Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
  • 12:28
    masa-ita synchronize #1003
  • 08:26
    googlebot labeled #1020
  • 08:26
    googlebot unlabeled #1020
  • 08:23
    googlebot labeled #1020
  • 08:23
    tfdocsbot labeled #1020
  • 08:23
    tfdocsbot labeled #1020
  • 08:23
    wckim review_requested #1020
  • 08:23
    wckim review_requested #1020
  • 08:23
    wckim opened #1020
  • 05:20
    truongsinh review_requested #959
  • 04:00
    NachoCodeX edited #1019
  • 03:59
    googlebot labeled #1019
  • 03:59
    tfdocsbot labeled #1019
  • 03:59
    tfdocsbot labeled #1019
  • 03:58
    NachoCodeX opened #1019
  • 03:58
    NachoCodeX review_requested #1019
  • 03:58
    NachoCodeX review_requested #1019
  • Sep 16 23:18
    IkerST edited #1018
  • Sep 16 23:16
    tfdocsbot labeled #1018
  • Sep 16 23:16
    tfdocsbot labeled #1018
HarikrishnanBalagopal
@HarikrishnanBalagopal
@lamberta is there any way to switch which version of the API you are viewing without going back to overview page?
for example in the python docs https://docs.python.org/3/howto/argparse.html
you can switch the version to 2.7 and it still stays on argparse without going back to the index
HarikrishnanBalagopal
@HarikrishnanBalagopal
@lamberta also is there a dedicated tensorflow help channel?
HarikrishnanBalagopal
@HarikrishnanBalagopal
the new @tf.function is very unintuitive, I am getting colocation errors between CPU and GPU 0 when I try to run this code https://bpaste.net/show/qiKt
dataset is a tf.data.Dataset
works with tf.function commented out
Crispinvz
@crispindev
Hi @lamberta I just submitted the first pull request for the /es directory. Any chance you can have a look at it for approval. It kept asking for a CLA. "Create Spanish /es Directory in TF Repo #945"
Billy Lamberta
@lamberta
@crispindev Awesome! I verified the CLA so it's no problem
Crispinvz
@crispindev
@lamberta Thanks!!
Anubhav
@anubh-v

Hi everyone,
I'm attempting to update the documentation for an operation in the Dataset API.
I have located the source file and the docstring within the source file.

I wish to update the documentation in both r1.14 and r2.0.
However, I am not clear which branch to make the PR against.

Should I be making the PR against the master branch - as mentioned sometime earlier in this chat ?

Billy Lamberta
@lamberta
@anubh-v Thanks for looking into this. Please make the updates on the master branch. Patches to older branches must get cherrypicked, docgen re-run, etc. I don't think there's much of an appetite for that with TF2 on the horizon
Anubhav
@anubh-v
I see. Thanks for the answer @lamberta
HarikrishnanBalagopal
@HarikrishnanBalagopal
@lamberta given a video dataset, how do you sample fixed length random patches from videos?
the shape of the dataset is (num_videos, video_length, height, width, channels) where video_length is different for each video in the dataset
tf.data.Dataset.from_tensor_slices doesn't work with this
Billy Lamberta
@lamberta
@HarikrishnanBalagopal probably a better question for stackoverflow. though i'd like to get a video dataset tutorial if we can get it to work in colab ok
HarikrishnanBalagopal
@HarikrishnanBalagopal
?
That is lacking information about the return value
What is the shape of the returned tensor?
@lamberta ^
is the returned tensor of shape (batch, in_planes, in_rows, in_cols, *ksizes) ?
HarikrishnanBalagopal
@HarikrishnanBalagopal
hey
anyone here?
tensorflow could really use its own support channel
Billy Lamberta
@lamberta
perhaps, but this ain't it :)
HarikrishnanBalagopal
@HarikrishnanBalagopal
I am getting weird behaviour with tf.function
if I define a training step as a function with the @tf.function annotation and then call the function in a loop the model trains
but if I move the loop into the function then it does not train the model
for instance this doesn't work https://bpaste.net/show/23C2
it fails to train the model
the documentation for @tf.function is not detailed enough to explain this
Billy Lamberta
@lamberta
work in progress. check out this detailed section on autograph/tf.function: https://github.com/tensorflow/tensorflow/blob/master/tensorflow/python/autograph/g3doc/reference/index.md
HarikrishnanBalagopal
@HarikrishnanBalagopal
it seems to be the while loop causing the issue
or not
very hard to debug
Billy Lamberta
@lamberta
Check out https://github.com/tensorflow/tensorflow/blob/master/tensorflow/python/autograph/g3doc/reference/debugging.md
I'm sure they would love feedback/PRs on it. Still early for their docs, though
HarikrishnanBalagopal
@HarikrishnanBalagopal
I am wondering if this is a google colab issue, if I include the loop inside the @tf.function then sometimes it goes into an infinite loop
might have crashed the runtime and colab just not updating
@lamberta uh the example they give there is slightly incorrect
original function
@tf.function
def f(a):
  pdb.set_trace()
  if a > 0:
    tf.print(a, 'is positive')
during debugging
>l
      8 def f(a):
      9   pdb.set_trace()
---> 10   tf.print(a)
     11   if a > 0:
     12     tf.print('is positive')
where did the extra tf.print(a) line come from?
HarikrishnanBalagopal
@HarikrishnanBalagopal
HarikrishnanBalagopal
@HarikrishnanBalagopal
@lamberta ok I figured it out, you can't use for i in range(64): or something like that since its python objects
you have to use for i in tf.range(64): for it to get converted to tf ops by @tf.function
Billy Lamberta
@lamberta
nice. might be worth a PR to those tf.function docs
HarikrishnanBalagopal
@HarikrishnanBalagopal
with that fix it no longer goes into an infinite loop