repository
stringclasses
156 values
issue title
stringlengths
1
1.01k
labels
stringclasses
8 values
body
stringlengths
1
270k
tensorflowtensorflow
error building with cuda after bazel server die
Bug
this be effectively reopen 4105 as this be another way to reproduce the issue that I be see repeatedly the fix for 4105 be to make sure after run configure clean fetch be run I see this issue almost every day so I m not sure if bazel flush some cache or the daemon be die but the issue reliably reproduce by k...
tensorflowtensorflow
documentation on how reusable variable sync between cpu and gpu
Bug
I haven t be able to find anything online that document how the sync of reusable variable happen between cpu and gpu specifically I m interested in if a cpu pin variable be use multiple time in a gpu computation without be update do it sync only once I e use a dirty flag or will it try sync again each time...
tensorflowtensorflow
code snippet in the tensorflow for poet codelab be confuse
Bug
base on this stack overflow question in section 5 4 the first python command have a lead bash python tensorflow example image retrain retrain py bottleneck dir tf file bottleneck how many training step 500 model dir tf file inception output graph tf file retra...
tensorflowtensorflow
tensorflow contrib learn tensorflowestimator fail under sklearn base clone
Bug
tensorflow contrib learn tensorflowestimator fail under sklearn base clone this prevent the possibility to perform cv use sklearn cross validation cross val score over this estimator the clone fail due to the fact the tensorflowestimator get param return a key name param that be not part of the constructor...
tensorflowtensorflow
wrong example script in the doc for preprocesse datum
Bug
I be read the doc for preprocessing preprocesse which be a small paragraph link to the cifar 10 network as an example however that script do not perform any preprocessing do we have a well example illustrate preprocesse step like datum normalization distort image etc
tensorflowtensorflow
tensorboard should update browser location on tab change
Bug
image right now change tab do not change the location which mean browser refresh etc lose the current tab
tensorflowtensorflow
can t configure due to undeclared package inside if android if io
Bug
I be have trouble configure the late master branch dbe7ee0dfa9e5ab26284522379f2747510fc267b when I run configure I get error tensorflow tensorflow contrib session bundle build 107 1 no such target tensorflow core android lib lite target android lib lite not declare in package t...
tensorflowtensorflow
tf image rot90 return none if the argument be not a python integer
Bug
what related github issue or stackoverflow thread have you find by search the web for your problem point out in this stack overflow question environment info 1 the commit hash git rev parse head bf5b2f0185d4b61329a0bf7a56827661ef6526a1 if possible provide a minimal reproducible exampl...
tensorflowtensorflow
minor documentation code mismatch in optimizer
Bug
the optimizer class have the follow method l388 python def valid dtype self valid type for loss variable and gradient default to float32 subclass should override to allow other type return valid type for loss variable and gradient return set d...
tensorflowtensorflow
incorrect documentation for dynamic rnn
Bug
the help for tf nn dynamic rnn state that the optional initial state kwarg if cell state size be an integer must be a tensor of appropriate type and shape batch size x cell state size this be incorrect the shape must be actually be batch size cell state size
tensorflowtensorflow
seg fault when compute gradient of 3d convolution filter with 1 1 1 kernel
Bug
minimal example to reproduce the bug python import numpy as np import tensorflow as tf sess tf session with sess as default input batch height width depth input channel x shape 3 85 65 83 8 filter kernel height kernel width depth input channel out...
tensorflowtensorflow
reduce max and maximum give different result for negative infinity
Bug
use tf maximum with negative inf input as follow tf maximum math inf math inf eval give the expect result inf however tf reduce max on the same input tf reduce max math inf math inf eval give 3 40282e 38 which be the min float32 for positive infinit...
tensorflowtensorflow
doc for inception model
Bug
tensorflow 0 10 0rc0 cpu on linux this be about the inception model and example that be include with tensorflow I be use this version of the inception model I m not sure whether to ask these question here or on stackoverflow but there be some related question there that haven t be answer and this be al...
tensorflowtensorflow
document how to add new customized file system
Bug
add customized file system be support but not well document tf inc python c import tensorflow as tf print tf sysconfig get include g std c 11 share test file system cc o test file system so fpic I tf inc add a section in how tos will help
tensorflowtensorflow
process hang when use tf sessionrun with multiple time the same input
Bug
it seem that if the same input appear multiple time in the input argument of tf sessionrun from c api h then the tf sessionrun call never return this issue can be reproduce by modify c api test cc and replace the line csession setinput feed int32tensor 3 with csession setinput feed int32...
