tensorflow / community

Stores documents used by the TensorFlow developer community

Documentation

TensorFlow is an end-to-end open source platform for machine learning. It has a comprehensive, flexible ecosystem of tools, libraries, and community resources that lets researchers push the state-of-the-art in ML and developers easily build and deploy ML-powered applications.

TensorFlow was originally developed by researchers and engineers working within the Machine Intelligence team at Google Brain to conduct research in machine learning and neural networks. However, the framework is versatile enough to be used in other areas as well.

TensorFlow provides stable Python and C++ APIs, as well as a non-guaranteed backward compatible API for other languages.

Keep up-to-date with release announcements and security updates by subscribing to [email protected]. See all the mailing lists.

Install

See the TensorFlow install guide for the pip package, to enable GPU support, use a Docker container, and build from source.

To install the current release, which includes support for CUDA-enabled GPU cards (Ubuntu and Windows):

$ pip install tensorflow

Other devices (DirectX and MacOS-metal) are supported using Device plugins.

A smaller CPU-only package is also available:

$ pip install tensorflow-cpu

To update TensorFlow to the latest version, add --upgrade flag to the above commands.

Nightly binaries are available for testing using the tf-nightly and tf-nightly-cpu packages on PyPi.

Try your first TensorFlow program

$ python

>>> import tensorflow as tf
>>> tf.add(1, 2).numpy()
3
>>> hello = tf.constant('Hello, TensorFlow!')
>>> hello.numpy()
b'Hello, TensorFlow!'

For more examples, see the TensorFlow tutorials.

Contribution guidelines

If you want to contribute to TensorFlow, be sure to review the contribution guidelines. This project adheres to TensorFlow's code of conduct. By participating, you are expected to uphold this code.

We use GitHub issues for tracking requests and bugs, please see TensorFlow Forum for general questions and discussion, and please direct specific questions to Stack Overflow.

The TensorFlow project strives to abide by generally accepted best practices in open-source software development.

Patching guidelines

Follow these steps to patch a specific version of TensorFlow, for example, to apply fixes to bugs or security vulnerabilities:

• Clone the TensorFlow repo and switch to the corresponding branch for your desired TensorFlow version, for example, branch r2.8 for version 2.8.
• Apply (that is, cherry-pick) the desired changes and resolve any code conflicts.
• Run TensorFlow tests and ensure they pass.
• Build the TensorFlow pip package from source.

Continuous build status

You can find more community-supported platforms and configurations in the TensorFlow SIG Build community builds table.

Official Builds

Build Type	Status	Artifacts
Linux CPU		PyPI
Linux GPU		PyPI
Linux XLA		TBA
macOS		PyPI
Windows CPU		PyPI
Windows GPU		PyPI
Android		Download
Raspberry Pi 0 and 1		Py3
Raspberry Pi 2 and 3		Py3
Libtensorflow MacOS CPU	Status Temporarily Unavailable	Nightly Binary Official GCS
Libtensorflow Linux CPU	Status Temporarily Unavailable	Nightly Binary Official GCS
Libtensorflow Linux GPU	Status Temporarily Unavailable	Nightly Binary Official GCS
Libtensorflow Windows CPU	Status Temporarily Unavailable	Nightly Binary Official GCS
Libtensorflow Windows GPU	Status Temporarily Unavailable	Nightly Binary Official GCS

Resources

• TensorFlow.org
• TensorFlow Tutorials
• TensorFlow Official Models
• TensorFlow Examples
• TensorFlow Codelabs
• TensorFlow Blog
• Learn ML with TensorFlow
• TensorFlow Twitter
• TensorFlow YouTube
• TensorFlow model optimization roadmap
• TensorFlow White Papers
• TensorBoard Visualization Toolkit
• TensorFlow Code Search

Learn more about the TensorFlow community and how to contribute.

Courses

• Coursera
• Udacity
• Edx

License

Apache License 2.0

community's Issues

Link to TensorFlow roadmap is broken

In https://www.tensorflow.org/community, the link "See product roadmap" at the end of the page is broken.

Best way to contact mentors GCI 2019

Hi!

Hi! Just asking about GCI and the contacting mentors part, is there any IRC channel or Slack channel that we should join, or should we just ask here (the chat and email list redirect to this repository.

Thanks so much!

