repository
stringclasses
156 values
issue title
stringlengths
1
1.01k
labels
stringclasses
8 values
body
stringlengths
1
270k
tensorflowtensorflow
syncreplicasoptimizer race condition strange behavior
Bug
it seem there be a strange race condition in syncreplicasoptimizer lead to strange behaviour I include below an example code to reproduce what seem to be a bug hopefully in my code as well as the command to reproduce it pretty much the same code as in mnist replica py I be try to implement synchronize...
tensorflowtensorflow
wrong command for instal protobuf at tutorial
Bug
the command for instal protobuf at tutorial be wrong bug they be use for instal tensorflow via pip but obviously what should be give be way to install google protobuf via pip please fix it here s the url instal with native pip
tensorflowtensorflow
add python graph transform example
Bug
the doc for the graph transform tool write your own transform talk about write your own transform but way to do so be in c it would be cool if there be an example of c hook writing transform in python as well for example here be one python transform by mrry
tensorflowtensorflow
runtimeerror no c shape function register for standard op nearestneighbor
Bug
for bug or installation issue please provide the follow information the more information you provide the more easily we will be able to offer help and advice what related github issue or stackoverflow thread have you find by search the web for your problem github issue 7524 stackoverflow q...
tensorflowtensorflow
can there any example about multi gpu or multi machine training
Bug
the news say 64 gpu can acheive 56 speed how can we build for training test can there any example
tensorflowtensorflow
erroneous number of channel in the guide to tf layer tutorial
Bug
the tutorial a guide to tf layer build a convolutional neural network seem to have the wrong number of channel in the tensor dimension give at the end of the paragraph convolutional layer 1 our output tensor produce by conv2d have a shape of batch size 28 28 1 the same width and height...
tensorflowtensorflow
test issue
Bug
note only file github issue for bug and feature request all other topic will be close 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 ...
tensorflowtensorflow
internal compiler error use 1 0 0 devel docker image
Bug
first pull the docker image docker pull tensorflow tensorflow 1 0 0 devel docker run it tensorflow tensorflow 1 0 0 devel bin bash then in the docker session cd tensorflow bazel build c opt tensorflow libtensorflow c so this succeed for a while but eventually outpu...
tensorflowtensorflow
break change to api
Bug
tf nn sample softmax loss and tf nn nce loss have both change their api such that you need to switch the input label to label input parameter you might want to put a note here under break change incase you want to check here be the api of the two version of tf although I do find ...
tensorflowtensorflow
primary install page return 404
Bug
404 that s an error the requested url be not find on this server that s all we know
tensorflowtensorflow
tf r1 0 stride slice documentation
Bug
in the doc the follow sentence seem incomplete incoherent if the ith bit of ellipsis mask be set to something as many unspecified dimension as need will be insert between other dimension only one non zero bit be allow in ellipsis mask if the ith bit of new axis mask be one then a begi...
tensorflowtensorflow
typo in validate your installation document
Bug
validateyourinstallation partway down if you be new to tensorflow see get start with tensorflow getting start with tensorflow note the extraneous square bracket look like a markdown typo interestingly the getting start with link still work which do not survive my copy paste into the is...
tensorflowtensorflow
segfault in runtime executor due to variable overflow
Bug
I m run tensorflow 1 0 and I m encounter a segfault in tensorflow with the follow output f tensorflow core common runtime executor cc 484 check fail e src output 32768 38774 vs 32768 abort core dump my program load a fairly large sparse matrix 1879549 sample 556926 feat...
tensorflowtensorflow
update performance guide compiler recommendation for macos user
Bug
hi as per the performance guide I try to build tf from source use gcc version 4 8 3 on a macbook I try gcc 4 9 and 6 3 homebrew version I get the follow error same error for 6 3 gcc 4 9 error unrecognized command line option fcolor diagnostic gcc 4 9 error unrecognized c...
tensorflowtensorflow
after type run in tensorflow debugger the terminal reappear and stuck
Bug
my program can run and show some trae result but for a classification question the accuracy of training set and validation set both keep at around 0 5 for 40 epoch so I want to use tensorflow debugger to watch what the variable be when I type run and push enter it jump out of the debugger and the ter...
