Autograph failure with `\` #35765

emailweixu · 2020-01-11T06:54:34Z

Please make sure that this is a bug. As per our GitHub Policy, we only address code/doc bugs, performance issues, feature requests and build/installation issues on GitHub. tag:bug_template

System information

Have I written custom code (as opposed to using a stock example script provided in TensorFlow):
OS Platform and Distribution (e.g., Linux Ubuntu 16.04): Ubuntu 18.04
Mobile device (e.g. iPhone 8, Pixel 2, Samsung Galaxy) if the issue happens on mobile device:
TensorFlow installed from (source or binary): binary
TensorFlow version (use command below): v2.1.0-rc2-17-ge5bf8de 2.1.0
Python version: 3.6
Bazel version (if compiling from source):
GCC/Compiler version (if compiling from source):
CUDA/cuDNN version:
GPU model and memory: Nvidia RTX 2080

Describe the current behavior
Tensorflow shows warning about failure of autograph

WARNING:tensorflow:AutoGraph could not transform <bound method C.f of <__main__.C object at 0x7f4a83904668>> 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: expected exactly one node node, found [<gast.gast.FunctionDef object at 0x7f4a47e0d5c0>, <gast.gast.Return object at 0x7f4a47e0df28>]

The warning seems to be caused by the backslash "\".

Describe the expected behavior
There should be no such warning

Code to reproduce the issue

import tensorflow as tf

class C(object):
    def f(self):
        # error disappear if \ in the following line is removed
        a = \
            1
        return a

obj = C()

@tf.function
def func():
    mem =  obj.f()
    return mem

def main():
    print(func())

if __name__ == "__main__":
    main()

The text was updated successfully, but these errors were encountered:

TF autograph has bug handling `\` (tensorflow/tensorflow#35765)

r-wheeler · 2020-01-13T20:53:00Z

Just ran into this as well, removing the slash fixed it.

jvishnuvardhan · 2020-01-13T21:57:07Z

I am closing this as the related PR fix merged. Thanks!

tensorflow-bot · 2020-01-14T19:47:42Z

Are you satisfied with the resolution of your issue?
Yes
No

emailweixu · 2020-01-14T22:39:14Z

@jvishnuvardhan which PR fixed this bug?

jvishnuvardhan · 2020-01-14T23:44:19Z

@emailweixu Sorry. I thought it was merged into master branch in tensorflow. It was my mistake. I am reopening it. Thanks!

ref. tensorflow/tensorflow#35765

TF autograph has bug handling `\` (tensorflow/tensorflow#35765)

mdanatg · 2020-01-15T23:59:26Z

Looks like a bug in the parser. dedent_block seems to be confused by the backslash continuation:

s = r'''
    def f():
        a = \
            1
        return a
'''

print(dedent_block(s))

def f():
    a = \
    1
return a

mdanatg · 2020-01-17T21:40:01Z

For a workaround until this is fixed, note that you can use parentheses to break expressions on multiple lines:

a = (
    1)

tensorflow-bot · 2020-01-20T13:54:25Z

Are you satisfied with the resolution of your issue?
Yes
No

Currently, for some function, TF2 autograph will fail. See tensorflow/tensorflow#35765 tensorflow/tensorflow#30149 tensorflow/autograph#3 If autograph failed, the functions will be run eagerly and TF cannot optimize them. So we'd better address them. ## Manually test ~~~python df1 = spark.range(100) from petastorm.spark import make_spark_converter # Set a cache directory on DBFS FUSE for intermediate data. spark.conf.set("petastorm.spark.converter.parentCacheDirUrl", "file:///dbfs/ml/tmp/petastorm/QA/bugs/") converter1 = make_spark_converter(df1) with converter1.make_tf_dataset(num_epochs=1) as dataset: for batch in dataset: print(batch.id) ~~~ * Before Output includes: ``` WARNING:tensorflow:AutoGraph could not transform <function _NamedtupleCache.get at 0x7f0bfbe6f200> 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: expected exactly one node node, found [<gast.gast.FunctionDef object at 0x7f0bfad5d050>, <gast.gast.Return object at 0x7f0bfad5d7d0>] WARNING:tensorflow:AutoGraph could not transform <function make_petastorm_dataset.<locals>.<lambda> at 0x7f0bf8da03b0> and will run it as-is. Cause: could not parse the source code: .map(lambda row: _set_shape_to_named_tuple(reader.schema, row, reader.batched_output)) This error may be avoided by creating the lambda in a standalone statement. ``` * After The warnings listed above disappear.