Building Tensorflow API in ./configure.py comma-separated list does not work correctly

The comma-separated lsit does not function correctly,
If I input the library folder first, it finds it but complains about the include folder,
If I inptu the include folder first, it complains about the library folder

Please specify the CUDA SDK version you want to use. [Leave empty to default to CUDA 10]:

Please specify the cuDNN version you want to use. [Leave empty to default to cuDNN 7]:

Please specify the comma-separated list of base paths to look for CUDA libraries and headers. [Leave empty to use the default]: C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.0\include, C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.0\lib\x64

Could not find any cudart.lib in any subdirectory:
''
'lib64'
'lib'
'lib/-linux-gnu'
'lib/x64'
'extras/CUPTI/'
of:
'C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.0\include'
Asking for detailed CUDA configuration...

Please specify the CUDA SDK version you want to use. [Leave empty to default to CUDA 10]:

Please specify the cuDNN version you want to use. [Leave empty to default to cuDNN 7]:

Please specify the comma-separated list of base paths to look for CUDA libraries and headers. [Leave empty to use the default]: C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.0\lib\x64 C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.0\include

Could not find any cuda.h matching version '10' in any subdirectory:
''
'include'
'include/cuda'
'include/*-linux-gnu'
'extras/CUPTI/include'
'include/cuda/CUPTI'
of:
Asking for detailed CUDA configuration...

Add a custom Op

How to add a custom op in Windows system?thank you!!!

SIG Add-ons: Finalize documentation and examples, and publish first package.

TF 2.0 might have deprecated tf.contrib, but much of its functionality is widely-used and important to the TensorFlow community. Addons is a repository of those bleeding edge contributions that conform to well-established API patterns, but implement new functionality not available in core TensorFlow.

For more information on Addons, check out their GitHub repo, their community charter, and their mailing group.

To become mentor for Google Code In

I wish to become a mentor for Google Code in 2019. I have done several projects in Machine Learning and few certifications too. It will be a nice exposure for me if I could get a chance to become a mentor for Google code at TensorFlow

My LinkedIn profile
My Github profile

NOTE: As I am not sure on how to approach TensorFlow on becoming mentor for Google Code in, I am creating an issue for it. Kindly let me know if there is some other way that I should follow for applying.

Tensorflow-gpu is not using any gpu power

Windows 10 x64
GTX 1060 6GB
CUDA - 10.0, 10.1
cuDNN - 7.6.1, 7.6.1
Python - 3.6.5
Anaconda newest version and updated using "conda update conda(-build)"

My installation was as follows:-

Python
Anaconda
Visual Studio - Desktop development with C++
CUDA and cuDNN in chronological order
pip install tensorflow-gpu (ignore installed and upgrade)

I am trying to work on a few simple TensorFlow projects, but over multiple installations, TensorFlow has not used the GPU power it has at its disposal. The specified installation was done after a clean Windows install as well. Tensorflow works in general, but it absolutely does not use any GPU processing power, but occupies 4.5GB of GPU memory.

I need to get this working as soon as possible. Thanks in advance.

Installation of @tensorflowtfjs-node is not possible

This is when I installed it.
I can't solve it.

E:\jslife_WX\wxjslife_rebuild_face>npm install

@tensorflow/[email protected] install E:\jslife_WX\wxjslife_rebuild_face\node_modules@tensorflow\tfjs-node
node scripts/install.js

• Downloading libtensorflow
  events.js:173
  throw er; // Unhandled 'error' event
  ^

Error: connect ETIMEDOUT 172.217.160.80:443
at TCPConnectWrap.afterConnect [as oncomplete] (net.js:1083:14)
Emitted 'error' event at:
at TLSSocket.socketErrorListener (_http_client.js:397:9)
at TLSSocket.emit (events.js:197:13)
at emitErrorNT (internal/streams/destroy.js:82:8)
at emitErrorAndCloseNT (internal/streams/destroy.js:50:3)
at processTicksAndRejections (internal/process/next_tick.js:76:17)
npm WARN [email protected] requires a peer of webpack@^2.0.0 || ^3.0.0 || ^4.0.0 but none is installed. You must install peer dependencies yourself.
npm WARN @tensorflow/[email protected] requires a peer of @tensorflow/[email protected] but none is installed. You must install peer dependencies yourself.
npm WARN @tensorflow/[email protected] requires a peer of @tensorflow/[email protected] but none is installed.
You must install peer dependencies yourself.
npm WARN @tensorflow/[email protected] requires a peer of @tensorflow/[email protected] but none is installed. You must install peer dependencies yourself.

npm ERR! code ELIFECYCLE
npm ERR! errno 1
npm ERR! @tensorflow/[email protected] install: node scripts/install.js
npm ERR! Exit status 1
npm ERR!
npm ERR! Failed at the @tensorflow/[email protected] install script.
npm ERR! This is probably not a problem with npm. There is likely additional logging output above.