tensorflowtensorflow
possible improvement in documentation
Bug
I m follow the documentation here compile the kernel for the gpu device specifically I find that run the second command in the code block of the section compile the kernel for the gpu device return a linker error complain about the lack of the lcudart library I have solve this issue by add the optio...
tensorflowtensorflow
tf import graph def graph def be invalid at node
Bug
I ve be try to import a frozen graph into a new program and do a simple forward pass but tf import graph def have be throw a valueerror that I really can t make sense of environment info operating system ubuntu 14 04 lt 64 bit instal version of cuda and cudnn none if instal from source provide...
tensorflowtensorflow
improve google indexing for the documentation
Bug
whenever I run a google search on a tensorflow functionality say tf reshape it give I the entire documentation not the specific documentation relate to that functionality currently the way I use be to run a search with ctrl f to find specific documentation relate to what I search for numpy ...
tensorflowtensorflow
refactore gpu installation from source issue ubuntu 16 04
Bug
it look like the gpu installation from source get refactore and be fail to compile on my ubuntu 16 04 cuda 8 0 cudnn 5 0 with gtx 1080 gpu I have the gtx 1080 work with the build from a few day ago but just pull from the head and now can not compile so I be not sure if this be the auto configure for the gpu ...
tensorflowtensorflow
tensorflow hang when enqueuee with session timeout specify
Bug
I be create a simple fifoqueue and enqueuee element to it everything work fine when the session timeout be not specify code as below with timeout comment out time python ex py enqueue enqueue enqueue enqueue enqueue 1 56user 0 24system 0 01 70elapsed 105 cpu everything finish correctly but ...
tensorflowtensorflow
document checkpoint file format
Bug
hi as far I can find the internal structure of checkpoint file be not document on tensorflow org I m try to write binding to tensorflow which can load checkpoint without make use of the python api so I be wonder if that information be available somewhere
tensorflowtensorflow
link to iris monitor py break
Bug
the link to iris monitor py return not find on tensorflow tutorial log and monitor basic log and monitor basic with tf contrib learn should it be redirect to github there
tensorflowtensorflow
l2 normalize break in master
Bug
tf nn l2 normalize tf constant np one 2 2 2 2 0 1 2 crash with the follow error valueerror shape 1 3 must have rank at most 1 the problem be that l2 normalize function convert the dimension list into 0 1 2 double bracket and it stop be support recently it work fine...
tensorflowtensorflow
unclear documentation and behavior for sampler in tensorflow
Bug
for the sampler implement in tensorflow e g tf nn fix unigram candidate sampler the behavior be not well define in the document for instance I would expect the label specify in true class will be exclude from the sample pool and the sampling will be conduct for each batch but accord to my experiment nei...
tensorflowtensorflow
unexpected name in namedoutputs tuple when use output collection with some layer
Bug
we be get unexpected name they lack outer name scope in namedoutputs tuple add to collection use tf contrib layer output collection the follow code demonstrate the issue import tensorflow as tf from tensorflow contrib import slim with tf name scope train with slim arg scope slim fu...
tensorflowtensorflow
tf round np round
Bug
look like tensorflow and numpy use different tie breaking rule tf round 0 5 1 np round 0 5 0 both be sound so actually there s no bug however since tf and np be tightly couple be worth consider to uniform they or at least to explictly mention it in the doc especially because np be often use...
tensorflowtensorflow
function defun can t be apply to tf variable
Bug
I need to specify a custom gradient for my custom op function I ve write the following example import numpy as np import tensorflow as tf from tensorflow python framework import function def custom op grad op grad return grad function defun x tf float32 python grad func custom op grad def ...
tensorflowtensorflow
error can t copy tensorflow model embed gen word2vec py doesn t exist
Bug
I be instal tf from source like I always do but I m have the follow issue this time when run bazel bin tensorflow tool pip package build pip package tmp tensorflow pkg I get bash python tensorflow bazel bin tensorflow tool pip package build pip package tmp tensorflow pkg sit aug 6 16 07 12 edt 2016 ...
