blocks|key|1379470|text|根据文档，一旦你的模型经过训练，你就可以使用Tensorboard来可视化图形。|type|unstyled|depth|inlineStyleRanges|entityRanges|offset|length|data|1379471|首先，定义模型并运行它。然后，打开Tensorboard并切换到Graph。|1379472|最小可编译示例|style|BOLD|1379473|此示例取自文档。首先，定义模型和数据。|1379474|#+Relevant+imports.
%25load_ext+tensorboard

from+__future__+import+absolute_import
from+__future__+import+division
from+__future__+import+print_function

from+datetime+import+datetime
from+packaging+import+version

import+tensorflow+as+tf
from+tensorflow+import+keras

#+Define+the+model.
model+=+keras.models.Sequential([
++++keras.layers.Flatten(input_shape=(28,+28)),
++++keras.layers.Dense(32,+activation='relu'),
++++keras.layers.Dropout(0.2),
++++keras.layers.Dense(10,+activation='softmax')
])

model.compile(
++++optimizer='adam',
++++loss='sparse_categorical_crossentropy',
++++metrics=['accuracy'])

(train_images,+train_labels),+_+=+keras.datasets.fashion_mnist.load_data()
train_images+=+train_images+/+255.0|code-block|syntax|javascript|1379475|接下来，训练你的模特。在这里，您需要为Tensorboard定义一个回调以用于可视化统计数据和图形。|1379476|#+Define+the+Keras+TensorBoard+callback.
logdir="logs/fit/"+%2B+datetime.now().strftime("%25Y%25m%25d-%25H%25M%25S")
tensorboard_callback+=+keras.callbacks.TensorBoard(log_dir=logdir)

#+Train+the+model.
model.fit(
++++train_images,
++++train_labels,+
++++batch_size=64,
++++epochs=5,+
++++callbacks=[tensorboard_callback])|1379477|训练结束后，在笔记本上跑步。|1379478|%25tensorboard+--logdir+logs|1379479|并切换到导航栏中的Graph：|1379480|​|1379481|📷|atomic|1379482|1379483|您将看到一个非常像这样的图表：|1379484|1379485|1379486|1379487|entityMap|0|LINK|mutability|MUTABLE|url|https://www.tensorflow.org/tensorboard/r2/graphs#op-level_graph|1|IMAGE|IMMUTABLE|imageUrl|https://developer.qcloudimg.com/http-save/yehe-900000/b17cbe62998ba12174d2ff0e19a3015e.png|imageAlt|2|https://developer.qcloudimg.com/http-save/yehe-900000/6aaf653ae7262621520771b1fe7f9158.png^0|2|2|0|0|0|0|7|0|0|0|0|0|0|0|0|0|0|1|1|0|0|0|0|0|1|2|0|0^^$0|@$1|2|3|4|5|6|7|1R|8|@]|9|@$A|1S|B|1T|1|1U]]|C|$]]|$1|D|3|E|5|6|7|1V|8|@]|9|@]|C|$]]|$1|F|3|G|5|6|7|1W|8|@$A|1X|B|1Y|H|I]]|9|@]|C|$]]|$1|J|3|K|5|6|7|1Z|8|@]|9|@]|C|$]]|$1|L|3|M|5|N|7|20|8|@]|9|@]|C|$O|P]]|$1|Q|3|R|5|6|7|21|8|@]|9|@]|C|$]]|$1|S|3|T|5|N|7|22|8|@]|9|@]|C|$O|P]]|$1|U|3|V|5|6|7|23|8|@]|9|@]|C|$]]|$1|W|3|X|5|N|7|24|8|@]|9|@]|C|$O|P]]|$1|Y|3|Z|5|6|7|25|8|@]|9|@]|C|$]]|$1|10|3|11|5|6|7|26|8|@]|9|@]|C|$]]|$1|12|3|13|5|14|7|27|8|@]|9|@$A|28|B|29|1|2A]]|C|$]]|$1|15|3|11|5|6|7|2B|8|@]|9|@]|C|$]]|$1|16|3|17|5|6|7|2C|8|@]|9|@]|C|$]]|$1|18|3|11|5|6|7|2D|8|@]|9|@]|C|$]]|$1|19|3|13|5|14|7|2E|8|@]|9|@$A|2F|B|2G|1|2H]]|C|$]]|$1|1A|3|11|5|6|7|2I|8|@]|9|@]|C|$]]|$1|1B|3|-4|5|6|7|2J|8|@]|9|@]|C|$]]]|1C|$1D|$5|1E|1F|1G|C|$1H|1I]]|1J|$5|1K|1F|1L|C|$1M|1N|1O|-4]]|1P|$5|1K|1F|1L|C|$1M|1Q|1O|-4]]]]

According to the <a href="https://www.tensorflow.org/tensorboard/r2/graphs#op-level_graph" rel="noreferrer">docs</a>, you can use Tensorboard to visualise graphs once your model has been trained. 