WARNING: The TensorFlow contrib module will not be included in TensorFlow 2.0. For more information, please see: * https://github.com/tensorflow/community/blob/master/rfcs/20180907-contrib-sunset.md * https://github.com/tensorflow/addons If you depend on functionality not listed there, please file an issue.

What about enable GitHub Pages for this repo?

As most materials here are documents (written in markdown), maybe a jekyll based website will be much more friendly for users to reach already existed rfcs and so on?

Invalid Email for Project Steward

I tried sending an email to project steward Sarah Novotny, please can the email be reviewed?

Application to become a Google Code-In 2019 mentor for TensorFlow

I wish to apply for becoming a mentor for Google Code-In 2019. I have done several projects in Deep Learning. I feel it will be an amazing opportunity for me to become a part of the Google Code-In event as mentor on behalf of TensorFlow, the framework that has helped me build some cool projects. I am providing links to my LinkedIn and GitHub profile:

LinkedIn
GitHub

P.S.: Since I am not sure about how to approach TensorFlow on becoming a mentor for Google Code-In, I am creating an issue for it. Kindly let me know if there is some other way that I should follow for becoming a mentor.

Where is staging package in tf2? Originally from tensorflow.contrib.staging import StagingArea

Community repo

The only issues you should file here concern the community's documentation or processes.

All other bugs should be filed on the appropriate repo, questions directed to the various email groups, or if it's a "how-to" question, StackOverflow.

Hi there,
I just see the list about tf.contrib package, and see that staging is deleted since redundant , i want to ask that which function or package could replace this?

Tracking Issue for PPC64LE builds

Tracking Issue for PPC64LE builds

Following the new process in https://github.com/tensorflow/community/blob/master/sigs/build/community-builds.md
creating this issue to track the existing CPU and GPU community supported builds for PPC64LE.

Test Plan:
CPU and GPU Builds for each commit triggered by GITscm polling.
Nightly build and unit test runs for CPU and GPU Builds.
Issues opened for failing test.

Documentation and examples:
Artifacts published as a results of these builds can be downloaded from the Jenkins server jobs for nightly artifacts and stable release artifacts.

We are working on uploading docker images to the ibmcom/tensorflow namespace.

Support plan:
@wdirons - Primary contact for GPU build/test failures
@sandipmgiri - Primary contact for CPU build/test failures
@tedhtchang - Primary contact for build failures that are environment related.
@jayfurmanek
@JonTriebenbach

Current builds:
https://powerci.osuosl.org/user/wdirons/my-views/view/TensorFlow%20Jobs/

Question: RNN layer vs. cell

See #15 "Unifying RNN"

Why do both RNN cells and layers exist, instead of having just RNN layers (containing the cell functionality)?

KeyError while loading image data.

Error:
File "DataCollection.py", line 38, in caption_image
return "Image (CC BY 2.0) " + ' - '.join(attributions[str(image_rel)].split(' - ')[:-1])
KeyError: 'sunflowers\6166888942_7058198713_m.jpg'

Function:
def caption_image(image_path):
image_rel = pathlib.Path(image_path).relative_to(data_root)
return "Image (CC BY 2.0) " + ' - '.join(attributions[str(image_rel)].split(' - ')[:-1])

It gives a KeyError while returning the parameter.
And every time it gives a KeyError with different image name and path.

Any idea on this?

Embedding and Partitioned Variable is finished in tf2.0?

Hi, guys:
the ref: https://github.com/tensorflow/community/blob/master/rfcs/20190116-embedding-partitioned-variable.md
No api in tf2.0 about the embedding ?

Becoming a mentor for Google Code-in'19

I wanted to become a mentor for Tensorflow for Google Code-in 2019. However, there seems to be no certain way to approach the organization, hence, why I am creating an issue here.
I have a great interest in Machine Learning and specialize in Computer Vision. Given the opportunity, I would be more than happy to submit my CV for review.

