Closed shuang1330 closed 7 years ago
BhagyeshVikani
commented
7 years ago Thank you for reporting the issue. Please reinstall with sudo python setup.py install and check whether the bug is fixed.
And please share details about the second model you created so that we can debug it.
shuang1330
commented
7 years ago Hi, thank you for helping me with that! I didn't use the setup.py , instead, I just copy the entire folder to run it... because when I used the setup.py file, I would install it to the system python path, but I'm using anaconda instead, and when I call python from the terminal, the annconda python will be used. I'm not sure why the setup.py install it into the other python path..
The model I'm using is copied from somewhere in github, can't find the source now.. But it's a pretty simple one:
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import argparse
import sys
from tensorflow.examples.tutorials.mnist import input_data
import tensorflow as tf
from tensorflow.python.client import timeline
import tf_cnnvis.tf_cnnvis
FLAGS = None
def deepnn(x):
x_image = tf.reshape(x, [-1, 28, 28, 1])
W_conv1 = weight_variable([5, 5, 1, 32])
b_conv1 = bias_variable([32])
h_conv1 = tf.nn.relu(conv2d(x_image, W_conv1) + b_conv1)
h_pool1 = max_pool_2x2(h_conv1)
W_conv2 = weight_variable([5, 5, 32, 64])
b_conv2 = bias_variable([64])
h_conv2 = tf.nn.relu(conv2d(h_pool1, W_conv2) + b_conv2)
h_pool2 = max_pool_2x2(h_conv2)
W_fc1 = weight_variable([7 * 7 * 64, 1024])
b_fc1 = bias_variable([1024])
h_pool2_flat = tf.reshape(h_pool2, [-1, 7*7*64])
h_fc1 = tf.nn.relu(tf.matmul(h_pool2_flat, W_fc1) + b_fc1)
keep_prob = tf.placeholder(tf.float32)
h_fc1_drop = tf.nn.dropout(h_fc1, keep_prob)
W_fc2 = weight_variable([1024, 10])
b_fc2 = bias_variable([10])
y_conv = tf.matmul(h_fc1_drop, W_fc2) + b_fc2
return y_conv, keep_prob
def conv2d(x, W):
return tf.nn.conv2d(x, W, strides=[1, 1, 1, 1], padding='SAME')
def max_pool_2x2(x):
return tf.nn.max_pool(x, ksize=[1, 2, 2, 1],
strides=[1, 2, 2, 1], padding='SAME')
def weight_variable(shape):
initial = tf.truncated_normal(shape, stddev=0.1)
return tf.Variable(initial)
def bias_variable(shape):
initial = tf.constant(0.1, shape=shape)
return tf.Variable(initial)
if __name__ == '__main__':
mnist = input_data.read_data_sets('MNIST_data/', one_hot=True)
x = tf.placeholder(tf.float32, [None, 784])
y_ = tf.placeholder(tf.float32, [None, 10])
y_conv, keep_prob = deepnn(x)
cross_entropy = tf.reduce_mean(
tf.nn.softmax_cross_entropy_with_logits(labels=y_, logits=y_conv))
train_step = tf.train.AdamOptimizer(1e-4).minimize(cross_entropy)
correct_prediction = tf.equal(tf.argmax(y_conv, 1), tf.argmax(y_, 1))
accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
run_options = tf.RunOptions(trace_level=tf.RunOptions.FULL_TRACE)
run_metadata = tf.RunMetadata()
with tf.Session() as sess:
sess.run(tf.global_variables_initializer())
batch = mnist.train.next_batch(50)
feed_dict = {x: batch[0], y_: batch[1], keep_prob: 0.5}
sess.run((y_conv, keep_prob), feed_dict, options = run_options, run_metadata=run_metadata)
tl = timeline.Timeline(run_metadata.step_stats)
ctf = tl.generate_chrome_trace_format()
with open('timeline.json', 'w') as f:
f.write(ctf)
tf_cnnvis.get_visualization(graph_or_path = tf.get_default_graph(), value_feed_dict = feed_dict, input_tensor=None, layers=['r','p','c]', path_logdir='./Log',
path_outdir='./Output', force=False, n=8)`
javiribera
commented
7 years ago I can confirm these errors even when not using anaconda, but a virtualenv. Way to reproduce it:
pip install Pillow numpy scipy tensorflow h5pyjupyter nbconvert --to script tf_cnnvis_Example1.ipynb(tf_cnnvis) ╭─javiribera@sonic ~/Downloads/tf_cnnvis/examples ‹master*›
╰─$ py tf_cnnvis_Example1.py
100% [......................................................................] 243904576 / 2439045762
Saved under ./alexnet_weights.h5
W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE3 instructions, but these are available on your machine and could speed up CPU computations.
