<blockquote>
<p>By checking the version of the installed package, conda installs Tensorflow version 2.1.
But as of today the latest version of Tensorflow is 2.3. Furthermore</p>
</blockquote>
<p>那只是因为你(可能?)在windows上。正如您所看到的,<a href="https://anaconda.org/anaconda/tensorflow" rel="nofollow noreferrer">here</a><code>tensorflow</code>作为2.3从<code>conda</code>默认通道中可用,但目前仅在linux上可用</p>
<p>你的网站上也说明了原因<a href="https://www.anaconda.com/blog/tensorflow-in-anaconda" rel="nofollow noreferrer">linked</a>(我的重点):</p>
<blockquote>
<p>Anaconda is proud of our efforts to deliver a simpler, faster experience using the excellent TensorFlow library. It takes significant time and effort to add support for the many platforms used in production, and to ensure that the accelerated code is still stable and mathematically correct. As a result, our TensorFlow packages <strong>may not be available concurrently with the official TensorFlow wheels</strong>. We are, however, committed to maintaining our TensorFlow packages, and work to have updates available as soon as we can.</p>
</blockquote>
<p>简而言之:Anaconda团队正在针对intel mkl库创建自定义tf版本,以加快CPU上的计算速度。早些时候在同一个网站上,他们还提到他们为不同的cuda版本创建版本</p>
<blockquote>
<p>Why Anaconda provides 2.1 in two different packages, given that the package should be the same for any version > 1.15?</p>
</blockquote>
<p><code>tensorflow-gpu</code>软件包只是一个元软件包,也就是说,它只用于安装具有不同依赖项的不同版本的<code>tensorflow</code>(也使您能够安装不同的cuda版本)。官方版本只允许<a href="https://www.tensorflow.org/install/source#gpu" rel="nofollow noreferrer">combinations</a>的tensorflow版本和cuda</p>
<blockquote>
<p>Which one should I install, the pip version or the conda version? An article in Anaconda blog specifies that the version provided with conda is faster, but the article is old (2018) and refers to an old version of Tensorflow (1.10)</p>
</blockquote>
<p>在阅读上述文章时,这种加速与针对英特尔mkl库的构建有关,后者可以加速CPU上的计算。考虑到您的设置,在使用<code>conda</code>时只能安装<code>tensorflow</code>2.1,您需要问问自己是否依赖最新的<code>tensorflow</code>版本,以及是否不需要加速的cpu代码。使用<code>pip</code>安装最新的tensorflow通常没有问题。只需确保您为上述tensorflow版本创建了一个新环境,并且仅在该环境中使用<code>pip</code>安装/更新tensorflow或其任何依赖项即可。有<a href="https://www.anaconda.com/blog/using-pip-in-a-conda-environment" rel="nofollow noreferrer">general advice</a>到不<strong>混合</strong><code>conda</code>和<code>pip</code>安装太多,因为一个安装可能会破坏另一个安装(因为它们使用不同的方法来解决依赖关系),但使用单独的env时应该很好</p>