If there is a proper path to approaching the organization, kindly inform me and I shall go there.

Training in tf1.8 (Cuda version 9 )

(tf1.8) C:\tensorflow1\models\research\object_detection>python train.py --logtostderr --train_dir=training/ --pipeline_config_path=training/faster_rcnn_inception_v2_pets.config
WARNING:tensorflow:From C:\Users\ISEE\Anaconda3\envs\tf1.8\lib\site-packages\tensorflow\python\platform\app.py:126: main (from main) is deprecated and will be removed in a future version.
Instructions for updating:
Use object_detection/model_main.py.
WARNING:tensorflow:From C:\tensorflow1\models\research\object_detection\legacy\trainer.py:266: create_global_step (from tensorflow.contrib.framework.python.ops.variables) is deprecated and will be removed in a future version.
Instructions for updating:
Please switch to tf.train.create_global_step
WARNING:tensorflow:num_readers has been reduced to 1 to match input file shards.
Traceback (most recent call last):
File "train.py", line 184, in
tf.app.run()
File "C:\Users\ISEE\Anaconda3\envs\tf1.8\lib\site-packages\tensorflow\python\platform\app.py", line 126, in run
_sys.exit(main(argv))
File "C:\Users\ISEE\Anaconda3\envs\tf1.8\lib\site-packages\tensorflow\python\util\deprecation.py", line 250, in new_func
return func(*args, **kwargs)
File "train.py", line 180, in main
graph_hook_fn=graph_rewriter_fn)
File "C:\tensorflow1\models\research\object_detection\legacy\trainer.py", line 291, in train
clones = model_deploy.create_clones(deploy_config, model_fn, [input_queue])
File "C:\tensorflow1\models\research\slim\deployment\model_deploy.py", line 193, in create_clones
outputs = model_fn(*args, **kwargs)
File "C:\tensorflow1\models\research\object_detection\legacy\trainer.py", line 204, in _create_losses
prediction_dict = detection_model.predict(images, true_image_shapes)
File "C:\tensorflow1\models\research\object_detection\meta_architectures\faster_rcnn_meta_arch.py", line 821, in predict
prediction_dict = self._predict_first_stage(preprocessed_inputs)
File "C:\tensorflow1\models\research\object_detection\meta_architectures\faster_rcnn_meta_arch.py", line 872, in _predict_first_stage
image_shape) = self._extract_rpn_feature_maps(preprocessed_inputs)
File "C:\tensorflow1\models\research\object_detection\meta_architectures\faster_rcnn_meta_arch.py", line 1250, in _extract_rpn_feature_maps
feature_map_shape[2])]))
File "C:\tensorflow1\models\research\object_detection\core\anchor_generator.py", line 103, in generate
anchors_list = self._generate(feature_map_shape_list, **params)
File "C:\tensorflow1\models\research\object_detection\anchor_generators\grid_anchor_generator.py", line 111, in _generate
with tf.init_scope():
AttributeError: module 'tensorflow' has no attribute 'init_scope'

This is the error when I run
python train.py --logtostderr --train_dir=training/ --pipeline_config_path=training/faster_rcnn_inception_v2_pets.config

Tensorflow Lite Quant: Error when using a TF Lite Quant model to classify 2 classes. Acurracy low of 1%

I used this notebook to train a network:
https://colab.research.google.com/github/tensorflow/examples/blob/master/community/en/flowers_tf_lite.ipynb

I basically have 2 classes to train:
Class 1 and Class 2;

At the end of the training, I converted the model to a quantized model using the code:
https://www.tensorflow.org/lite/performance/post_training_quantization

import tensorflow as tf
converter = tf.lite.TFLiteConverter.from_saved_model (saved_model_dir)
converter.optimizations = [tf.lite.Optimize.OPTIMIZE_FOR_SIZE]
tflite_quant_model = converter.convert ()

When passing the converted quantized model to the Android example. He just doesn't recognize my classes when it comes to rating and making a rating below 1%.

The FLOAT model, without optimization, worked well.

Could anyone tell me what's going on?

Repository that I got the Android code from:
https://github.com/tensorflow/examples/tree/master/lite/examples/image_classification/android

Even after redoing the process several times, it is still a problem.