tensorflowtensorflow
0 10 0rc0 contrib distribution crash when sample n be scalar
Bug
github issue be for bug installation problem feature request for general support from the community see stackoverflow to make bug and feature request more easy to find and organize we close issue that be deem out of scope for github issue and point people to stackoverflow for bug or installation i...
tensorflowtensorflow
problem in implement the gradient in python doc
Bug
refer to the doc here implement the gradient in python first of all the chain rule be completely wrong also the zeroout example be not a very good one imo would be much well to give e g a simple derivative of a polynomial or something just a fyi edit I d like to emphasize that the chain rul...
tensorflowtensorflow
unable to import frozen graph with batchnorm
Bug
error when load the frozen graph with tensorflow contrib layer python layers batch norm valueerror graph def be invalid at node u batchnorm cond assignmovingavg switch input tensor batchnorm move mean 0 can not convert a tensor of type float32 to an input of type float32 ref freeze graph py does...
tensorflowtensorflow
miss doc tf contrib layer embed column
Bug
tutorial refer to tf contrib layer embed column no documentation for this be available at api docs the same applie for tf contrib learn dnnlinearcombinedclassifier use later in that tutorial
tensorflowtensorflow
documentation for add a new op
Bug
this be a feature request currently I find the documentation for add an op quite minimal there be a few important question that remain unanswered how should one use multithreade cpu code in an operation can we use openmp how many thread should an op use it be hard to find this in the implementation ...
tensorflowtensorflow
a link in tensorflow s website be ineffective
Bug
the link in download the tutorial code from tensorflow linear model tutorial fail be the source code of the website not open I didn t find it
tensorflowtensorflow
dnn autoencoder iris py example error
Bug
the example in skflow folder give this error traceback most recent call last file home rmasad test dnn autoencoder iris py line 33 in transform autoencoder fit transform iris data file usr local lib python2 7 dist package tensorflow contrib learn python learn estimator base py l...
tensorflowtensorflow
gradient computation fail concatenation in while loop body
Bug
description when I be try to implement rnn with while loop I try to concatenate output to a matrix this work in forward pass but not in apply gradient also I see that there be discussion 2237 about support recursive nn there s a workaround by transform tree structure to a matrix for example ...
tensorflowtensorflow
tf cond and tf case execute all branch
Bug
when I run python import tensorflow as tf x1 tf variable 1 x2 tf variable 1 cond value tf variable true assign 1 tf assign x1 2 assign 2 tf assign x2 2 with tf control dependency assign 1 result 1 tf identity x1 with tf control dependency assign 2 r...
tensorflowtensorflow
broadcast 0 rank tensor when compute gradient for tf nn relu
Bug
environment info operating system osx macos cpu only version 0 9 0 perhaps this be desire behavior but I would have much appreciate a more descriptive warning at least which would have save much debugging I haven t find a small reproducible case for this but in the code I originally fi...
tensorflowtensorflow
dnn linear combine py be not compatible with python 3
Bug
in python 3 x line 215 of dnn linear combine py master version result in an error line 215 be return sorted set self dnn feature column if self dnn feature column else none the error be builtin typeerror unorderable type sparsecolumnhashe str to reproduce this error in ...
tensorflowtensorflow
documentation of session argument seem outdated
Bug
in readme md a session be create as tf session local at the same time the docstre l888 say no value other than an empty string be support this be clearly an inconsistency
tensorflowtensorflow
float division by zero
Bug
it seem that if you only run 1 training step at a time then you can come to a point where it be too fast traceback most recent call last file main py line 6 in connection start socket 8089 callback handler message processor file mnt d workspace sketchrecognitionwithtensorflow src main...
tensorflowtensorflow
get an 0 8 whl when instal from source from master branch
Bug
I m get an 0 8 whl when instal from source right now current master 6 27 ubuntu 15 1 python3 5 be try it from scratch again check the repo readme md release 0 9 0 cudnn 7 5 be that correct bazel version 0 2 3 output from ls l usr local cuda lib64 cudn aug 1...
tensorflowtensorflow
allow use of gitter I m for tensorflow tensorflow repo
Bug
as far as I could tell there be no place for the community to discuss tensorflow in realtime the recommendation all seem to be about ask specific question in a detailed async fashion it would be nice if there be a gitter I m or irc channel for tensorflow so that the community could interact more easily not eve...