Autograph isn't able to trace functions with a backslash continuation, see: tensorflow/tensorflow#35765

Currently, for some function, TF2 autograph will fail. See tensorflow/tensorflow#35765 tensorflow/tensorflow#30149 tensorflow/autograph#3 If autograph failed, the functions will be run eagerly and TF cannot optimize them. So we'd better address them. ## Manually test ~~~python df1 = spark.range(100) from petastorm.spark import make_spark_converter # Set a cache directory on DBFS FUSE for intermediate data. spark.conf.set("petastorm.spark.converter.parentCacheDirUrl", "file:///dbfs/ml/tmp/petastorm/QA/bugs/") converter1 = make_spark_converter(df1) with converter1.make_tf_dataset(num_epochs=1) as dataset: for batch in dataset: print(batch.id) ~~~ * Before Output includes: ``` WARNING:tensorflow:AutoGraph could not transform <function _NamedtupleCache.get at 0x7f0bfbe6f200> 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: expected exactly one node node, found [<gast.gast.FunctionDef object at 0x7f0bfad5d050>, <gast.gast.Return object at 0x7f0bfad5d7d0>] WARNING:tensorflow:AutoGraph could not transform <function make_petastorm_dataset.<locals>.<lambda> at 0x7f0bf8da03b0> and will run it as-is. Cause: could not parse the source code: .map(lambda row: _set_shape_to_named_tuple(reader.schema, row, reader.batched_output)) This error may be avoided by creating the lambda in a standalone statement. ``` * After The warnings listed above disappear.

emailweixu added a commit to HorizonRobotics/alf that referenced this issue Jan 11, 2020

Fix autograph failure

f1da958

TF autograph has bug handling `\` (tensorflow/tensorflow#35765)

emailweixu mentioned this issue Jan 11, 2020

Fix autograph failure HorizonRobotics/alf#304

Merged

emailweixu added a commit to HorizonRobotics/alf that referenced this issue Jan 12, 2020

Fix autograph failure (#304)

d7e3a46

TF autograph has bug handling `\` (tensorflow/tensorflow#35765)

oanush self-assigned this Jan 13, 2020

oanush added comp:autograph Autograph related issues TF 2.1 for tracking issues in 2.1 release type:bug Bug labels Jan 13, 2020

oanush assigned jvishnuvardhan and unassigned oanush Jan 13, 2020

jvishnuvardhan closed this as completed Jan 13, 2020

jvishnuvardhan reopened this Jan 14, 2020

jvishnuvardhan closed this as completed Jan 14, 2020

tensorflow deleted a comment from tensorflow-bot bot Jan 14, 2020

jvishnuvardhan reopened this Jan 14, 2020

MokkeMeguru added a commit to MokkeMeguru/TFGENZOO that referenced this issue Jan 15, 2020

remove slash

0055b11

ref. tensorflow/tensorflow#35765

Haichao-Zhang pushed a commit to Haichao-Zhang/alf that referenced this issue Jan 15, 2020

Fix autograph failure (HorizonRobotics#304)

8543209

TF autograph has bug handling `\` (tensorflow/tensorflow#35765)

jvishnuvardhan assigned mdanatg and unassigned jvishnuvardhan Jan 15, 2020

jvishnuvardhan added the stat:awaiting tensorflower Status - Awaiting response from tensorflower label Jan 15, 2020

tensorflowbutler removed the stat:awaiting tensorflower Status - Awaiting response from tensorflower label Jan 16, 2020

This was referenced Jan 17, 2020

Autograph transformation warning #35810

Closed

AutoGraph unexpected indent in tf-nightly-gpu-2.1.0.dev20191103 #33962

Closed

tensorflow-copybara closed this as completed in 8e3adf7 Jan 20, 2020

emailweixu mentioned this issue Feb 4, 2020

Implementation of Model-based RL HorizonRobotics/alf#325

Merged

goldiegadde added this to Done in TensorFlow 2.2.0 Feb 6, 2020

zhangxuan1918 added a commit to zhangxuan1918/tensorflow3DMMRendering that referenced this issue Mar 15, 2020

remove backslash from code; see tensorflow/tensorflow#35765

847b6e6

zhuzilin mentioned this issue Mar 24, 2020

[Autograph] Fix Autograph '\' error #37851

Closed

jvishnuvardhan mentioned this issue Mar 24, 2020

Autograd cannot handle if conditions with line brakes when accessing fields #36899

Closed

WeichenXu123 mentioned this issue Apr 15, 2020

Address several autograph failed issues for TF2 uber/petastorm#542

Merged

mdanatg mentioned this issue Jun 22, 2020

AutoGraph "could not transform" warning when code contains a multi-line string with backslashes #40589

Closed

liuhenry added a commit to liuhenry/gkp-rl that referenced this issue Aug 19, 2020

Fix autograph failure in oscillator_qubit

475ece5

Autograph isn't able to trace functions with a backslash continuation, see: tensorflow/tensorflow#35765

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Autograph failure with `\` #35765

Autograph failure with `\` #35765

emailweixu commented Jan 11, 2020

r-wheeler commented Jan 13, 2020

jvishnuvardhan commented Jan 13, 2020

tensorflow-bot bot commented Jan 14, 2020

emailweixu commented Jan 14, 2020

jvishnuvardhan commented Jan 14, 2020

mdanatg commented Jan 15, 2020

mdanatg commented Jan 17, 2020

tensorflow-bot bot commented Jan 20, 2020

Autograph failure with \ #35765

Autograph failure with \ #35765

Comments

emailweixu commented Jan 11, 2020

r-wheeler commented Jan 13, 2020

jvishnuvardhan commented Jan 13, 2020

tensorflow-bot bot commented Jan 14, 2020

emailweixu commented Jan 14, 2020

jvishnuvardhan commented Jan 14, 2020

mdanatg commented Jan 15, 2020

mdanatg commented Jan 17, 2020

tensorflow-bot bot commented Jan 20, 2020

Autograph failure with `\` #35765

Autograph failure with `\` #35765