Model TF Lite Quantized:

Modelo TF Lite Float:

TypeError: The added layer must be an instance of class Layer. Found: <keras.layers.core.Dropout object

Getting error in Tensor Flow while running it in Google colab. I am getting this: ====>

TypeError: The added layer must be an instance of class Layer. Found: <keras.layers.core.Dropout object

THE Code is :============>
`
import numpy as np
import tensorflow as tf
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers.core import Dense, Dropout
from tensorflow.keras.layers.recurrent import LSTM

mnist = tf.keras.datasets.mnist
(x_train ,y_train),(x_test,y_test) = mnist.load_data()

model = tf.keras.Sequential();
model.add(Dropout(0.2))

model.add(LSTM(128, activation='relu'))
model.add(Dropout(0.1))

model.add(Dense(32, activation='relu'))
model.add(Dropout(0.2))

model.add(Dense(10, activation='softmax'))`

MY versions are :--------
TensorFLow ~~ 1.14
Keras ~~ 2.24

Since the code which I am using is quite correct(I think) the problem must be with TensorFlow or Keras....

The Docker host machine only requires Nvidia Drivers

Currently, the Docker install GPU support instructions here:
https://www.tensorflow.org/install/install_linux#gpu_support

Indicates it requires the full CUDA toolkit, cuDNN, and Nvidia drivers as outlined in:
https://www.tensorflow.org/install/install_linux#NVIDIARequirements

The document should be changed to indicate the host machine (running the container) requires only Nvidia drivers. See:
https://github.com/nvidia/nvidia-docker/wiki/CUDA#requirements

The container itself requires only the CUDA toolkit, cuDNN and Docker file can be found here:
https://hub.docker.com/r/nvidia/cuda/

Where is stop_if_no_increase_hook gone?

I use tensorflow.contrib.estimator.stop_if_no_increase_hook in my code to perform an early stopping, which works smoothly before. But today it throws out an error:
The TensorFlow contrib module will not be included in TensorFlow 2.0.
For more information, please see:

• https://github.com/tensorflow/community/blob/master/rfcs/20180907-contrib-sunset.md
• https://github.com/tensorflow/addons
• https://github.com/tensorflow/io (for I/O related ops)
  If you depend on functionality not listed there, please file an issue.

Traceback (most recent call last):
File ".\main.py", line 215, in
hook = tf.contrib.estimator.stop_if_no_increase_hook(
AttributeError: module 'tensorflow.contrib.estimator' has no attribute 'stop_if_no_increase_hook'

I noticed that estimator has been moved to tensorflow/estimator. But it doesn't support this method now. Would you tell me where to find it or other workaround to perform early stoppoing?

Better Tensorboard Logging

Maybe that issue is only personal due to the nature of my problem.
I am training (using keras) in triple loops and I have the inner loop executed just 5 times. Even if I said the Tensorboard to log data every 100+ loops, it will log the data of each first time of the triple loop.

Full explanation:
I have 400 simulation experiments that have 1000 timestamps each. The training must reset its states each time that the simulation finishes and then the outer loops. Training happens mainly in inner loops so I have to set "inner loops - simulation number - external loops".
That will result in a really big training epochs, usually 5(inner) x 400(simulations) x 1000(outer epochs) = 2.000.000. The data logging is becoming really difficult because I would just like to have a log every 100 simulations? maybe even less but because the Tensorboard must note each time that the inner loop is called, I end up with 400.000 datalogs (each of 2-3MB, add up to 1TB per training!)
A work around (but messy) is to have a flag that I called every 100 simulation to enable the Tensorboard for that instant.

Proposed solution:
A frequency variable (at least in keras) that will actually count the number of epochs by counting the times that the callback is used. I will assume that this problem also imply to bigger networks that have hundreds inner loops but for them the problem can be solved by increasing the histogram_freq but my proposition is for a more general use.

Discord Discourse

For a so large community we have really a very low traffic Dev mailing lists so I suppose something It Is not working in the community comunication infra.

I propose with a rationale quite to LLVM community to setup ad adopt an unique Discord and Discourse for Dev and the SIGs.

Adopt contributor covenant

Can we adopt the contributor covenant in Tensorflow?

Becoming a mentor for Google Code-In 2019

I wish to become a mentor for Google Code-in 2019. I have done a number of ML projects and I've been a Google Code-In Finalist in 2018. My main areas of expertise are Computer Vision. I also do digital design. I'd be happy to send in a resume and more details as well, please let me know if there's an official email or form.