tensorflowtensorflow
deep mnist tutorial would benefit from some diagram of the neural network architecture
Bug
the basic version of this tutorial have a nice diagram that make it really clear what model the code be attempt to reproduce the deep mnist for expert tutorial however doesn t have such an image especially for the multilayer convolutional network that be the heart of the article I have a hard time and...
tensorflowtensorflow
unboundlocalerror when tf newsessionoption fail in session py
Bug
the following fail with unboundlocalerror after session tf session tf configproto log device placement true traceback most recent call last file opt error py line 3 in session tf session tf configproto log device placement true file user yaroslavvb tfimmediate hood tensorflo...
tensorflowtensorflow
unhandled api callback message when use gpu tracing
Bug
use the gpu tracer e g via session run with runoption full trace result in warning like the follow w tensorflow core common runtime gpu gpu tracer cc 513 unhandled api callback for 2 41 w tensorflow core common runtime gpu gpu tracer cc 513 unhandled api callback for 2 41 these be believe to be h...
tensorflowtensorflow
dequeuee immediately after start thread fail
Bug
from a question on stack overflow the follow code fail python import tensorflow as tf import time with tf graph as default filename list datum batch mat format i 1 for I in range 5 filename queue tf train string input producer filename list with tf session as sess...
tensorflowtensorflow
unable to download nightly binary
Bug
all the link in readme such as this one give I http error 404 problem access view nightly job nigntly matrix linux gpu tf build container type gpu tf build be opt opt tf build be pip pip tf build python version python2 label gpu linux lastsuccessfulbuild artifact pip test whl tensorflow 0 8 0 cp27 non...
tensorflowtensorflow
hard code 1 for cuda version break exist macos gpu build with cuda 7 5
Bug
commit break exist gpu build on macos with cuda 7 5 it now fail I tensorflow stream executor cuda cuda gpu executor cc 1078 fail to find libcuda so on this system fail precondition could not dlopen dso libcuda 1 dylib dlerror dlopen libcuda 1 dylib 5 image not find a work around be to create a...
tensorflowtensorflow
fixedlenfeature with dimension of unknown size fail with strange error
Bug
python import tensorflow as tf example tf train example feature tf train feature feature a tf train feature int64 list tf train int64list value 1 serialize example serializetostre feature a tf fixedlenfeature none tf int64 parse tf parse single example seriali...
tensorflowtensorflow
cuda error misalign address on mnist example
Bug
summary what might be cause this error when run python tensorflow model image mnist convolutional py e tensorflow stream executor cuda cuda event cc 49 error polling for event status fail to query event cuda error misalign address environment info operating system linux lounge 4 5 6 200...
tensorflowtensorflow
poor numeric in cpu tf multinomial
Bug
the cpu implementation of tf multinomial seem to have numerical underflow overflow when logit be outside of a certain narrow range the usable logit range on my system be about 88 to 88 which be unacceptably small for my use case where the logit come from an automatically train hide layer for example ...
tensorflowtensorflow
provide variable scope documentation
Bug
this be to track a documentation request from an external user the section on variable scope be pretty helpful and we have be use it to make more complicated net more manageable I think some information could be informal like a blog post on good practice that perhaps googler be encourage to follow in structu...
tensorflowtensorflow
exponentialmovingaverage average duplicate the current scope name
Bug
use tensorflow nightly python import tensorflow as tf with tf name scope scope x tf variable 42 dtype tf float32 ema tf train exponentialmovingaverage decay 0 9 apply op ema apply x average ema average x print average name scope scope variable exponentialm...
tensorflowtensorflow
mention that gpu reduction be nondeterministic in doc
Bug
the problem I be try out the mnist for expert tutorial deep mnist for expert and I have inconsistent result on the gpu what do I mean by inconsistent with the exactly same network parameter and randomness remove read below in the post every time I run the complete train then test process the ...
tensorflowtensorflow
tf learn bug when run dnnclassifi example
Bug
from sklearn import dataset metric iris dataset load iris classifier learn dnnclassifi hide unit 10 20 10 n class 3 classifier fit iris data iris target score metric accuracy score iris target classifier predict iris data print accuracy f score traceback most recent...
