2025_NIPS_Accelerating Multimodal Large Language Models via Dynamic Visual-Token Exit and Empirical

文章主要内容总结

本文聚焦多模态大语言模型(MLLMs)的视觉冗余问题,通过分析注意力行为发现其推理过程存在三个核心阶段:早期融合、模态内建模、多模态推理。研究揭示,视觉 tokens 在文本 tokens 获取足够视觉信息后便不再对推理有实质贡献。基于此,提出动态视觉 token 退出方法(DyVTE),通过轻量级超网络感知文本 token 状态,自动决策视觉 tokens 的退出时机,在不损失性能的前提下降低计算开销。该方法与现有 token 级压缩方法正交且互补,在 LLaVA、VILA 等多个模型及9个基准测试中验证有效,最高可减少 45.7% 计算量,且具备跨模型泛化能力。

创新点

  1. 首次从注意力行为视角系统分析 MLLMs 推理模式,明确三个通用阶段及视觉 tokens 的冗余特性,为效率优化提供理论依据。
  2. 提出 DyVTE 方法,通过动态退出视觉 tokens 而非裁剪/合并单个 token,实现全局级冗余优化,且不改变模型原有推理流程。
  3. 设计独立于 MLLMs 梯度计算的高效训练机制,训练成本低,同时支持与现有 token 剪枝方法结合,进一步提升效率(最高减少 51.5% 计算量)。
  4. 验证了方法的通用性,在不同规模、不同架构的 MLLMs 上均有效,且具备跨模型迁移能力,无需针对特定模型重新训练。

翻译部分(Markdown 格式)

Abstract

本文从注意力行为视角研究多模态大语言模

船载 OpenHarmony AIS 航海工作台:告警与防碰撞预警系统 船载OpenHarmony AIS航海工作台的告警与防碰撞系统,集成十余类告警(超速、偏航、会遇、SOS等),通过网格空间索引实现高效邻近判定,保障实时性。告警管理器统一调度规则、去重、通知与日志,支持悬浮球、语音、跑马灯、弹窗多渠道触达,确保航行安全。SOS紧急通道独立运行,避免干扰。系统设计强调性能优化与多场景适配,为航海安全提供核心支撑。 阅读详情

相关推荐

AI 应用开发:四维架构落地实战——以 HR Agent 为例拆解感知·推理·记忆·执行【意图路由、规划编排、长期记忆、工具执行、面试题】

本文基于四维架构(感知·推理·记忆·执行),以HR Agent为例,系统拆解大模型应用从“聊天”到“生产可用”的落地路径。通过Spring AI实现意图路由、规划编排、RAG检索与工具调用,结合真实踩坑复盘,揭示架构化设计对安全、成本与可维护性的关键价值,为AI Agent工程化提供可复用的实战框架。

王大师企业官方博客 767

通达信筹码分布主图源码分享

NOTEXT下区成本:IF(CURRBARSCOUNT<=CONST(XA_26+1),CONST(IF(1,XA_94,XA_94)),DRAWNULL),DOTLINE,COLORYELLOW;NOTEXT上区成本:IF(CURRBARSCOUNT<=CONST(XA_26+1),CONST(IF(1,XA_81,XA_81)),DRAWNULL),DOTLINE,COLORRED;FFG1:DRAWBAND(C线,RGB(255,0,0),XA_33,RGB(255,0,0));

zhibiaogong的博客 3359

Android逆向实战:抖音聊天记录数据库解析与导出(附完整SQL查询)

本文提供了一份Android逆向实战指南,详细解析了抖音聊天记录数据库的结构与导出方法。通过定位核心的`im.db`数据库文件,深度解读conversation_list、msg等关键表结构,并附上完整的SQL查询语句,帮助开发者或研究人员安全、合规地提取和分析本地聊天数据,适用于移动安全研究与数据分析场景。

mango的博客 860

Genesis物理引擎实战指南:轻量级确定性刚体仿真核心

刚体物理仿真作为机器人控制、嵌入式运动学验证和确定性游戏开发的基础技术,其核心在于数值稳定性、跨平台可复现性与约束求解精度。本文围绕线性互补问题(LCP)求解、半隐式欧拉积分和零依赖C++实现三大关键技术展开,解析如何在无GUI、低算力环境下构建可单步调试、帧级可控的确定性仿真系统。重点覆盖碰撞检测数学建模、关节局部坐标系绑定、质量-惯性张量一致性校验等工程落地关键点,适用于STM32、树莓派及教育实验场景。内容深度结合Genesis Physics Engine的源码实践,直击‘穿模’‘NaN’‘关节发软