tensorflowtensorflow
tensorflow for poet section 5 label image py on site be different than curl version
Bug
link 4 on the site the label image py script use relative path for file while the version the tutorial ask user to curl use absolute path for file if the user be knowledgeable enough to prefer copy the script on the website vs curl ing it from then the user might run into an issue where they first...
tensorflowtensorflow
tensorflow for poet section 5 miss a verb
Bug
in 4 you might warning be miss a verb maybe you might get warning or you might see warning
tensorflowtensorflow
break link to doc in extra
Bug
in the doc on nec loss the target doc of the candidate sample algorithm reference of which the url be be miss and give just a 404 error
tensorflowtensorflow
make the url in the documentation clickable
Bug
it would be nice if the url in the documentation be clickable example image
tensorflowtensorflow
tensorflow sequence to sequence model
Bug
hello everyone I execute the code translate py from sequence to sequence model tutorial of tensorflow for 20 hour and do not get output can anyone tell I how much time will it take to display the output system specification I instal ubuntu 16 04 on vmware with a disk space of 100 gb seqgoo
tensorflowtensorflow
the function layer convolutional py do not follow tf graphkeys convention
Bug
in convolutional py bias and weight variable be create and name kernel and bias however accord to the convention define in tf graphkey md those variable should be name weight and bias and assign the corresponding collection there be some third party functionality which utilize this con...
tensorflowtensorflow
tensorflow example mutilbox detector bazel build error
Bug
I think there be lot of out of date doc in this repo for example tensorflow example mutilbox detector this example show build mutilbox detector use bazel and generate an binary executable file however I just get bunch of error warn config value be not define in any rc file opt error ...
tensorflowtensorflow
slim doesn t play well with the new tf layer v 1 0
Bug
we use tf contrib slim to build nn architecture after upgrade to tensorflow 1 0 the be some trouble when building model with slim for example slim arg scope doesn t work with the tf layer component and when we keep use slim layer we get a runtime error tensorflow python framework error impl inv...
tensorflowtensorflow
api documentation with link to source code
Bug
I think it would be a really great improvement if the api documentation would contain link to the correspond source code file on github similar to python package documentation on readthedocs org e g this would also make it much fast to create pull request that fix error in the documentation
tensorflowtensorflow
can t enforce shape invariant with tensorarray in while loop
Bug
I can t enforce shape invariant in a while loop if one of the input be a tensorarray here s a minimal example import tensorflow as tf def body I ta ta ta write I 1 0 return i 1 ta arr size 10 ta tf tensorarray tf float32 size arr size I tf constant 0 tf int32 ...
tensorflowtensorflow
image problem in tensorflow documentation at tensorflow org error 404
Bug
I m not sure I should put the issue here but when I click issue tracker on tensorflow org it redirect I here I realize that the directory structure for tf documentation have change a lot for tf 1 0 on now there be problem for web page with image like a 404 error be report while load the image
tensorflowtensorflow
doc for offset in extract glimpse don t match implementation
Bug
docs state offset a tensor of type float32 a 2 d integer tensor of shape batch size 2 contain the x y location of the center of each window whereas the implementation state l92 call tensorflow operate with y x as index notice x y vs y x he...
tensorflowtensorflow
how to reproduce the 58x performance improve claim by tf r1 0
Bug
tf r1 0 claim that it can achieve 58x performance improve by 64 gpu for inception v3 be there any guideline or sample code to help we to reproduce the result
tensorflowtensorflow
error message could be improve
Bug
if your try and initialize a dynamic rnn without specify an initial state or dtype you get this enigmatic error message if no initial state be provide dtype must be I don t think that message be very clear something well would be along the line of if no initial state be provide ...
tensorflowtensorflow
the tutorial log and monitor basic with tf contrib learn have error
Bug
when I use the code snippet in the section customize the evaluation metric with metricspec of the tutorial log and monitor basic with tf contrib learn the code snippet be python validation metric accuracy tf contrib learn metric spec metricspec metric fn tf contr...