NOTE: As I am not sure how to approach TensorFlow on becoming a mentor for Google Code-in, I am creating an issue for it. Kindly let me know if there is some other way that I should follow for applying.

How do i connect with GSOC Mentors?

Community repo

HI, I am Ayushman Kumar, a GSoC 2020 aspirant. I am really struggling to connect myself with any mentor. I have joined the Gitter channel, mailed in the given email address but i did not get any response. If someone can help me connect with the mentor, i will be helped a lot.
Thank you.

Import Error: cannot import name 'model_utils'

import tensorflow as tf
tf.version( got 1.13.1)
import tensorflow_probability as tfp

got error:

File "c:\program files\python36\lib\site-packages\tensorflow_estimator\python\estimator\model_fn.py", line 33, in
from tensorflow.python.saved_model import model_utils as export_utils

ImportError: cannot import name 'model_utils'

basically, there is no model_utils subfolder or file under tensorflow/python/saved_model

thanks.

typo in section Functions that create state

Small typo in the second sentence of section Functions that create state
This is makes it clear that the code will have the same semantics once wrapped.
should read:
This makes it clear that the code will have the same semantics once wrapped.

[Translation] Translation for Korean

We've translated this part of guidelines to Korean on our repository with github pages.

This is URL of Korean repository : https://github.com/BangSeongJin/translation_to_korean.git

How to get connected?

I am a gsoc 2020 aspirant and want to work on introducing a updated tokenizer and clustering algorithm module for tensorflow.
I mailed but got no response even on public platform.
Please help me.

TensorFlow C API tutorial

There is a tutorial for TF in python, a smaller guide for the C++ API, but there is absolutely nothing for the C API.

Is there any help for starting to use TF in C?

Misleading BidirectionalRNN

For bidirectional rnns, the TF 1.x API is tf.nn.bidirectional_dynamic_rnn(cell_fw, cell_bw), while the TF 2.x API is keras.layers.Bidirectional(tf.keras.layers.RNN(cell, unroll=True)).

My point is, keras.layers.Bidirectional should accept two RNNCell instances, one forward and one backward. With current interface, keras.layers.Bidirectional implicitly create a forward and a backward layer, making things out of control.

I think the better way is to expose cell_fw and cell_bw, which is more consistent with the original TF 1.x API, so that users know they are using two different RNN cells for forward and backward purpose. And users can even share forward & backward RNN cell if they really wish. This interface design also enables more complicated use cases, e.g.: using a 2-layer forward LSTM with hidden size 1024 each layer & a 3-layer backward GRU with hidden size 512 each layer.

For simplicity & compatibility concern, the cell_fw and cell_bw can be designated as key word arguments, making sure i) higher order users know what they are doing and ii) normal users are not influenced.

Switching default [float | int] types in Tensorflow 2.0

Hello everyone,

Not be able to switch between default types has been an issue for such projects like GPflow since the beginning of the TensorFlow. I guess a way of changing default types in TensorFlow will make a lot of things easier.

That question was raised before, e.g. tensorflow/tensorflow#9781, where usually maintainers argued that GPU is much faster on float32, default type cannot (should not) be changed because of backwards compatibility reasons. The thing is that precision is very important for such operations like cholesky, solvers and etc. If project uses only such operations, you have to specify type everywhere, it gets even worse when you start using other frameworks or small libraries which follow standard type settings and sometimes they become useless. The policy of "changing types locally will solve your problems" becomes cumbersome.

It would be great to have methods tf.set_default_float and tf.set_default_int in TensorFlow 2.0 and I believe that such a small change will make TensorFlow a bit more appealing.

ERROR Importing TensorFlow

C:\Users\fpython>python
Python 3.6.5 |Anaconda, Inc.| (default, Mar 29 2018, 13:32:41) [MSC v.1900 64 bit (AMD64)] on win32
Type "help", "copyright", "credits" or "license" for more information.