ciya3282的博客 1592

《低功耗方法学》翻译——第一章:介绍

第一章:介绍

记录所学,分享知识,结识挚友 2069

【AI智能体】Dify集成 Echarts实现数据报表展示实战详解

Dify集成 Echarts 实现智能数据报表集成与展示实战详解

congge 5962

基于图像形态学处理的交通灯检测识别matlab仿真

图像形态学处理是基于数学形态学的理论,利用结构元素(如矩形、圆形等)对图像进行处理,通过将结构元素与图像进行特定的逻辑运算,来提取图像中的目标特征、去除噪声、改善图像质量等。在交通灯检测识别中,主要利用交通灯的颜色、形状等特征,通过形态学操作突出这些特征,抑制背景和噪声干扰,从而实现对交通灯的准确检测和识别。图像采集与预处理采集:通过安装在车辆或道路上的摄像头获取交通场景图像。灰度化:将彩色图像转换为灰度图像,减少数据量,便于后续处理。

平时工作较为繁忙,私信消息一般晚上回复,谢谢大家~~ 160

Coze智能体开发:扣子智能体设置多Agent模式

在扣子中创建智能体后,智能体默认使用单 Agent 模式。在单 Agent 模式下处理复杂任务时,你必须编写非常详细和冗长的提示词,而且你可能需要添加各种插件和工作流等,这增加了调试智能体的复杂性。调试时任何一处细节改动,都有可能影响到智能体的整体功能,实际处理用户任务时,处理结果可能与预期效果有较大出入。

王国平的博客 3713

计算机控制技术及应用

《计算机控制技术及应用》通过总结作者的科研成果和吸收国内外的先进理论、方法和技术,以计算机控制系统的网络化、开放化、智能化和集成化发展趋势为主线索,主要介绍计算机控制系统的基本概念和计算机控制系统的典型形式,计算机控制系统的数学描述与基本分析方法,以及数字PID控制器、最少拍控制系统、达林算法、状态变量设计法等设计方法及其各种改进算法。重点突出计算机控制系统输入输出通道接口的设计方法和典型控制系统的设计案例与实现方法等。

