官术网_书友最值得收藏!

TensorFlow

TensorFlow was initially developed by the Google Brain team to be used internally at Google, and was then to be released to the larger public. On November 9, 2015, it was distributed under the Apache 2.0 open source license, and since then it has become the most widespread open source software library for high-performance numerical computation (mostly used for deep learning). It is capable of computations across a variety of platforms (systems with multiple CPUs, GPUs, and TPUs), and from desktops to clusters of servers to mobile and edge devices.

In this book, we will use TensorFlow as the backend of Keras, that is, we won't use it directly, but we will need to have it running on our system:

Installing TensorFlow on a CPU system is quite straightforward: just use pip install tensorflow. But if you have an NVIDIA GPU (you actually need a GPU card with CUDA Compute Capability 3.0 or higher) on your system, the requirements ramp up and you first have to install the following:

  • CUDA Toolkit 9.0
  • The NVIDIA drivers associated with CUDA Toolkit 9.0
  • cuDNN v7.0

For each operation, you need to accomplish various steps depending on your system, as detailed on the NVIDIA website. You can find all the directions for installation depending on your system (Ubuntu, Windows, or macOS) at https://www.tensorflow.org/install/.

After having accomplished all the necessary steps, pip install tensorflow-gpu will install the TensorFlow package that's optimized for GPU computations.

主站蜘蛛池模板: 阿坝县| 荣成市| 郓城县| 宾阳县| 吴忠市| 什邡市| 辛集市| 柳州市| 忻城县| 山阴县| 浙江省| 汝南县| 普兰县| 莱西市| 凤冈县| 塔城市| 西吉县| 宁河县| 清河县| 名山县| 兴业县| 克拉玛依市| 苗栗市| 临颍县| 来宾市| 武强县| 黄平县| 奉节县| 合川市| 莫力| 邵阳县| 黄浦区| 太谷县| 罗山县| 满城县| 苍溪县| 高邑县| 上杭县| 广河县| 沂南县| 临沧市|