tensorflowtensorflow
tf nn have no attribute rnn cell in version 1 0 1
Bug
what related github issue or stackoverflow thread have you find by search the web for your problem environment info operating system ubuntu 16 04 instal version of cuda and cudnn please attach the output of ls l path to cuda lib libcud if instal from binary pip package provide...
tensorflowtensorflow
cci good practice
Bug
google be a member of cci what do you think to add cci badge to tf repository
tensorflowtensorflow
add description how to generate op wrapper
Bug
generate instruction of op wrapper be useful for development of go bind
tensorflowtensorflow
tensorboard embed with t sne crash
Bug
when I run t sne on my dataset on small or big one with tensorboard embed the app crash all the time after thereabout 500 iteration I can t re run or stop the operation to change the parameter the iteration number freeze and the other functionality on embed part crash also anyone have the same proble...
tensorflowtensorflow
tfdbg dump root directory do not exist
Bug
I run python m tensorflow python debug example debug mnist debug on my laptop which have window 10 and gtx1070 but after I type first run it return an exception like this traceback most recent call last file d anaconda3 lib runpy py line 184 in run module as main main mod s...
tensorflowtensorflow
3d convolution unnaturally slow on cpu
Bug
3d convolution on cpu seem unnaturally slow I can t use gpu due to memory limit so I m look at cpu execution 2d convolution in tensorflow seem to be well optimize all cpu core be use performance be just few time below gpu with 3d convolution the difference be order of magnitude also I ve...
tensorflowtensorflow
the use of multirnncell
Bug
doc be in I find the line stack lstm tf contrib rnn multirnncell lstm number of layer I need to use multirnncell but I write those line a tf nn rnn cell basiclstmcell 10 3 print i d a 0 I d a 1 its output be 4648063696 4648063696 can multirnncell use the same object...
tensorflowtensorflow
break class hierarchy in sample code in get start tutorial
Bug
the last example in do not work when run it give the follow error message in model feature label mode param 18 modelfnop connect subgraph we build to the 19 appropriate functionality 20 return tf contrib learn estimator model fn modelfnops 21 ...
tensorflowtensorflow
tar unrecognized archive format error when try to unpack flower photo tgz tf tutorial on osx
Bug
this be relate to the how to retrain inception s final layer for new category tutorial I m try to unpack the flower photo tgz after curl it use curl o tar xzf flower photo tgz this be from the image retrain tutorial for tensorflow result from curl total receive ...
tensorflowtensorflow
the update documentation site do not contain set up tensorflow for development
Bug
recently the tensorflow web site be update and some content be re organize since I always forget how to install tf from source such as bazel build then bazel bin I frequently refer the official web site however after the update set up tensorflow for development section have go away w...
tensorflowtensorflow
addition be much slow on non last axis non fuse batch norm with nchw
Bug
I notice this from observe my model train many time slow when use non fused batch norm and the nchw data format when look at the timeline it s dominate by addition and multiplication operation I can mostly work around this by use the fuse batch norm but densenet model in principle and in practice when ...
tensorflowtensorflow
small mistake in tensorflow document on the website
Bug
I don t know whether it s right to report document mistake in this place I don t find other place to report mistake on tensorflow s website in the custom model section of get start with tensorflow page it s mention as tf contrib learn linearregressor be actually a sub class of tf contrib learn linear...
tensorflowtensorflow
tensorflow version 1 0 0 rc2 on windows opkernel op bestsplit device type cpu for unknown op bestsplit with test code
Bug
I instal tensorflow version 1 0 0 rc2 on window 7 sp1 x64 ultimate python 3 5 2 anaconda custom 64 bit use pip install upgrade when I try run the test script from test the tensorflow installation in eclipse 4 5 or in the console import tensorflow as tf print tensorflow versi...