深度学习工具包 DeepLearningToolbox

深度学习工具包Deprecation notice.-----This toolbox is outdated and no longer maintained.There are much better tools available for deep learning than this toolbox, e.g. [Theano](http://deeplearning.net/software/theano/), [torch](http://torch.ch/) or [tensorflow](http://www.tensorflow.org/)I would suggest you use one of the tools mentioned above rather than use this toolbox.Best, Rasmus.DeepLearnToolbox================A Matlab toolbox for Deep Learning.Deep Learning is a new subfield of machine learning that focuses on learning deep hierarchical models of data.It is inspired by the human brain's apparent deep (layered, hierarchical) architecture.A good overview of the theory of Deep Learning theory is[Learning Deep Architectures for AI](http://www.iro.umontreal.ca/~bengioy/papers/ftml_book.pdf)For a more informal introduction, see the following videos by Geoffrey Hinton and Andrew Ng.* [The Next Generation of Neural Networks](http://www.youtube.com/watch?v=AyzOUbkUf3M) (Hinton, 2007)* [Recent Developments in Deep Learning](http://www.youtube.com/watch?v=VdIURAu1-aU) (Hinton, 2010)* [Unsupervised Feature Learning and Deep Learning](http://www.youtube.com/watch?v=ZmNOAtZIgIk) (Ng, 2011)If you use this toolbox in your research please cite [Prediction as a candidate for learning deep hierarchical models of data](http://www2.imm.dtu.dk/pubdb/views/publication_details.php?id=6284)```@MASTERSTHESIS\{IMM2012-06284, author = "R. B. Palm", title = "Prediction as a candidate for learning deep hierarchical models of data", year = "2012",}```Contact: rasmusbergpalm at gmail dot comDirectories included in the toolbox-----------------------------------`NN/` - A library for Feedforward Backpropagation Neural Networks`CNN/` - A library for Convolutional Neural Networks`DBN/` - A library for Deep Belief Networks`SAE/` - A library for Stacked Auto-Encoders`CAE/` - A library for Convolutional Auto-Encoders`util/` - Utility functions used by the libraries`data/` - Data used by the examples`tests/` - unit tests to verify toolbox is workingFor references on each library check REFS.mdSetup-----1. Download.2. addpath(genpath('DeepLearnToolbox'));Example: Deep Belief Network---------------------```matlabfunction test_example_DBNload mnist_uint8;train_x = double(train_x) / 255;test_x = double(test_x) / 255;train_y = double(train_y);test_y = double(test_y);%% ex1 train a 100 hidden unit RBM and visualize its weightsrand('state',0)dbn.sizes = [100];opts.numepochs = 1;opts.batchsize = 100;opts.momentum = 0;opts.alpha = 1;dbn = dbnsetup(dbn, train_x, opts);dbn = dbntrain(dbn, train_x, opts);figure; visualize(dbn.rbm{1}.W'); % Visualize the RBM weights%% ex2 train a 100-100 hidden unit DBN and use its weights to initialize a NNrand('state',0)%train dbndbn.sizes = [100 100];opts.numepochs = 1;opts.batchsize = 100;opts.momentum = 0;opts.alpha = 1;dbn = dbnsetup(dbn, train_x, opts);dbn = dbntrain(dbn, train_x, opts);%unfold dbn to nnnn = dbnunfoldtonn(dbn, 10);nn.activation_function = 'sigm';%train nnopts.numepochs = 1;opts.batchsize = 100;nn = nntrain(nn, train_x, train_y, opts);[er, bad] = nntest(nn, test_x, test_y);assert(er < 0.10, 'Too big error');```Example: Stacked Auto-Encoders---------------------```matlabfunction test_example_SAEload mnist_uint8;train_x = double(train_x)/255;test_x = double(test_x)/255;train_y = double(train_y);test_y = double(test_y);%% ex1 train a 100 hidden unit SDAE and use it to initialize a FFNN% Setup and train a stacked denoising autoencoder (SDAE)rand('state',0)sae = saesetup([784 100]);sae.ae{1}.activation_function = 'sigm';sae.ae{1}.learningRate = 1;sae.ae{1}.inputZeroMaskedFraction = 0.5;opts.numepochs = 1;opts.batchsize = 100;sae = saetrain(sae, train_x, opts);visualize(sae.ae{1}.W{1}(:,2:end)')% Use the SDAE to initialize a FFNNnn = nnsetup([784 100 10]);nn.activation_function = 'sigm';nn.learningRate = 1;nn.W{1} = sae.ae{1}.W{1};% Train the FFNNopts.numepochs = 1;opts.batchsize = 100;nn = nntrain(nn, train_x, train_y, opts);[er, bad] = nntest(nn, test_x, test_y);assert(er < 0.16, 'Too big error');```Example: Convolutional Neural Nets---------------------```matlabfunction test_example_CNNload mnist_uint8;train_x = double(reshape(train_x',28,28,60000))/255;test_x = double(reshape(test_x',28,28,10000))/255;train_y = double(train_y');test_y = double(test_y');%% ex1 Train a 6c-2s-12c-2s Convolutional neural network %will run 1 epoch in about 200 second and get around 11% error. %With 100 epochs you'll get around 1.2% errorrand('state',0)cnn.layers = { struct('type', 'i') %input layer struct('type', 'c', 'outputmaps', 6, 'kernelsize', 5) %convolution layer struct('type', 's', 'scale', 2) %sub sampling layer struct('type', 'c', 'outputmaps', 12, 'kernelsize', 5) %convolution layer struct('type', 's', 'scale', 2) %subsampling layer};cnn = cnnsetup(cnn, train_x, train_y);opts.alpha = 1;opts.batchsize = 50;opts.numepochs = 1;cnn = cnntrain(cnn, train_x, train_y, opts);[er, bad] = cnntest(cnn, test_x, test_y);%plot mean squared errorfigure; plot(cnn.rL);assert(er<0.12, 'Too big error');```Example: Neural Networks---------------------```matlabfunction test_example_NNload mnist_uint8;train_x = double(train_x) / 255;test_x = double(test_x) / 255;train_y = double(train_y);test_y = double(test_y);% normalize[train_x, mu, sigma] = zscore(train_x);test_x = normalize(test_x, mu, sigma);%% ex1 vanilla neural netrand('state',0)nn = nnsetup([784 100 10]);opts.numepochs = 1; % Number of full sweeps through dataopts.batchsize = 100; % Take a mean gradient step over this many samples[nn, L] = nntrain(nn, train_x, train_y, opts);[er, bad] = nntest(nn, test_x, test_y);assert(er < 0.08, 'Too big error');%% ex2 neural net with L2 weight decayrand('state',0)nn = nnsetup([784 100 10]);nn.weightPenaltyL2 = 1e-4; % L2 weight decayopts.numepochs = 1; % Number of full sweeps through dataopts.batchsize = 100; % Take a mean gradient step over this many samplesnn = nntrain(nn, train_x, train_y, opts);[er, bad] = nntest(nn, test_x, test_y);assert(er < 0.1, 'Too big error');%% ex3 neural net with dropoutrand('state',0)nn = nnsetup([784 100 10]);nn.dropoutFraction = 0.5; % Dropout fraction opts.numepochs = 1; % Number of full sweeps through dataopts.batchsize = 100; % Take a mean gradient step over this many samplesnn = nntrain(nn, train_x, train_y, opts);[er, bad] = nntest(nn, test_x, test_y);assert(er < 0.1, 'Too big error');%% ex4 neural net with sigmoid activation functionrand('state',0)nn = nnsetup([784 100 10]);nn.activation_function = 'sigm'; % Sigmoid activation functionnn.learningRate = 1; % Sigm require a lower learning rateopts.numepochs = 1; % Number of full sweeps through dataopts.batchsize = 100; % Take a mean gradient step over this many samplesnn = nntrain(nn, train_x, train_y, opts);[er, bad] = nntest(nn, test_x, test_y);assert(er < 0.1, 'Too big error');%% ex5 plotting functionalityrand('state',0)nn = nnsetup([784 20 10]);opts.numepochs = 5; % Number of full sweeps through datann.output = 'softmax'; % use softmax outputopts.batchsize = 1000; % Take a mean gradient step over this many samplesopts.plot = 1; % enable plottingnn = nntrain(nn, train_x, train_y, opts);[er, bad] = nntest(nn, test_x, test_y);assert(er < 0.1, 'Too big error');%% ex6 neural net with sigmoid activation and plotting of validation and training error% split training data into training and validation datavx = train_x(1:10000,:);tx = train_x(10001:end,:);vy = train_y(1:10000,:);ty = train_y(10001:end,:);rand('state',0)nn = nnsetup([784 20 10]); nn.output = 'softmax'; % use softmax outputopts.numepochs = 5; % Number of full sweeps through dataopts.batchsize = 1000; % Take a mean gradient step over this many samplesopts.plot = 1; % enable plottingnn = nntrain(nn, tx, ty, opts, vx, vy); % nntrain takes validation set as last two arguments (optionally)[er, bad] = nntest(nn, test_x, test_y);assert(er < 0.1, 'Too big error');```[![Bitdeli Badge](https://d2weczhvl823v0.cloudfront.net/rasmusbergpalm/deeplearntoolbox/trend.png)](https://bitdeli.com/free "Bitdeli Badge")