First, define your model and run it. Then, open Tensorboard and switch to the Graph tab.

<hr>

Minimal Compilable Example

This example is taken from the docs. First, define your model and data.

<pre><code># Relevant imports.
%load_ext tensorboard

from __future__ import absolute_import
from __future__ import division
from __future__ import print_function

from datetime import datetime
from packaging import version

import tensorflow as tf
from tensorflow import keras

# Define the model.
model = keras.models.Sequential([
 keras.layers.Flatten(input_shape=(28, 28)),
 keras.layers.Dense(32, activation='relu'),
 keras.layers.Dropout(0.2),
 keras.layers.Dense(10, activation='softmax')
])

model.compile(
 optimizer='adam',
 loss='sparse_categorical_crossentropy',
 metrics=['accuracy'])

(train_images, train_labels), _ = keras.datasets.fashion_mnist.load_data()
train_images = train_images / 255.0
</code></pre>

Next, train your model. Here, you will need to define a callback for Tensorboard to use for visualising stats and graphs.

<pre><code># Define the Keras TensorBoard callback.
logdir="logs/fit/" + datetime.now().strftime("%Y%m%d-%H%M%S")
tensorboard_callback = keras.callbacks.TensorBoard(log_dir=logdir)

# Train the model.
model.fit(
 train_images,
 train_labels, 
 batch_size=64,
 epochs=5, 
 callbacks=[tensorboard_callback])
</code></pre>

After training, in your notebook, run 

<pre><code>%tensorboard --logdir logs
</code></pre>

And switch to the Graph tab in the navbar:

<a href="https://i.stack.imgur.com/15yFx.png" rel="noreferrer"><img src="https://i.stack.imgur.com/15yFx.png" alt="enter image description here"></a>

You will see a graph that looks a lot like this:

<a href="https://i.stack.imgur.com/Bwjuf.png" rel="noreferrer"><img src="https://i.stack.imgur.com/Bwjuf.png" alt="enter image description here"></a>

blocks|key|2009786|text|您可以可视化任何tf.function修饰函数的图形，但首先必须跟踪它的执行情况。|type|unstyled|depth|inlineStyleRanges|offset|length|style|CODE|entityRanges|data|2009787|可视化Keras模型的图形意味着可视化它的call方法。|2009788|默认情况下，此方法不是tf.function修饰的，因此必须将模型调用包装在一个正确修饰的函数中并执行它。|2009789|import+tensorflow+as+tf

model+=+tf.keras.Sequential(
++++[
++++++++tf.keras.layers.Flatten(input_shape=(28,+28)),
++++++++tf.keras.layers.Dense(32,+activation="relu"),
++++++++tf.keras.layers.Dropout(0.2),
++++++++tf.keras.layers.Dense(10,+activation="softmax"),
++++]
)


@tf.function
def+traceme(x):
++++return+model(x)


logdir+=+"log"
writer+=+tf.summary.create_file_writer(logdir)
tf.summary.trace_on(graph=True,+profiler=True)
#+Forward+pass
traceme(tf.zeros((1,+28,+28,+1)))
with+writer.as_default():
++++tf.summary.trace_export(name="model_trace",+step=0,+profiler_outdir=logdir)|code-block|syntax|javascript|2009790|entityMap^0|8|B|0|L|4|0|B|B|0|0^^$0|@$1|2|3|4|5|6|7|Q|8|@$9|R|A|S|B|C]]|D|@]|E|$]]|$1|F|3|G|5|6|7|T|8|@$9|U|A|V|B|C]]|D|@]|E|$]]|$1|H|3|I|5|6|7|W|8|@$9|X|A|Y|B|C]]|D|@]|E|$]]|$1|J|3|K|5|L|7|Z|8|@]|D|@]|E|$M|N]]|$1|O|3|-4|5|6|7|10|8|@]|D|@]|E|$]]]|P|$]]

You can visualize the graph of any <code>tf.function</code> decorated function, but first, you have to trace its execution.

Visualizing the graph of a Keras model means to visualize it's <code>call</code> method.

By default, this method is not <code>tf.function</code> decorated and therefore you have to wrap the model call in a function correctly decorated and execute it.