tensorflowtensorflow
persistent tensor deletion fail when tensor be move
Bug
tldr handle mover register python sessionhandle instead of string session handle which cause internalerror unable to get element from the feed when garbage collection be trigger on move persistent tensor the solution be to replace self register dead handle handle with self register dead handle ...
tensorflowtensorflow
tf image decode png return wrong value for uint16 image
Bug
environment info operating system ubuntu 14 04 instal version of cuda and cudnn cuda 7 5 cudnn r4 if instal from source provide the commit hash d8eb4bb6470d4cb3d0f67f2111a39fa50f1c28e5 step to reproduce please see the below code after save numpy array as uint16 png image I load it use tf...
tensorflowtensorflow
udacity readme docker command should not remove container
Bug
the udacity readme at suggest start docker like this docker run p 8888 8888 it rm b gcr io tensorflow udacity assignment 0 5 0 the rm cause the container to get remove once you end the session which delete all your progress so far in the class it would be well to not suggest rm so that ...
tensorflowtensorflow
avg max pool3d description have a bug
Bug
in file tensorflow g3doc api doc python function and class shard0 tf nn avg pool3d md tensorflow g3doc api doc python function and class shard4 tf nn max pool3d md original ksize a list of int that have length 5 1 d tensor of length 5 the size of the window for each dimension of ...
tensorflowtensorflow
question word txt link break for word2vec
Bug
the link for the word2vec analogy question text file be borken
tensorflowtensorflow
virtualenv installation failure
Bug
environment info operating system os x 10 10 3 we follow the download and setup doc and try to create a virtualenv environment in the directory tensorflow but it didn t work well virtualenv system site package tensorflow use base prefix application canopy app appdata canopy 1 7 2 3327 mac...
tensorflowtensorflow
confuse embed lookup docstre incorrect use of the word partition
Bug
the wording in the docstring for nn embed lookup be confusing and refer to subset as partition embed lookup a partition of a set x be a set of nonempty subset of x such that every element x in x be in exactly one of these subset here be an example of incorrect usage for instance 13 id be split across 5 p...
tensorflowtensorflow
incorrect error message
Bug
this error message be out of date valueerror can not execute operation use run no default session be register use with default session sess or pass an explicit session to run session sess as far as I can tell there be no long any default session function instead one must call sess...
tensorflowtensorflow
install instruction for os x gpu
Bug
with the wonderful addition of 664 installation installation instruction for os x with gpu support would be very helpful
tensorflowtensorflow
method to check for gpu op support
Bug
so give the basic word2vec example be bind to cpu 514 we can see that tf nn embed lookup doesn t work on gpu therefore op that use embed lookup internally doesn t support gpu either for example nce loss can we have explicit info in the documentation on which operation be currently gpu capab...
tensorflowtensorflow
nan value management in max pool
Bug
hi I face a strange behavior use the max pool op return 3 40282347e 38 when nan be provide a simple example run on cpu be show bellow import tensorflow as tf import numpy as np x tf placeholder tf float32 none 2 1 1 pool tf nn max pool x ksize 1 2 1 1 stride 1 2 1 1 ...
tensorflowtensorflow
set up tensorflow for development break on linux
Bug
tldr instead of ln s bazel bin tensorflow tool pip package build pip package runfiles org tensorflow I have to do ln s bazel bin tensorflow tool pip package build pip package runfile main ubuntu 15 04 bazel 0 2 3 week old tensorflow from head on macos bazel 0 2 1 homebrew...
tensorflowtensorflow
tf pad docstring be miss description of the mode argument
Bug
the docstring give example of use constant reflect and symmetric mode but never describe what they actually be for example what value be use to pad in constant mode the answer be 0 what exactly be the difference between reflect and symmetric I don t know
tensorflowtensorflow
tensorflow segfault on attempt to save a large variable
Bug
the checkpoint format write the value of a variable into a protocol buffer which have a 2 gb limit the saving mechanism do not validate the size of a variable mdash which can be much large than 2 gb mdash before attempt to write it into the protocol buffer the follow code can cause a segfault python with t...