W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.1 instructions, but these are available on your machine and could speed up CPU computations.
W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.2 instructions, but these are available on your machine and could speed up CPU computations.
W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX instructions, but these are available on your machine and could speed up CPU computations.
No Layer with layer name = conv1
No Layer with layer name = conv2_1
No Layer with layer name = conv2_2
No Layer with layer name = conv3
No Layer with layer name = conv4_1
No Layer with layer name = conv4_2
No Layer with layer name = conv5_1
No Layer with layer name = conv5_2
Skipping. Too many featuremap. May cause memory errors.
Skipping. Too many featuremap. May cause memory errors.
No Layer with layer name = MaxPool
No Layer with layer name = MaxPool_1
No Layer with layer name = MaxPool_2
No Layer with layer name = MaxPool_3
No Layer with layer name = MaxPool_4
No Layer with layer name = Conv2D
No Layer with layer name = Conv2D_1
No Layer with layer name = Conv2D_2
No Layer with layer name = Conv2D_3
No Layer with layer name = Conv2D_4
No Layer with layer name = Conv2D_5
No Layer with layer name = Conv2D_6
No Layer with layer name = Conv2D_7
Total Time = 37.623156No file is created in the current directory.
javiribera
commented
7 years ago For some reason the error above is fixed after this pull request...: https://github.com/InFoCusp/tf_cnnvis/pull/2
(test) ╭─javiribera@sonic ~/Downloads/tf_cnnvis/examples ‹master*›
╰─$ py tf_cnnvis_Example1.py
W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE3 instructions, but these are available on your machine and could speed up CPU computations.
W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.1 instructions, but these are available on your machine and could speed up CPU computations.
W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.2 instructions, but these are available on your machine and could speed up CPU computations.
W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX instructions, but these are available on your machine and could speed up CPU computations.
Reconstruction Completed for conv1 layer. Time taken = 3.557765 s
Reconstruction Completed for conv2_1 layer. Time taken = 10.330736 s
Reconstruction Completed for conv2_2 layer. Time taken = 11.265965 s
Reconstruction Completed for conv3 layer. Time taken = 38.331542 s
Reconstruction Completed for conv4_1 layer. Time taken = 19.666729 s
Reconstruction Completed for conv4_2 layer. Time taken = 20.610766 s
Reconstruction Completed for conv5_1 layer. Time taken = 10.944213 s
Reconstruction Completed for conv5_2 layer. Time taken = 10.348010 s
Skipping. Too many featuremap. May cause memory errors.
Skipping. Too many featuremap. May cause memory errors.
Reconstruction Completed for MaxPool layer. Time taken = 4.010168 s
Reconstruction Completed for MaxPool_1 layer. Time taken = 10.653442 s
Reconstruction Completed for MaxPool_2 layer. Time taken = 11.627388 s
Reconstruction Completed for MaxPool_3 layer. Time taken = 11.479884 s
Reconstruction Completed for MaxPool_4 layer. Time taken = 11.486936 s
Reconstruction Completed for Conv2D layer. Time taken = 4.330865 s
Reconstruction Completed for Conv2D_1 layer. Time taken = 13.503956 s
Reconstruction Completed for Conv2D_2 layer. Time taken = 12.696401 s
Reconstruction Completed for Conv2D_3 layer. Time taken = 46.248193 s
Reconstruction Completed for Conv2D_4 layer. Time taken = 23.508405 s
Reconstruction Completed for Conv2D_5 layer. Time taken = 23.400738 s
Reconstruction Completed for Conv2D_6 layer. Time taken = 16.414462 s
Reconstruction Completed for Conv2D_7 layer. Time taken = 17.022638 s
Total Time = 333.895082Hi @shuang1330 Here's the slightly modified script you'll need to use- explanation added below.
