TensorFlow v0.10.0 RC0发布,一个表达机器学习算法的接口

jopen 5年前
   <p style="text-align: center;"><img alt="" src="https://simg.open-open.com/show/326eaa9a539ec5575022ac39de585d3d.png" /></p>    <p>TensorFlow 是一个表达机器学习算法的接口,并且是执行算法的实现框架。使用 TensorFlow 表示的计算可以在众多异构的系统上方便地移植,从移动设别如手机或者平板电脑到成千的GPU计算集群上都可以执行。该系统灵活,可以被用来表示很多的算法包括,深度神经网络的训练和推断算法,也已经被用作科研和应用机器学习系统在若干的计算机科学领域或者其他领域中,例如语言识别、计算机视觉、机器人、信息检索、自然语言理解、地理信息抽取和计算药物发现。</p>    <p style="text-align: center;"><img alt="" src="https://simg.open-open.com/show/4a67e12961d71d510c83c2aa35a8febb.gif" /></p>    <h2>更新日志</h2>    <h3>主要功能和改进</h3>    <ul>     <li>Added support for C++ shape inference</li>     <li>Added graph-construction C API</li>     <li>Major revision to the graph-construction C++ API</li>     <li>Support makefile build for iOS</li>     <li>Added Mac GPU support</li>     <li>Full version of TF-Slim available as <code>tf.contrib.slim</code></li>     <li>Added k-Means clustering and WALS matrix factorization</li>    </ul>    <h3>重大Bug修复和其他更改</h3>    <ul>     <li>Allow gradient computation for scalar values.</li>     <li>Performance improvements for gRPC</li>     <li>Improved support for fp16</li>     <li>New high-level ops in tf.contrib.{layers,metrics}</li>     <li>New features for TensorBoard, such as shape display, exponential smoothing</li>     <li>Faster and more stable Google Cloud Storage (GCS) filesystem support</li>     <li>Support for zlib compression and decompression for TFRecordReader and TFRecordWriter</li>     <li>Support for reading (animated) GIFs</li>     <li>Improved support for SparseTensor</li>     <li>Added support for more probability distributions (Dirichlet, Beta, Bernoulli, etc.)</li>     <li>Added Python interfaces to reset resource containers.</li>     <li>Many bugfixes and performance improvements</li>     <li>Many documentation fixes</li>    </ul>    <h2>下载</h2>    <ul>     <li><a href="/misc/goto?guid=4958992721413282923" rel="nofollow"><strong>Source code</strong> (zip)</a></li>     <li><a href="/misc/goto?guid=4958992721518944403" rel="nofollow"><strong>Source code</strong> (tar.gz)</a></li>    </ul>