ML-DL-各种资源汇总
1.Used Libraries, Datasets, and Models
1.1 Libraries
- TensorFlow (from Google): https://www.tensorflow.org/
- Theano (from U Montreal): https://github.com/Theano/Theano
- Caffe (from Berkeley): http://caffe.berkeleyvision.org/
- Torch (from Facebook): http://torch.ch/
- Nervana Graph (from Intel):
- MXNet (from Amazon): http://mxnet.io/
- Deeplearning4j (from Skymind): https://deeplearning4j.org/
- CNTK (from Microsoft): https://github.com/Microsoft/CNTK/wiki
- Keras (high-level API for both TensorFlow and Theano): https://keras.io/
- TensorFlow-Slim (high-level API for TensorFlow)
- Various other high-level APIs for Tensorflow, like TFLearn, TensorLayer
- Spark + Caffe and/or TensorFlow:
- SparkNet with Caffe as backend for each worker (from Berkeley):
1.2 Datasets
- ImageNet (large scale image-classification; pre-training)
- MS COCO (large scale object bounding box detection, image captioning, visual question answering)
- Cifar10 (small scale image classification)
- MNIST(small scale digits classification)
- Visual Genome https://visualgenome.org/ (large scale multi-task image understanding)
1.3 Models Structure
- Inception
- ResNet
- VGG
- AlexNet
- MobileNet
- SqueezeNet
1.3 Available pre-trained model
- tensorflow / slim
1) https://github.com/tensorflow/models/tree/master/slim (inceptionv1-v4; resnet50, 101, 152; vgg16,19; inception-resnet-v2) - caffe
1) https://github.com/BVLC/caffe/wiki/Model-Zoo (model zoo)
2) https://github.com/beniz/deepdetect/issues/89 (resnet, inception-v1, vgg) - keras
1) https://keras.io/applications/ (xception, vgg16, vgg19, resnet50, inceptionv3, mobilenet) - mxnet
1) model galaxy https://github.com/dmlc/mxnet-model-gallery
2. TensorFlow Tips
2.1 Using GPUs
By default, TensorFlow eats all your GPUs' memory (Yes, every single bit of them!), even when you explicitly specify which gpu(s) to use. To limit this "intended" behavior, one needs to set GPUOptions to your session, and we recommend the following. One can refer to here for the rationale.
gpu_options = tf.GPUOptions(per_process_gpu_memory_fraction=0., allow_growth=True)config=tf.ConfigProto(gpu_options=gpu_options, <your other configuraions>)sess = tf.Session(config=config, <your other session settings>)
with sess.as_default():
<rest of session code>
The above has been tested under TF version r0.12 on DL Workstation 1, 2, 3. The testing code is here (tested with both single-gpu training and multi-gpu training). Note that the code is different from the example provided in TF examples, and it is modified by us to be compatible with r0.12.
As of 2017/01/11, experimenting such GPU setting with Keras using TF backend is inconclusive. Further investigation is needed.
2.1.1 To limit the GPU to use
Use CUDA_VISIBLE_DEVICES, see http://stackoverflow.com/questions/34775522/tensorflow-mutiple-sessions-with-mutiple-gpus
$ CUDA_VISIBLE_DEVICES=0 python my_script.py # Uses GPU 0.
$ CUDA_VISIBLE_DEVICES=1 python my_script.py # Uses GPU 1.
$ CUDA_VISIBLE_DEVICES=2,3 python my_script.py # Uses GPUs 2 and 3.
2.2 Upgrade to TF1.0
https://www.tensorflow.org/install/migration#how_to_upgrade
ML-DL-各种资源汇总的更多相关文章
- ML&DL视频教程资源
作者:Bruce链接:https://www.zhihu.com/question/49909565/answer/345894856来源:知乎著作权归作者所有.商业转载请联系作者获得授权,非商业转载 ...
- Kinect开发资源汇总
Kinect开发资源汇总 转自: http://www.sigvc.org/bbs/forum.php?mod=viewthread&tid=254&highlight=kinec ...
- 微信小程序(应用号)资源汇总整理
微信小应用资源汇总整理 开源项目 WeApp - 微信小程序版的微信 wechat-weapp-redux-todos - 微信小程序集成Redux实现的Todo list wechat-weapp- ...
