding and Generation via State Space Models [36.0400717590138]
  We present OmniMamba, the first linear-architecture-based multimodal generation model. It generates both text and images through a unified next-token prediction paradigm. It achieves competitive performance with JanusFlow while surpassing Show-o across benchmarks.
  arXiv  Detail & Related papers  (2025-03-11T17:59:46Z)
• LightGen: Efficient Image Generation through Knowledge Distillation and Direct Preference Optimization [37.236005953016175]
  LightGen is an efficient training paradigm for image generation models.<n>It distills knowledge from state-of-the-art (SOTA) text-to-image models into a compact Masked Autoregressive architecture.<n>Experiments confirm that LightGen achieves image generation quality comparable to SOTA models.
  arXiv  Detail & Related papers  (2025-03-11T16:58:02Z)
• Migician: Revealing the Magic of Free-Form Multi-Image Grounding in Multimodal Large Language Models [79.59567114769513]
  We introduce Migician, the first multi-image grounding model capable of performing free-form and accurate grounding across multiple images. Our model achieves significantly superior multi-image grounding capabilities, outperforming the best existing MLLMs by 24.94% and even surpassing much larger 70B models.
  arXiv  Detail & Related papers  (2025-01-10T07:56:23Z)
• Selective State Space Memory for Large Vision-Language Models [0.0]
  State Space Memory Integration (SSMI) is a novel approach for efficient fine-tuning of LVLMs. SSMI captures long-range dependencies and injects task-specific visual and sequential patterns effectively. experiments on benchmark datasets, including COCO Captioning, VQA, and Flickr30k, demonstrate that SSMI achieves state-of-the-art performance.
  arXiv  Detail & Related papers  (2024-12-13T05:40:50Z)
• SynerGen-VL: Towards Synergistic Image Understanding and Generation with Vision Experts and Token Folding [66.74446220401296]
  We propose SynerGen-VL, a simple yet powerful encoder-free MLLM capable of both image understanding and generation.<n>We introduce the token folding mechanism and the vision-expert-based progressive alignment pretraining strategy, which effectively support high-resolution image understanding.<n>Our code and models shall be released.
  arXiv  Detail & Related papers  (2024-12-12T18:59:26Z)
• Dynamic-VLM: Simple Dynamic Visual Token Compression for VideoLLM [28.64108439552772]
  We introduce a large-scale synthetic dataset created from proprietary models. We also explore a dynamic visual token compression architecture that strikes a balance between computational efficiency and performance. Our proposed model achieves state-of-the-art results across various video tasks and shows impressive generalization.
  arXiv  Detail & Related papers  (2024-12-12T18:20:41Z)
• Multimodal Instruction Tuning with Hybrid State Space Models [25.921044010033267]
  Long context is crucial for enhancing the recognition and understanding capabilities of multimodal large language models. We propose a novel approach using a hybrid transformer-MAMBA model to efficiently handle long contexts in multimodal applications. Our model enhances inference efficiency for high-resolution images and high-frame-rate videos by about 4 times compared to current models.
  arXiv  Detail & Related papers  (2024-11-13T18:19:51Z)
• ADEM-VL: Adaptive and Embedded Fusion for Efficient Vision-Language Tuning [38.26304604660713]
  ADEM-VL is an efficient vision-language method that tunes models based on pretrained large language models. Our framework surpasses existing methods by an average accuracy of 0.77% on ScienceQA dataset.
  arXiv  Detail & Related papers  (2024-10-23T11:31:06Z)
• PDF-WuKong: A Large Multimodal Model for Efficient Long PDF Reading with End-to-End Sparse Sampling [63.93112754821312]
  Document understanding is a challenging task to process and comprehend large amounts of textual and visual information. Recent advances in Large Language Models (LLMs) have significantly improved the performance of this task. We introduce PDF-WuKong, a multimodal large language model (MLLM) which is designed to enhance multimodal question-answering (QA) for long PDF documents.