import tensorflow as tf
Traceback (most recent call last):
File "C:\Users\fpython\Anaconda3\lib\site-packages\tensorflow\python\pywrap_tensorflow.py", line 58, in
from tensorflow.python.pywrap_tensorflow_internal import *
File "C:\Users\fpython\Anaconda3\lib\site-packages\tensorflow\python\pywrap_tensorflow_internal.py", line 28, in
_pywrap_tensorflow_internal = swig_import_helper()
File "C:\Users\fpython\Anaconda3\lib\site-packages\tensorflow\python\pywrap_tensorflow_internal.py", line 24, in swig_import_helper
_mod = imp.load_module('_pywrap_tensorflow_internal', fp, pathname, description)
File "C:\Users\fpython\Anaconda3\lib\imp.py", line 243, in load_module
return load_dynamic(name, filename, file)
File "C:\Users\fpython\Anaconda3\lib\imp.py", line 343, in load_dynamic
return _load(spec)
ImportError: DLL load failed: Uma rotina de inicialização da biblioteca de vínculo dinâmico (DLL) falhou.

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "", line 1, in
File "C:\Users\fpython\Anaconda3\lib\site-packages\tensorflow_init_.py", line 28, in
from tensorflow.python import pywrap_tensorflow # pylint: disable=unused-import
File "C:\Users\fpython\Anaconda3\lib\site-packages\tensorflow\python_init_.py", line 49, in
from tensorflow.python import pywrap_tensorflow
File "C:\Users\fpython\Anaconda3\lib\site-packages\tensorflow\python\pywrap_tensorflow.py", line 74, in
raise ImportError(msg)
ImportError: Traceback (most recent call last):
File "C:\Users\fpython\Anaconda3\lib\site-packages\tensorflow\python\pywrap_tensorflow.py", line 58, in
from tensorflow.python.pywrap_tensorflow_internal import *
File "C:\Users\fpython\Anaconda3\lib\site-packages\tensorflow\python\pywrap_tensorflow_internal.py", line 28, in
_pywrap_tensorflow_internal = swig_import_helper()
File "C:\Users\fpython\Anaconda3\lib\site-packages\tensorflow\python\pywrap_tensorflow_internal.py", line 24, in swig_import_helper
_mod = imp.load_module('_pywrap_tensorflow_internal', fp, pathname, description)
File "C:\Users\fpython\Anaconda3\lib\imp.py", line 243, in load_module
return load_dynamic(name, filename, file)
File "C:\Users\fpython\Anaconda3\lib\imp.py", line 343, in load_dynamic
return _load(spec)
ImportError: DLL load failed: Uma rotina de inicialização da biblioteca de vínculo dinâmico (DLL) falhou.

Failed to load the native TensorFlow runtime.

See https://www.tensorflow.org/install/errors

for some common reasons and solutions. Include the entire stack trace
above this error message when asking for help.

Could someone please help me solving this problem?

What about migrating the @tensorflow.org mailing list to GitHub?

Personally, I think mailing list is pretty efficient and very good at managing discussion scope and people who involved.

But, the tensorflow.org mailing list is not so well managed - I am not blame the people in charge, but the infrastructure system is sort not catching up the developing style nowadays. The fact is, people are more likely to use infrastructure like GitHub, which we can cross reference many things/code and integrate tools together, I guess that's one of the reasons that the code is served here. And, we have observed so many issues in this repo asking about tensorflow bugs some what, which, I believe has bring much effort for the maintainers :(

Yes, there is plenty things to handle as everything is double-bladed (两面性). I am only to raise this issue, please feel free to close anytime :)

Where is tf.contrib.training.HParams ?

I am using AutoAugment util from tensorflow object api for data augmentation.
That util use tf.contrib.training.HParams and I want to use it in tf 2.0 rc.
Can't find where that function is hidden now, any help?

Tensorflow js with Node js?

Hello,

Do you also need to install nodejs in order to use tensorflow js?

Or is it only enough to use tensorflow js?

Thank you!

SIG Add-ons: Complete the move from contrib to addons as specified from RFC.

TF 2.0 might have deprecated tf.contrib, but much of its functionality is widely-used and important to the TensorFlow community. Addons is a repository of those bleeding edge contributions that conform to well-established API patterns, but implement new functionality not available in core TensorFlow.

RFC: Move from tf.contrib to tensorflow/addons.

For more information on Addons, check out their GitHub repo, their community charter, and their mailing group.