tensorflowtensorflow
update faq
Bug
the faq here still say that multi machine isn t available yet but it be available now
tensorflowtensorflow
pow gradient create inf at 0 0
Bug
hi I have a pipeline where I be first clamp to 0 1 and then do x 2 4 convert to from gamma rgb to linear rgb y tf minimum tf maximum y 0 0 1 0 gamma tf constant 1 0 2 4 tf float32 y tf pow y gamma however this instantly create inf value and nans in the next step se...
tensorflowtensorflow
documentation update virtualenv should not be run from
Bug
hi the documentation suggest that you make a virtualenv essentially by cd mkdir tensorflow virtualenv tensorflow unfortunately for some reason I can not comprehend this make virtualenv try to scan every html file possibly also other in in my case this hit a 22 mb file I have lie around ...
tensorflowtensorflow
nan checker do not check output of optimizer
Bug
hi I notice this while debug fp16 support for the adam optimizer so I think I d write it down here before it be forget I have problem with nan in my pipeline but couldn t really figure out where they come from so I call tf add check numeric op but it turn out that if the adam optimizer create a nan in...
tensorflowtensorflow
protobuf message overflow on try distribute
Bug
I m try to build an rnn on multi machine follow the distribute tensorflow when I use with sv manage session server target as sess it show error attributeerror supervisor object have no attribute manage session so I follow the code of inception with sv prepare or wait for session server tar...
tensorflowtensorflow
word2vec overflow int32 corpus size
Bug
hello I m try to build a big word model and I m get overflow in the corpus size declare here l119 the error be be throw here l191 because negative number be less than positive number I m go to change they to uint64 in my own build if it doesn t cause problem should I make a pr
tensorflowtensorflow
float point exception when compute gradient
Bug
hi tensorflow developer team I just want to first let you know that I really appreciate all the work you be put into this amazing open source system it have be a terrific system for I so far however I have come into a roadblock when use the tf gather function in gradient I be use environment info op...
tensorflowtensorflow
tf nn dynamic rnn fail when batch size be 0
Bug
environment info operating system ubuntu 15 10 instal version of cuda and cudnn rw r r 1 root root 322936 aug 15 2015 usr local cuda lib64 libcudadevrt a lrwxrwxrwx 1 root root 16 aug 15 2015 usr local cuda lib64 libcudart so libcudart so 7 5 lrwxrwxrwx 1 root root ...
tensorflowtensorflow
manual placement on gpu of a custom operator with both cpu and gpu implementation will always run the cpu version
Bug
environment info operating system ubuntu 14 04 64 bit instal version of cuda and cudnn 7 5 and 4 ls l usr local cuda lib64 libcud rw r r 1 root root 322936 aug 15 2015 usr local cuda lib64 libcudadevrt a lrwxrwxrwx 1 root root 16 aug 15 2015 usr local cuda lib64 libcudart s...
tensorflowtensorflow
unclear what parallel iteration do in while scan doc issue possible functional op issue
Bug
it s unclear from both the code and documentation why scan default to 10 parallel iteration doesn t that cause it to sometimes run the function before the accumulate value be available I also poke around and read the documentation for while and that didn t clear thing up thing it look like it might do ru...
tensorflowtensorflow
some test gpu one hang for about 300 second before run
Bug
a lot of test seem to run very slowly like for 5 minute I think it s the one that use the gpu it always hang at the start of the test right after I tensorflow core common runtime gpu gpu device cc 755 create tensorflow device gpu 0 device 0 name geforce gtx 750 ti pci bus i d 0000 01 00...
tensorflowtensorflow
tensorboard issue on mac 0 8 0rc0
Bug
environment info operating system mac os x use virtualenv instal version of cuda and cudnn none please attach the output of ls l path to cuda lib libcud if instal from binary pip package provide 1 which pip package you instal sudo pip install upgrade 2 the output from python...
tensorflowtensorflow
update website with correct cuda info
Bug
I receive this on twitter can you please help update tf website to state tf gpu need nvidia compute cap 3 5 and to install cudnn v4 not 5 thank I m not use cuda myself so I m not sure of the detail here but want to get it log so somebody more knowledgeable can take a look
tensorflowtensorflow
profiling tool for open source tensorflow
Bug
it would be very useful to understand which part of the computational graph be cause computational bottleneck this question have be ask on stack overflow here and here but it seem more appropriate as a feature request as this would allow systematic discussion the tensorflow white paper talk about very ad...