tensorflowtensorflow
training use multiple gpu return inf value for loss and nan for grad
Bug
I have two tesla k80 card 2 gpu per card and I spend few day test a mnist classification model use multiple gpu what I find be that the training process would always diverge get nan for grad and inf for loss when I use two gpu which be in the same card however when I allocate two gpu to my training operatio...
tensorflowtensorflow
add continuous integration review requirement to contribute md
Bug
could some basic instruction and suggestion be add to contribute md example note to add correctly instal and configure a compliant python c linter link and explanation of other guideline that must be follow how to run test on your own machine
tensorflowtensorflow
add a bit more description on add an op
Bug
hi I m try to add my custom op and feel that the document doesn t give enough information on how it work I struggle a lot and finally I m able to figure thing out by read the code of some header file can anyone take a look if the change to the document in this pr will make it well for new people
tensorflowtensorflow
crash in tf sparse tensor dense matmul
Bug
what related github issue or stackoverflow thread have you find by search the web for your problem possibly relate to 4282 also so seem to suggest that abort trap 6 be relate to access memory that tf doesn t own both problem boil down to bound error crash the python session instead of th...
tensorflowtensorflow
no attribute outer context when calculate gradient from import graph
Bug
it seem when you import a graph with a while loop you can t calculate gradient as you could on the original graph e g python import tensorflow as tf I tf constant 0 name input out tf while loop lambda I tf less I 5 lambda I tf add I 1 I name output graph def tf get d...
tensorflowtensorflow
doc break link miss py file in model file guide
Bug
on the page a tool developer s guide to tensorflow model file github md link in the graphdef section there s a link to graph metric py point to the url this url do not exist perhaps it s intend to point at a different file in that folder I don t see an obvious suitable replacement from s...
tensorflowtensorflow
seg fault when use tf session with opencv 3
Bug
hi we notice that when we try to use tensorflow with opencv 3 it consistently seg fault and crash the command be import cv2 import numpy as np import tensorflow as tf with tf session as sess img cv2 imread messi5 jpg 0 row col img shape m np float32 ...
tensorflowtensorflow
documentation for tf contrib layers stack replace by documentation for tf stack
Bug
documentation for tf contrib layer stack appear to be replace by documentation for tf stack see stack stack
tensorflowtensorflow
tensorflow retrain model compression fail with error terminate call after throw an instance of std bad alloc what std bad alloc abort core dump
Bug
I be do a college project on tensorflow I have successfully retrain use rerain py file to use that model in android program I be try to compress the model use command in the follow link shrink file size but it be get terminate throw the below error error terminate call after throw an instance of ...
tensorflowtensorflow
api documentation core graph data structure
Bug
a few thing be wrong in the documentation around tfgraphadd to collectionname value graphadd to collection url near the end of tf graph name scope the line about valueerror be incomplete the paragraph at the very end of tf graph name scope should be in tf graph add to collection both tf graph a...
tensorflowtensorflow
tf upgrade do not update concat tf image resize image
Bug
there be 2 issue about concat use concat dim instead of axis incorrect position of the explicit signature if the parameter be e g tf concat 0 tf concat 0 tile y for y in range 4 become tf concat concat dim 0 value tf concat concat dim 0 value tile y ...
tensorflowtensorflow
hclhkbu dlbench show tensorflow be slow than other framework
Bug
base on issuecomment 276648478 recent update of benchmarke state of the art deep learn software tool by shyhuai freemanx xiaowec if I get it right show some performance issue for example see table 7 alexnet r be significantly 10 time slow in tf than in other framework an it s ev...
tensorflowtensorflow
miss fclose
Bug
contrib pi example label image label image cc line 105 miss fclose infile
tensorflowtensorflow
cpu performance issue desynchronize core
Bug
from time to time I have 600 cpu utilization and great performance while feed batch into my model but after few minute or even hour cpu stick on near 100 and everything become too slow while it s 600 utilization cpu history show great synchronization between all my core after it degrade it seem li...
tensorflowtensorflow
pytorch 2 5x fast on vgg16
Bug
what related github issue or stackoverflow thread have you find by search the web for your problem start on so and be tell to post here so post comment70901342 41832779 environment info operating system ubuntu 14 04 maxwell titan x instal version of cuda and cudnn cuda 8 0 cu...
