WARNING:tensorflow:AutoGraph could not transform <function format_example at ...> and will run it as-is

Question

My question is related to this and this one here. I am using PyCharm on Windows and Python 3.7.8 and Tensorflow 2.2.0:

print (sys.version)
3.7.8 (tags/v3.7.8:4b47a5b6ba, Jun 28 2020, 08:53:46) [MSC v.1916 64 bit (AMD64)]

print(tf.__version__)
2.2.0

When I run this code from this colab tutorial:

import os

import numpy as np

import matplotlib.pyplot as plt
import tensorflow as tf
import tensorflow_datasets as tfds

(raw_train, raw_validation, raw_test), metadata = tfds.load(
    'cats_vs_dogs',
    split=['train[:80%]', 'train[80%:90%]', 'train[90%:]'],
    with_info=True,
    as_supervised=True,
)
IMG_SIZE = 160 # All images will be resized to 160x160

def format_example(image, label):
  image = tf.cast(image, tf.float32)
  image = (image/127.5) - 1
  image = tf.image.resize(image, (IMG_SIZE, IMG_SIZE))
  return image, label

and then try to run this line:

train = raw_train.map(format_example)

I get the WARNING:

WARNING:tensorflow:AutoGraph could not transform <function format_example at 0x00000265A2DB4E58> and will run it as-is.
Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.
Cause: Unable to locate the source code of <function format_example at 0x00000265A2DB4E58>. Note that functions defined in certain environments, like the interactive Python shell do not expose their source code. If that is the case, you should to define them in a .py source file. If you are certain the code is graph-compatible, wrap the call using @tf.autograph.do_not_convert. Original error: could not get source code
To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert

I only get this WARNING when using PyCharm on my local pc. When I run it in colab no problems. What is the issue here and is this relevant or can I ignore the WARNING?

Answer 1

Like OP I get this warning in JupyterLab but not in Colab. But this only happens if I use a lambda.

train = raw_train.map(lambda x: x)

If I use a named function like the OP is doing here the warning goes away.

train = raw_train.map(format_example)

DataSet.map() always builds a graph even if you are in the eager mode. I think the graph builder is running into some kind of confusion.