  arXiv  Detail & Related papers  (2024-10-08T12:17:42Z)
• EMMA: Empowering Multi-modal Mamba with Structural and Hierarchical Alignment [39.870809905905325]
  We propose Empowering Multi-modal Mamba with Structural and Hierarchical Alignment (EMMA) to extract fine-grained visual information. Our model shows lower latency than other Mamba-based MLLMs and is nearly four times faster than transformer-based MLLMs of similar scale during inference.
  arXiv  Detail & Related papers  (2024-10-08T11:41:55Z)
• NVLM: Open Frontier-Class Multimodal LLMs [64.00053046838225]
  We introduce NVLM 1.0, a family of frontier-class multimodal large language models (LLMs) that achieve state-of-the-art results on vision-language tasks. We propose a novel architecture that enhances both training efficiency and multimodal reasoning capabilities. We develop production-grade multimodality for the NVLM-1.0 models, enabling them to excel in vision-language tasks.
  arXiv  Detail & Related papers  (2024-09-17T17:59:06Z)
• KV Cache Compression, But What Must We Give in Return? A Comprehensive Benchmark of Long Context Capable Approaches [52.02764371205856]
  Long context capability is a crucial competency for large language models (LLMs) This work provides a taxonomy of current methods and evaluating 10+ state-of-the-art approaches across seven categories of long context tasks.
  arXiv  Detail & Related papers  (2024-07-01T17:59:47Z)
• LOOK-M: Look-Once Optimization in KV Cache for Efficient Multimodal Long-Context Inference [32.20654044142376]
  LOOK-M is a pioneering, fine-tuning-free approach that efficiently reduces the multimodal KV cache size. It achieves up to 1.5x faster decoding and also maintains or even enhances performance across a variety of long context multimodal tasks.
  arXiv  Detail & Related papers  (2024-06-26T07:44:24Z)
• Scaling Diffusion Mamba with Bidirectional SSMs for Efficient Image and Video Generation [41.54814517077309]
  We introduce a novel diffusion architecture, Diffusion Mamba (DiM), which foregoes traditional attention mechanisms in favor of a scalable alternative. DiM achieves rapid inference times and reduced computational load, maintaining linear complexity with respect to sequence length. Results affirm the scalability and efficiency of DiM, establishing a new benchmark for image and video generation techniques.
  arXiv  Detail & Related papers  (2024-05-24T18:50:27Z)
• DiM: Diffusion Mamba for Efficient High-Resolution Image Synthesis [56.849285913695184]
  Diffusion Mamba (DiM) is a sequence model for efficient high-resolution image synthesis. DiM architecture achieves inference-time efficiency for high-resolution images. Experiments demonstrate the effectiveness and efficiency of our DiM.
  arXiv  Detail & Related papers  (2024-05-23T06:53:18Z)
• From Text to Pixel: Advancing Long-Context Understanding in MLLMs [70.78454154014989]
  We introduce SEEKER, a multimodal large language model designed to tackle this issue. SEEKER aims to optimize the compact encoding of long text by compressing the text sequence into the visual pixel space via images. Our experiments on six long-context multimodal tasks demonstrate that SEEKER can leverage fewer image tokens to convey the same amount of textual information compared with the OCR-based approach.
  arXiv  Detail & Related papers  (2024-05-23T06:17:23Z)
• SPHINX: The Joint Mixing of Weights, Tasks, and Visual Embeddings for Multi-modal Large Language Models [86.478087039015]
  We present a versatile multi-modal large language model (MLLM) with a joint mixing of model weights, tuning tasks, and visual embeddings. Based on our proposed joint mixing, we propose an efficient strategy aiming to better capture fine-grained appearances of high-resolution images. We hope our work may cast a light on the exploration of joint mixing in future MLLM research.
  arXiv  Detail & Related papers  (2023-11-13T18:59:47Z)
• Matryoshka Diffusion Models [38.26966802461602]
  Diffusion models are the de facto approach for generating high-quality images and videos. We introduce Matryoshka Diffusion Models, an end-to-end framework for high-resolution image and video synthesis. We demonstrate the effectiveness of our approach on various benchmarks, including class-conditioned image generation, high-resolution text-to-image, and text-to-video applications.
  arXiv  Detail & Related papers  (2023-10-23T17:20:01Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.