tensorflowtensorflow
what s the difference betwee tf nn seq2seq and tf python op seq2seq
Bug
what s the difference betwee tf nn seq2seq and tf python op seq2seq
tensorflowtensorflow
bug tfdbg multi gpu give valueerror duplicate node name n 0
Bug
hello tensorflow team I have be start to use your tensorflow debugger but have run into the issue that when I try and use it on a multi gpu model I get valueerror duplicate node name n 0 inspect thing close I see that the issue originate from the run metadata whose partition graph have many s...
tensorflowtensorflow
notfounderror tf file retrain graph pb
Bug
when I try to follow the tensorflow poet tutorial step I get this error when try to run the label image py file with classifier python tf file label image py tf file flower photo daisy 21652746 cc379e0eea m jpg traceback most recent call last file tf file label image py line 15 in g...
tensorflowtensorflow
multiple dequeue op be optimize away in late tf
Bug
multiple op dequeue op from the same queue start get optimize away in late tensorflow the follow executes dequeue once in jan17 head but 3 time in 12 1 sess run q dequeue q dequeue q dequeue one get expect behavior 3 dequeue when graph optimization be turn off tf configproto...
tensorflowtensorflow
resourceexhaustederror while convert imagenet dataset to tfrecord
Bug
determine list of input file and label from run medium root d9ecddfe d069 4252 aade b75fa707b3d2 imagenet data raw data train traceback most recent call last file cgpit model inception bazel bin inception build imagenet datum runfile inception inception datum build imagenet datum py line 704 in ...
tensorflowtensorflow
error in python double free or corruption prev
Bug
I m consistently get this error when stop training ctrl c on version build from head on jan17 on other hand run on version from jan5 head do not exhibit this behavior tf git version 0 12 1 1934 g27fca7d dirty session run complete in 0 01 sec with 0 500000 acc session run complete in 0...
tensorflowtensorflow
xla random number be the same across session run call
Bug
not sure if that s intend but that change behavior of train pipeline from tensorflow contrib compiler import jit jit scope jit experimental jit scope with jit scope compile op true x tf random uniform sess tf session print sess run x print sess run x outpu...
tensorflowtensorflow
well documentation for tf extract image patch
Bug
in this issue issuecomment 272525783 someone request a reverse operation to tf extract image patch the comment suggest that there exist a gradient operation l575 for this purpose which be add with this pr unfortunately it be not obvious how to apply this gradient operation and there be no inf...
tensorflowtensorflow
upgrade highwayhash
Bug
the highwayhash module which be download as an external dependency in tensorflow produce different hash result on big endian and little endian architecture this cause the test teststringtohashbucketsstrong from tensorflow python kernel test string to hash bucket op test to fail on big endian after raise a...
tensorflowtensorflow
python documentation for tensorflow ab out of sync
Bug
I check out the master branch tonight d4b5c606fc9fbd1a20b5b113b4bc831f31d889a3 I instal the code with pip into a conda environment it seem the tensorflow complex abs function have be merge into the overloaded tensorflow ab function the docstring along with the rest of the python documentation do...
tensorflowtensorflow
wrong argument order for input producer
Bug
the signature of input producer be python def input producer input tensor element shape none num epoch none shuffle true seed none capacity 32 share name none ...
tensorflowtensorflow
defun break error impl notfounderror op type not register
Bug
this code fail import tensorflow as tf from tensorflow python framework import function function defun tf float32 def custom op x return x x tf variable 1 dtype tf float32 sess tf session sess run tf global variable initializer print sess run custom op x sess...
tensorflowtensorflow
dynamic seq2seq add example of use context state
Bug
tf v1 0 in contrib seq2seq seq2seq test py the code do not show how to use context state so we can get good path as output instead of rnn output can we add sample code
tensorflowtensorflow
go unable to create two op of the same type use generate op wrapper function
Bug
as report in issuecomment 272045853 the follow test at head d4b5c606fc9fbd1a20b5b113b4bc831f31d889a3 go package bug import testing tf github com tensorflow tensorflow tensorflow go github com tensorflow tensorflow tensorflow go op func testbug t ...