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import argparse
import sys
from tensorflow.examples.tutorials.mnist import input_data
import tensorflow as tf
from tensorflow.python.client import timeline
import tf_cnnvis.tf_cnnvis
FLAGS = None
def deepnn(x):
x_image = tf.reshape(x, [-1, 28, 28, 1])
W_conv1 = weight_variable([5, 5, 1, 32])
b_conv1 = bias_variable([32])
h_conv1 = tf.nn.relu(conv2d(x_image, W_conv1) + b_conv1)
h_pool1 = max_pool_2x2(h_conv1)
W_conv2 = weight_variable([5, 5, 32, 64])
b_conv2 = bias_variable([64])
h_conv2 = tf.nn.relu(conv2d(h_pool1, W_conv2) + b_conv2)
h_pool2 = max_pool_2x2(h_conv2)
W_fc1 = weight_variable([7 * 7 * 64, 1024])
b_fc1 = bias_variable([1024])
h_pool2_flat = tf.reshape(h_pool2, [-1, 7*7*64])
h_fc1 = tf.nn.relu(tf.matmul(h_pool2_flat, W_fc1) + b_fc1)
keep_prob = tf.placeholder(tf.float32)
h_fc1_drop = tf.nn.dropout(h_fc1, keep_prob)
W_fc2 = weight_variable([1024, 10])
b_fc2 = bias_variable([10])
y_conv = tf.matmul(h_fc1_drop, W_fc2) + b_fc2
return x_image, y_conv, keep_prob
def conv2d(x, W):
return tf.nn.conv2d(x, W, strides=[1, 1, 1, 1], padding='SAME')
def max_pool_2x2(x):
return tf.nn.max_pool(x, ksize=[1, 2, 2, 1],
strides=[1, 2, 2, 1], padding='SAME')
def weight_variable(shape):
initial = tf.truncated_normal(shape, stddev=0.1)
return tf.Variable(initial)
def bias_variable(shape):
initial = tf.constant(0.1, shape=shape)
return tf.Variable(initial)
if __name__ == '__main__':
mnist = input_data.read_data_sets('MNIST_data/', one_hot=True)
x = tf.placeholder(tf.float32, [None, 784])
y_ = tf.placeholder(tf.float32, [None, 10])
x_image, y_conv, keep_prob = deepnn(x)
cross_entropy = tf.reduce_mean(
tf.nn.softmax_cross_entropy_with_logits(labels=y_, logits=y_conv))
train_step = tf.train.AdamOptimizer(1e-4).minimize(cross_entropy)
correct_prediction = tf.equal(tf.argmax(y_conv, 1), tf.argmax(y_, 1))
accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
run_options = tf.RunOptions(trace_level=tf.RunOptions.FULL_TRACE)
run_metadata = tf.RunMetadata()
with tf.Session() as sess:
sess.run(tf.global_variables_initializer())
batch = mnist.train.next_batch(50)
feed_dict = {x: batch[0], y_: batch[1], keep_prob: 0.5}
sess.run((y_conv, keep_prob), feed_dict, options = run_options, run_metadata=run_metadata)
tl = timeline.Timeline(run_metadata.step_stats)
ctf = tl.generate_chrome_trace_format()
with open('timeline.json', 'w') as f:
f.write(ctf)
tf_cnnvis.get_visualization(graph_or_path = tf.get_default_graph(), value_feed_dict = feed_dict, input_tensor=x_image, layers=['r','p','c'], path_logdir='./Log', path_outdir='./Output', force=False, n=8)The tf_cnnvis library is designed to compute the reconstructed image for input tensors which are 4D (samples x height x width x channels). The above network was taking in the images in a flattened format, i.e. (samples x (h*w*c)) at the first placeholder. These were being reshaped at a later point. The only change needed was to pass this reshaped tensor as the input tensor argument into get_visualization . That's been done in the above script and also thanks to the changes by @javiribera -- this works now.
Thanks to @shuang1330, we have now also included this in our examples directory.
BhagyeshVikani
commented
7 years ago Please test and close the issue if the bug has been resolved.
Hi! I got this error when I tried to run the example.. It says:
No Layer with layer name = conv1 No Layer with layer name = conv2_1 No Layer with layer name = conv2_2 No Layer with layer name = conv3 No Layer with layer name = conv4_1 No Layer with layer name = conv4_2 No Layer with layer name = conv5_1 No Layer with layer name = conv5_2 Skipping. Too many featuremap. May cause memory errors. Skipping. Too many featuremap. May cause memory errors. No Layer with layer name = MaxPool No Layer with layer name = MaxPool_1 No Layer with layer name = MaxPool_2 No Layer with layer name = MaxPool_3 No Layer with layer name = MaxPool_4 Total Time = 39.663317
When I tried to use the command: tf_cnnvis.get_visualization(graph_or_path = tf.get_default_graph(), value_feed_dict = feed_dict, input_tensor=None, layers=['r','p','c]', path_logdir='./Log', path_outdir='./Output', force=False, n=8) I tried it in a simple model with only 2 conv layers, 1 max_pool and 2 fc. It doesn't generate any outputs/log files..
Thank you in advance for looking into the problem I'm having.