<pre class="lang-py prettyprint-override"><code>import tensorflow as tf

model = tf.keras.Sequential(
 [
 tf.keras.layers.Flatten(input_shape=(28, 28)),
 tf.keras.layers.Dense(32, activation="relu"),
 tf.keras.layers.Dropout(0.2),
 tf.keras.layers.Dense(10, activation="softmax"),
 ]
)


@tf.function
def traceme(x):
 return model(x)


logdir = "log"
writer = tf.summary.create_file_writer(logdir)
tf.summary.trace_on(graph=True, profiler=True)
# Forward pass
traceme(tf.zeros((1, 28, 28, 1)))
with writer.as_default():
 tf.summary.trace_export(name="model_trace", step=0, profiler_outdir=logdir)
</code></pre>

blocks|key|2009824|text|下面是基于tf.keras.callbacks.TensorBoard代码的当前对我有用的内容(TF+2.0.0)：|type|unstyled|depth|inlineStyleRanges|entityRanges|offset|length|data|2009825|#+After+model+has+been+compiled
from+tensorflow.python.ops+import+summary_ops_v2
from+tensorflow.python.keras.backend+import+get_graph
tb_path+=+'/tmp/tensorboard/'
tb_writer+=+tf.summary.create_file_writer(tb_path)
with+tb_writer.as_default():
++++if+not+model.run_eagerly:
++++++++summary_ops_v2.graph(get_graph(),+step=0)|code-block|syntax|javascript|2009826|entityMap|0|LINK|mutability|MUTABLE|url|https://www.tensorflow.org/api_docs/python/tf/keras/callbacks/TensorBoard^0|5|U|0|0|0^^$0|@$1|2|3|4|5|6|7|Q|8|@]|9|@$A|R|B|S|1|T]]|C|$]]|$1|D|3|E|5|F|7|U|8|@]|9|@]|C|$G|H]]|$1|I|3|-4|5|6|7|V|8|@]|9|@]|C|$]]]|J|$K|$5|L|M|N|C|$O|P]]]]

Here's what is working for me at the moment (TF 2.0.0), based on the <a href="https://www.tensorflow.org/api_docs/python/tf/keras/callbacks/TensorBoard" rel="nofollow noreferrer">tf.keras.callbacks.TensorBoard</a> code:

<pre><code># After model has been compiled
from tensorflow.python.ops import summary_ops_v2
from tensorflow.python.keras.backend import get_graph
tb_path = '/tmp/tensorboard/'
tb_writer = tf.summary.create_file_writer(tb_path)
with tb_writer.as_default():
 if not model.run_eagerly:
 summary_ops_v2.graph(get_graph(), step=0)
</code></pre>

blocks|key|2009845|text|另一种选择是使用这个网站：https://lutzroeder.github.io/netron/|type|unstyled|depth|inlineStyleRanges|entityRanges|offset|length|data|2009846|生成带有.h5或.tflite文件的图形。|2009847|它所基于的github回购可以在这里找到：https://github.com/lutzroeder/netron|2009848|entityMap|0|LINK|mutability|MUTABLE|url|https://lutzroeder.github.io/netron/|1|https://github.com/lutzroeder/netron^0|D|10|0|0|0|L|10|1|0^^$0|@$1|2|3|4|5|6|7|R|8|@]|9|@$A|S|B|T|1|U]]|C|$]]|$1|D|3|E|5|6|7|V|8|@]|9|@]|C|$]]|$1|F|3|G|5|6|7|W|8|@]|9|@$A|X|B|Y|1|Z]]|C|$]]|$1|H|3|-4|5|6|7|10|8|@]|9|@]|C|$]]]|I|$J|$5|K|L|M|C|$N|O]]|P|$5|K|L|M|C|$N|Q]]]]

Another option is to use this website: <a href="https://lutzroeder.github.io/netron/" rel="nofollow noreferrer">https://lutzroeder.github.io/netron/</a>

which generate a graph with a .h5 or .tflite file.