visio安装包

Microsoft Office Visio 是一款专业的制图软件。利用经过更新的形状、协作工具和数据链接图表,创建专业级图表以简化复杂的信息。

Data-Structures-and-Algorithm-Analysis-in-Java.pdf

作者: Weiss, Mark A. 出版社: Addison Wesley Data Structures and Algorithm Analysis in Java is an “advanced algorithms” book that fits between traditional CS2 and Algorithms Analysis courses. In the old ACM Curriculum Guidelines, this course was known as CS7. This text is for readers who want to learn good programming and algorithm analysis skills simultaneously so that they can develop such programs with the maximum amo...

上一篇: 2025_NIPS_Grammars of Formal Uncertainty: When to Trust LLMs in Automated Reasoning Tasks
下一篇: 2025_NIPS_Regression-adjusted Monte Carlo Estimators for Shapley Values and Probabilistic Values
UnknownBody
博客等级 码龄16年 5601粉丝 4742原创
评论
成就一亿技术人!
拼手气红包6.0元
还能输入1000个字符
 
 条评论被折叠 查看
添加红包

请填写红包祝福语或标题

个

红包个数最小为10个

元

红包金额最低5元

当前余额3.43元 前往充值 >
需支付:10.00元
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包

打赏作者

UnknownBody

你的鼓励将是我创作的最大动力

¥1 ¥2 ¥4 ¥6 ¥10 ¥20
扫码支付:¥1
获取中
扫码支付

您的余额不足,请更换扫码支付或充值

打赏作者

实付元
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值