tensorflowtensorflow
tf reverse sequence doesn t work with argument of totally unknown shape
Bug
from stack overflow I be try to reverse my input with array op reverse sequence before send it to dynamic rnn the inference graph can be build with no problem but when build the training graph I get the follow error traceback most recent call last file bin trainer py line...
tensorflowtensorflow
request well documentation on tfrecord and their use
Bug
I be use tfrecord to use the sharde queue machinery for simple problem for which the example can be clone it be easy to use but recently I try to do something more complex train example where there be multiple label I post a question on stackoverflow about the specific issue I have it would be nice to hav...
tensorflowtensorflow
optimizer incompatible with sample miss doc
Bug
hi perhaps tensorflow doc should mention that 5 out of 7 available optimizer will not work with sample loss for now they fail with mysterious message the follow script will fail for momentum adagrad adadelta rmsprop and ftrl also where should I look if I d like to implement my own optimizer for gpu...
tensorflowtensorflow
use tensor as bool in cifar10 example
Bug
hello I have just recompile tf to commit f952246 and get an error on cifar10 example line if grad should be update to if grad be not none due to recent change just minor stuff
tensorflowtensorflow
error run example on gpu
Bug
run bazel bin tensorflow cc tutorial example trainer use gpu with late development code result in I tensorflow stream executor dso loader cc 105 successfully open cuda library libcubla so 7 0 locally I tensorflow stream executor dso loader cc 105 successfully open cuda library libcudnn so 4 0 7 locall...
tensorflowtensorflow
math processing error in web doc
Bug
see tensorflow org version r0 7 tutorial mnist beginner index html training section math processing error appear in multiple place in page use osx safari firefox and chrome screen shot attach screen shoot 2016 03 14 at 3 15 14 pm
tensorflowtensorflow
docbug
Bug
this tutorial will how you how to run the example script on your own image and will explain some of the option you have to help control the training process find at
tensorflowtensorflow
need well doc about the difference between name scope and variable scope in tensorflow and what s the context manager sove
Bug
hi although stack overflow have a question about this but right now the answer be not so clear and the doc also not tell the difference too but I believe understand the difference between name scope and variable scope and context manager be beneficial to program for newbie use tensorflow
tensorflowtensorflow
seq2seq tutorial example not work for python 3 4
Bug
environment info operating system ubuntu 14 04 1 which pip package you instal sudo pip3 install upgrade tmp pip tensorflow whl 2 the output from python c import tensorflow print tensorflow version 0 7 1 build from commit 263d00d271077 with tf unofficial set 1 config...
tensorflowtensorflow
image op do not all accept numpy array as argument
Bug
from stack overflow when I run the line flip image tf image random flip leave right image I get the follow error attributeerror numpy ndarray object have no attribute get shape my tensor image be an ndarray shape batch I m size I m size channel of batch ...
tensorflowtensorflow
c documentation break
Bug
none of the link on this page work
tensorflowtensorflow
document trick with slash in scope name
Bug
the documentation for tf name scope be rather ambiguous on what happen if the scope name end with merely state what happen if it doesn t end with a the fact that the behavior be different specifically that end with a do not result in automatically uniqueifye the scope name be very important and...
tensorflowtensorflow
test in tensorflow core distribute runtime sometimes time out
Bug
some of the test in this directory in particular master test and rpc grpc session test spawn subprocesse contain tensorflow server that listen on unused port there be a potential toctou bug in this code because the process be 1 pick unused port in the parent 2 fork process and instruct they to bin...
tensorflowtensorflow
fix build bug come from 8041c54
Bug
for bug issue please fill in the follow the more information you provide the more likely we can help you environment info operating system osx if instal from source provide the commit hash 8041c54 tensorflow contrib linear optimizer build linkopt wl bsymbolic l...
tensorflowtensorflow
sigmoid cross entropy doesn t verify shape
Bug
hello I have find a misleading behaviour in sigmoid cross entropy with logit function if I pass a tensor of shape n 1 as logit and a tensor of shape n as target it output a tensor of n n which be obviously different from tensor n if both tensor have the same shape n althou...
tensorflowtensorflow
download link of pdf version of the documentation get startre tutorial etc
Bug
try a lot of search but I be still not able to find a pdf version of the tensorflow doc kindly provide the same