The github repo it's based on may be found here:
<a href="https://github.com/lutzroeder/netron" rel="nofollow noreferrer">https://github.com/lutzroeder/netron</a>

blocks|key|2009869|text|下面是tf2.x与子类模型/层的解决方案|type|unstyled|depth|inlineStyleRanges|offset|length|style|BOLD|entityRanges|data|2009870|import+tensorflow+as+tf
print("TensorFlow+version:",+tf.__version__)

from+tensorflow.keras.layers+import+Dense,+Flatten,+Conv2D
from+tensorflow.keras+import+Model,Input


class+MyModel(Model):
++++def+__init__(self,+dim):
++++++++super(MyModel,+self).__init__()
++++++++self.conv1+=+Conv2D(16,+3,+activation='relu')
++++++++self.conv2+=+Conv2D(32,+3,+activation='relu')
++++++++self.conv3+=+Conv2D(8,+3,+activation='relu')
++++++++self.flatten+=+Flatten()
++++++++self.d1+=+Dense(128,+activation='relu')
++++++++self.d2+=+Dense(1)

++++def+call(self,+x):
++++++++x+=+self.conv1(x)
++++++++x+=+self.conv2(x)
++++++++x+=+self.conv3(x)
++++++++x+=+self.flatten(x)
++++++++x+=+self.d1(x)
++++++++return+self.d2(x)
++++def+build_graph(self):
++++++++x+=+Input(shape=(dim))
++++++++return+Model(inputs=[x],+outputs=self.call(x))


dim+=+(28,+28,+1)

#+Create+an+instance+of+the+model
model+=+MyModel((dim))
model.build((None,+*dim))
model.build_graph().summary()


tf.keras.utils.plot_model(model.build_graph(),+to_file="model.png",
+++++++++++expand_nested=True,+show_shapes=True)|code-block|syntax|javascript|2009871|输出为|2009872|TensorFlow+version:+2.5.0

Model:+"model"
_________________________________________________________________
Layer+(type)+++++++++++++++++Output+Shape++++++++++++++Param+#+++
=================================================================
input_1+(InputLayer)+++++++++[(None,+28,+28,+1)]+++++++0+++++++++
_________________________________________________________________
conv2d+(Conv2D)++++++++++++++(None,+26,+26,+16)++++++++160+++++++
_________________________________________________________________
conv2d_1+(Conv2D)++++++++++++(None,+24,+24,+32)++++++++4640++++++
_________________________________________________________________
conv2d_2+(Conv2D)++++++++++++(None,+22,+22,+8)+++++++++2312++++++
_________________________________________________________________
flatten+(Flatten)++++++++++++(None,+3872)++++++++++++++0+++++++++
_________________________________________________________________
dense+(Dense)++++++++++++++++(None,+128)+++++++++++++++495744++++
_________________________________________________________________
dense_1+(Dense)++++++++++++++(None,+1)+++++++++++++++++129+++++++
=================================================================
Total+params:+502,985
Trainable+params:+502,985
Non-trainable+params:+0|2009873|这里还有一个图形可视化|2009874|​|2009875|📷|atomic|2009876|2009877|entityMap|0|LINK|mutability|MUTABLE|url|https://https//fantashit.com/graph-visualization-of-subclassed-model-layer/|1|IMAGE|IMMUTABLE|imageUrl|https://developer.qcloudimg.com/http-save/yehe-900000/df490309914b6bc2793bc7edfdfab5d8.png|imageAlt^0|3|5|9|6|0|0|0|0|3|0|0|6|5|0|0|0|1|1|0|0^^$0|@$1|2|3|4|5|6|7|1A|8|@$9|1B|A|1C|B|C]]|D|@$9|1D|A|1E|1|1F]]|E|$]]|$1|F|3|G|5|H|7|1G|8|@]|D|@]|E|$I|J]]|$1|K|3|L|5|6|7|1H|8|@$9|1I|A|1J|B|C]]|D|@]|E|$]]|$1|M|3|N|5|H|7|1K|8|@]|D|@]|E|$I|J]]|$1|O|3|P|5|6|7|1L|8|@$9|1M|A|1N|B|C]]|D|@]|E|$]]|$1|Q|3|R|5|6|7|1O|8|@]|D|@]|E|$]]|$1|S|3|T|5|U|7|1P|8|@]|D|@$9|1Q|A|1R|1|1S]]|E|$]]|$1|V|3|R|5|6|7|1T|8|@]|D|@]|E|$]]|$1|W|3|-4|5|6|7|1U|8|@]|D|@]|E|$]]]|X|$Y|$5|Z|10|11|E|$12|13]]|14|$5|15|10|16|E|$17|18|19|-4]]]]