tensorflowtensorflow
dynamic rnn slow use master code then 0 12 1 release code
Bug
for master code git rev parse head ec7929b878926c39255254e9aea992f0bc65aa68 same code run for 0 12 1 batch size 256 batch s 4 76 inst s 1217 74 batch size 256 batch s 5 39 inst s 1379 17 batch size 256 batch s 5 11 inst s 1306 94 batch size 256 batch s 5 05...
tensorflowtensorflow
how to train multibox object detector include in the tf detect android demo
Bug
there be a way to train a custom model for the multibox object detector that be include in the tf detect android demo
tensorflowtensorflow
text classification character rnn py be use gru cell at old location compile error
Bug
note only file github issue for bug and feature request all other topic will be close 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 ...
tensorflowtensorflow
incorrect gradient for categorical distribution entropy
Bug
the categorical distribution class provide an awesome entropy operator but apparently the gradient calculation w r t the input operator doesn t work python logit tf variable initial value 1 2 3 2 5 1 probability tf nn softmax logit log probability ...
tensorflowtensorflow
deadlock when decode tfrecord
Bug
I be store my training example as variable length tfrecord use the following function for encode def convert to tf example const exonic seq const intronic seq alt exonic seq alt intronic seq psi distribution psi std a...
tensorflowtensorflow
tensorflow doc have sigmoid cross entropy with logit args flip
Bug
label come before logit please excuse if I m way off base not sure where these doc be generate from
tensorflowtensorflow
conv2d kernel size input do not raise valueerror anymore
Bug
I notice that tensorflow 0 10 0 raise a valueerror when I pass a kernel size that be large than the first input dimension the error occur when I try to pass datum in nchw format while the default datum format be nhwc I use this code for replication of the problem import tensorflow as tf nhwc ...
tensorflowtensorflow
can not run a distribute training example with tensorflow v0 12 1
Bug
hi I be try to run a distribute tensorflow example in the official repository with v0 12 1 current late release I can run asynchronous version without problem but when I turn on sync replicas tag some error occur please check the follow log in detail this example code can be run succ...
tensorflowtensorflow
python3 pickle treat tf gfile gfile wrong
Bug
note only file github issue for bug and feature request all other topic will be close 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 ...
tensorflowtensorflow
be disallow all search engine crawl www tensorflow org
Bug
user agent disallow can you please allow search engine thank fabrice canel microsoft bing
tensorflowtensorflow
histogram fix width type error and gpu issue
Bug
with the follow script python import tensorflow as tf with tf device gpu 0 image tf zero 200 200 3 dtype tf int32 hist range tf constant 0 255 dtype tf int32 hist histogram fix width image hist range nbin 16 dtype tf int32 with tf session as...
tensorflowtensorflow
typeerror can not create initializer for non float point type
Bug
environment info operating system tensorflow docker ubuntu 14 04 version tensorflow 0 12 1 tensorflow gpu 0 12 1 if possible provide a minimal reproducible example we usually don t have time to read hundred of line of your code from official example log or other ...
tensorflowtensorflow
not register tensorarrayv2 on 12 1 on macos cpu
Bug
I m get tensorflow python framework error impl notfounderror op type not register tensorarrayv2 after upgrade to 0 12 1 on macos same code work on 0 11 on mac and 0 12 1 on ubuntu cpu this op seem to be create by dynamic rnn code node name global rnn tensorarray 1 op tensorar...
tensorflowtensorflow
incorrect gradient when use tf dynamic stitch and tf gather
Bug
in tensorflow 0 12 I find the discrepancy of gradient in two mathematically equivalent training procedure of lstm probably due to the use of tf gather and tf dynamic stitch one be the normal procedure use the whole batch of training example to unroll the lstm in each step the other first use tf gather to selec...