- 【特别推荐】Node.js 入门教程和学习资源汇总
这篇文章与大家分享一批很有用的 Node.js 入门教程和学习资源.Node 是一个服务器端的 JavaScript 解释器,它将改变服务器应该如何工作的概念.它的目标是帮助程序员构建高度可伸缩的应用 ...
- Github上PHP资源汇总大全,php学习的好资料
Github上PHP资源汇总大全,php学习的好资料 国外程序员ziadoz 在Github上收集整理了PHP的各种资源,内容包括模板.框架.数据库.安全等方面的库和工具.汇总了各种PHP资源,供各位 ...
- 知名杀毒软件Mcafee(麦咖啡)个人版 资源汇总兼科普(来自卡饭)
虽然早已不是用咖啡了,但我也实时关注的咖啡的一举一动,潜水看帖日久,发现小白众多,好多有价值的帖子淹没于帖海当中,甚是惋惜. 我有如下建议 1.咖啡区管理层,能否吧一些优秀的资源教程 ...
- GitHub最全的前端资源汇总仓库(包括前端学习、开发资源、求职面试等)
在GitHub上收集的最全的前端资源汇总(包括前端学习.前端开发资源.前端求职面试等) 个人结合github上各位大神分享的资源进行了简单的汇总整理,每一个条目下面都有丰富的资料,是前端学习.工作的好 ...
- 数据可视化的优秀入门书籍有哪些,D3.js 学习资源汇总
习·D3.js 学习资源汇总 除了D3.js自身以外,许多可视化工具包都是基于D3开发的,所以对D3的学习就显得很重要了,当然如果已经有了Javascript的经验,学起来也会不费力些. Github ...
- KbmMW资源汇总(更新中…)
KbmMW框架是收费的,不在此提供下载,如需购买,请自行联系作者Kim Madsen. 网址资源: 官网主页:http://www.components4programmers.com/product ...
- ENode简介与各种资源汇总
ENode简介与各种资源汇总 ENode是什么 ENode是一个.NET平台开源的应用开发框架,为开发人员提供了一套完整的基于DDD+CQRS+ES+(in-memory)+EDA架构风格的解决方案. ...
随机推荐
- stm8时钟
为使系统快速启动,复位后时钟控制器自动使用HSI的8分频(HSI/8)做为主时钟(2M).其原因为HSI的稳定时间短,而8分频可保证系统在较差的VDD条件下安全启动.一旦主时钟源稳定,用户程序可将主时 ...
- postman 脚本学习
pm的脚本断言库默认似乎是集成chaijs的.所以重点也要掌握chaijs的用法,其实和其他断言库类似.玩着玩着就会了.推荐看看 简书 chaijs 中文文档 传送门: # pm 脚本的教程 http ...
- Python 字符串过滤
需求: str1 = " """<div class="m_wrap clearfix"><ul class=" ...
- 最全PyCharm教程
最全PyCharm教程--for python PyCharm简介: PyCharm是由JetBrains打造的一款Python IDE,VS2010的重构插件Resharper就是出自JetBrai ...
- There is no Action mapped for namespace / and action name . - [unknown location]
今天碰到了这个问题,原因不明白. 在webContent下建立了index.jsp后重启项目不报错了. 原因未知.
- xctool + oclint 安装使用
使用brew 安装Xctool 先跟新brew : sudo brew update brew install xctool --HEAD OK. 使用请参照 文档 如: xctool -works ...
- 复习:JSP基本的语法(JSP凝视 + JSP指令 + JSP脚本元素 + JSP动作元素)
JSP原理: 1. 对于每个请求.jsp容器都会创建一个新的线程来处理它: 2. Servlet容器载入jsp后转换成的servlet(.class文件)是常驻内存的,所以对应速度一般比較 ...
- 设计模式-建造者模式(Builder Pattern)
建造者模式(Builder Pattern):将一个复杂对象的构建与它的表示分离,使得同样的构建过程可以创建不同的表示. 建造者模式要求建造过程中是稳定的. Android 用到的 Builder 模 ...
- 我的IT之路2013(一)
一眨眼又到了写总结的时候了.废话不多说了,直接切入正题. 春节过后 从春节前开始大概半个月的时间就开始在TKY做物资管理项目,中间穿插了两个考试和J2EE的学习:结束TKY工作后继续深入学习J2EE, ...
- [Windows Azure] Learn SQL Reporting on Windows Azure (9-Step Tutorial)
Learn SQL Reporting on Windows Azure (9-Step Tutorial) 4 out of 4 rated this helpful - Rate this top ...