Here is the solution for tf2.x with <a href="https://https://fantashit.com/graph-visualization-of-subclassed-model-layer/" rel="nofollow noreferrer">Graph visualization of subclassed model/layer</a>
<pre><code>import tensorflow as tf
print(&quot;TensorFlow version:&quot;, tf.__version__)

from tensorflow.keras.layers import Dense, Flatten, Conv2D
from tensorflow.keras import Model,Input


class MyModel(Model):
 def __init__(self, dim):
 super(MyModel, self).__init__()
 self.conv1 = Conv2D(16, 3, activation='relu')
 self.conv2 = Conv2D(32, 3, activation='relu')
 self.conv3 = Conv2D(8, 3, activation='relu')
 self.flatten = Flatten()
 self.d1 = Dense(128, activation='relu')
 self.d2 = Dense(1)

 def call(self, x):
 x = self.conv1(x)
 x = self.conv2(x)
 x = self.conv3(x)
 x = self.flatten(x)
 x = self.d1(x)
 return self.d2(x)
 def build_graph(self):
 x = Input(shape=(dim))
 return Model(inputs=[x], outputs=self.call(x))


dim = (28, 28, 1)

# Create an instance of the model
model = MyModel((dim))
model.build((None, *dim))
model.build_graph().summary()


tf.keras.utils.plot_model(model.build_graph(), to_file=&quot;model.png&quot;,
 expand_nested=True, show_shapes=True)
</code></pre>
the output is
<pre><code>TensorFlow version: 2.5.0

Model: &quot;model&quot;
_________________________________________________________________
Layer (type) Output Shape Param # 
=================================================================
input_1 (InputLayer) [(None, 28, 28, 1)] 0 
_________________________________________________________________
conv2d (Conv2D) (None, 26, 26, 16) 160 
_________________________________________________________________
conv2d_1 (Conv2D) (None, 24, 24, 32) 4640 
_________________________________________________________________
conv2d_2 (Conv2D) (None, 22, 22, 8) 2312 
_________________________________________________________________
flatten (Flatten) (None, 3872) 0 
_________________________________________________________________
dense (Dense) (None, 128) 495744 
_________________________________________________________________
dense_1 (Dense) (None, 1) 129 
=================================================================
Total params: 502,985
Trainable params: 502,985
Non-trainable params: 0
</code></pre>
Here is also a graph visualization
<a href="https://i.stack.imgur.com/aWnBe.png" rel="nofollow noreferrer"><img src="https://i.stack.imgur.com/aWnBe.png" alt="Here is also a graph visualization" /></a>

I upgraded to Tensorflow 2.0 and there is no <code>tf.summary.FileWriter("tf_graphs", sess.graph)</code>. I was looking through some other StackOverflow questions on this and they said to use <code>tf.compat.v1.summary etc</code>. Surely there must be a way to graph and visualize a tf.keras model in Tensorflow version 2. What is it? I'm looking for a tensorboard output like the one below. Thank you!

<a href="https://i.stack.imgur.com/Zhc3g.png" rel="noreferrer"><img src="https://i.stack.imgur.com/Zhc3g.png" alt="enter image description here"></a>

How to graph tf.keras model in Tensorflow-2.0?

翻译质量差，导致语言生硬或混乱。

没有提供实际的解决方法或示例。

解答不清晰，无法理解或解决问题。

页面排版不美观，阅读体验差。

文章

问答

视频

学习中心

腾讯云实验室

直播

竞赛

腾讯云代码分析专区

腾讯iOA零信任安全管理系统专区

腾讯云架构师技术同盟交流圈

腾讯云数据库专区

腾讯云顾问专区

腾讯云原生专区

腾讯混元专区

腾讯云TCE专区

腾讯云Lighthouse专区

腾讯云HAI专区

腾讯云Edgeone专区

腾讯云存储专区

腾讯云智能专区

腾讯轻联专区 

腾讯云开发专区

TAPD专区

腾讯轻量云游戏服专区

腾讯云最具价值专家

腾讯云架构师技术同盟

腾讯云创作之星

腾讯云开发者先锋

腾讯云代码助手

云原生构建

TAPD 敏捷项目管理

Cloud Studio

SDK中心

API中心

命令行工具

涵盖代码开发、场景应用、自动测试全流程，助你从零构建专属AI助手

一站式MCP教程库，解锁AI应用新玩法

我升级到了Tensorflow 2.0，而且没有tf.summary.FileWriter("tf_graphs", sess.graph)。我正在研究关于这个的其他一些StackOverflow问题，他们说要使用tf.compat.v1.summary etc。当然，在Tensorflow版本2中，必须有一种方法来绘...

问如何在Tensorflow-2.0中绘制tf.keras模型？
EN

Stack Overflow用户

社区

活动

圈层

关于

腾讯云开发者

热门产品

热门推荐

更多推荐

问如何在Tensorflow-2.0中绘制tf.keras模型？EN