tensorflowtensorflow
doc a mismatch bug in api doc
Bug
there be a mismatch bug in api doc about tf graph graph it s easy to find that tf graph init graph init be a error and I check the original script op py l2003 it seem like the graph init be add by other script at the same time it cause the index url graph init ineff...
tensorflowtensorflow
minor doc issue param order in nce loss in word2vec tutorial
Bug
in section build the graph of g3doc tutorial word2vec index md python compute the nce loss use a sample of the negative label each time loss tf reduce mean tf nn nce loss nce weight nce bias embe train label num sample vocabulary size whereas the actua...
tensorflowtensorflow
cudnnlstm doesn t work with adamoptimizer
Bug
I be test how to use cudnnlstm there be not a lot of documentation on this in my own experiment I find when use adamoptimizer with cudnnlstm it always raise the follow exception I also find another repository use cudnnlstm and upload it here it also raise the same exception file ptb word lm ...
tensorflowtensorflow
fake quant with min max args have odd behavior
Bug
description on a simple linear regression example fake quant with min max args be not work if I change to fake quant with min max var with trainable quantization min max range it work just fine the min max value be the same in both approach reproducer include below what related gi...
tensorflowtensorflow
word2vec number of step in example
Bug
in the word2vec basic example master tensorflow example tutorial word2vec word2vec basic py the number of step be set to 100001 be this an arbitrary number should there not be a len datum 2 skip window 1 batch size num skip number of step to go through the datum set
tensorflowtensorflow
could not destroy cudnn handle cudnn status bad param check fail stream parent getconvolvealgorithm algorithm
Bug
I m try to use tensorflow for this project I ve get to the point where when I import tensorflow I get the message that all the cuda library be successfully open locally I can run the follow python code from run tensorflow from the command line and it work fine import tensorflow as tf hello ...
tensorflowtensorflow
fatal error tensorflow stream executor lib status h no such file or directory
Bug
I try to write my own op and I have instal tensorflow 0 12 0 with gpu support on linux this code fail include tensorflow core platform stream executor h with error fatal error tensorflow stream executor lib status h no such file or directory that file do not exist some mo...
tensorflowtensorflow
about the implementation of attention seq2seq decoder function
Bug
I be very pleased to hear the news that the attention decoder function be add recently to the master branch however the documentation of the function do not match with the actual implementation now 1 clarification on luong option inspect the code briefly I guess the paper that the luong option...
tensorflowtensorflow
widenet in example learn throw error on ubuntu 16 04 lt
Bug
be run tensorflow 0 11 0rc0 on ubuntu 16 04 lt instal via pip I be go through the guide at I think that my code have an error but then run the example script at also throw the same error be the problem my environment the script or ubuntu 16 04 I haven t try it in 14 04 training datum be do...
tensorflowtensorflow
classify image py file not find at tensorflow model image imagenet
Bug
it seem the tutorial be out of date
tensorflowtensorflow
file io get match file fail if any unrelated non matching sub directory be non readable
Bug
let filename tmp mymodel 005 file io get match file filename will raise an exception like permissiondeniederror tmp xrdp xrdp sesman jjz1ox for I I think this be a bug as this be an unrelated non matching directory so it do not matter and it will anyway not match the filenam...
tensorflowtensorflow
tf loss softmax cross entropy be deviously inconsistent
Bug
there be at least three variant of softmax cross entropy in tensorflow 1 tf nn softmax cross entropy with logit accept logit as the first argument and label as the second argument 2 tf contrib loss softmax cross entropy accept logit as the first argument and onehot label ...
tensorflowtensorflow
possible variable typo in vector representation of word documentation
Bug
the variable training input and training label in the section train the model train the model for input label in generate batch feed dict training input input training label label cur loss session run optimizer loss feed dict feed dict appear ...
tensorflowtensorflow
issue with tf one hot in 0 12 0 in gpu mode cuda error illegal address
Bug
I m use a lenet 5 mnist example from udacity s course link to the source code be below training work ok on a cpu config tf configproto device count gpu 0 but fail in a gpu mode with the follow cuda error illegal address error I c tf jenkins home workspace release win device gpu os w...