ImportError: cannot import name 'check_scalar'

ImportError Traceback (most recent call last)
in ()

----> 3 import sklearn
4 from sklearn.cluster import KMeans as km
5 from sklearn.preprocessing import StandardScaler

./anaconda3/envs/py35/lib/python3.5/site-packages/sklearn/init.py in ()
74 else:
75 from . import __check_build
---> 76 from .base import clone
77 from .utils._show_versions import show_versions
78

./anaconda3/envs/py35/lib/python3.5/site-packages/sklearn/base.py in ()
14
15 from . import version
---> 16 from .utils import _IS_32BIT
17
18 _DEFAULT_TAGS = {

./anaconda3/envs/py35/lib/python3.5/site-packages/sklearn/utils/init.py in ()
18 from ..exceptions import DataConversionWarning
19 from .deprecation import deprecated
---> 20 from .validation import (as_float_array,
21 assert_all_finite,
22 check_random_state, column_or_1d, check_array,

ImportError: cannot import name 'check_scalar'

Broken link in Tensorflow 2.0 overview

The Tensorflow 2.0 overview in the section "General", subsection "Should I start learning TensorFlow 1.x or TF 2.0?", links to https://www.deeplearning.ai/tensorflow-from-basics-to-mastery/ under the anchor "DeepLearning.ai Course". However, this link results in Error 404 "No Results Found". Is it the same as this course on coursera?

Scope of MLIR

Hi,
Is MLIR going to be focusing on just Tensorflow or open to enabling projects like for example, Halide or tensor-contractions or graphics (DX12 shaders) etc

@tatianashp

Ecosystem Issue/PR Grooming Sprints and public roadmap

I receive many mails of repeated Nagging assignee without maintainers activity.
How do you are handling Nagging Assignees?
Can you add an automatic label to have a quick internal overview of these issues?
Continuous ping to inactive maintainers doesn't give a good feeling about the library.

I have trained a .pb model which is in tf1.2 and I want to use it in the environment of tf2.0, but I fail to use it successfully. Anyone knows how to solve the proble

A few links are broken

Links in the following sentences are broken, please consider fixing them:

Pix2Pix (Example Colab)

NMT Model (Example Colab)

Dcgan (Example Colab)

... take a look at tf.Module. (Variables, Custom Layers) ...

... For further information, see our tutorials here and

... and running a saved model in TensorFlow 2.0.

TF 2.0 How to set a different learning rate for each layer?

Hi, I would like to ask a question about how to set different learning rates for several layers. In my implementation, I would like to set one learning rate for some layers while the other learning rate for some other layers. Is there some example codes? Thanks in advance.

Proposed Keras RNN interface does not provide `sequence_length`

I am not sure if I am writing to a correct place, so please point me to a better place if this is not it.

According to the https://github.com/tensorflow/community/blob/master/rfcs/20180920-unify-rnn-interface.md RFC the TF RNN should become obsolete. However, it has a major important functionality -- the sequence_length parameter allowing to process sequences of different length in a single batch. For NLP, this is quite crucial. However, as far as I can see, the Keras RNN does not offer such functionality. Are there any plans to add it?

As it stands, many of our NLP models will not be expressable in TF 2.0 APIs.

Will tensorflow 2.0 directly support sequence-to-sequence RNN models?

I saw that tf.contrib.seq2seq has been moved to tensorflow addons, so is that the recommended way to use seq2seq rnn models? Or maybe tensorflow 2.0 will directly support it?

Using four spaces for indentation in the Python docs? As a sign of good will towards the open source community.

This issue is absolutely not a debate about if four spaces is better than two spaces for indentation in python. Please avoid comments mentioning that.

This is more about the image that the tensorflow project gives to contributors and the open source community in general.

Currently this image is more "We are making this cool tool at Google, you can use it for free, and help us make it better".

We should try to move towards "Tensorflow is community driven, it's open source first. Google is heavily involved in it, so it's expected that the dev workflows of Googlers goes first. "

I understand that moving to four spaces per indentation for Python in the codebase is hard because of the git blame being overwritten and maybe other tools at Google. Also huge git conflicts in perspective.

But for the documentation it's different. First it's seen by a lot more people, so the impact is much much bigger for the image of TF. Some beginners start Python by learning Tensorflow. Let's give them Python practices recommended by pep8, the open source standard, not Google. There is less activity per line of code, so less chances of conflicts. And the git blame is not critical in the documentation.
I believe that changing the indentation in the docs would not be that hard technically speaking. Maybe time consuming at worst.

Let's make Tensorflow feel open-source first!

@dynamicwebpaige do you think I should do a pull request with a RFC? If so, would you endorse it?