experiments/STT-tensorflow
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Update the documentation of batch_jacobian to make the default behavior more explicit.

Browse Source 
PiperOrigin-RevId: 284641599
Change-Id: Ia9594c16d1fbc4006810357a509ca3679147abe8
This commit is contained in:
A. Unique TensorFlower 2019-12-09 14:58:12 -08:00 committed by TensorFlower Gardener
parent 2581436b1c
commit 22467f155c
1 changed files with 5 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
tensorflow/python/ops/parallel_for/gradients.py
View File

@ -95,8 +95,11 @@ def batch_jacobian(output, inp, use_pfor=True, parallel_iterations=None):
inp: A tensor with shape [b, x1, ..., x_m]
use_pfor: If true, uses pfor for computing the Jacobian. Else uses a
tf.while_loop.
parallel_iterations: A knob to control how many iterations and dispatched in
parallel. This knob can be used to control the total memory usage.
parallel_iterations: A knob to control how many iterations are vectorized
and dispatched in parallel. The default value of None, when use_pfor is
true, corresponds to vectorizing all the iterations. When use_pfor is
false, the default value of None corresponds to parallel_iterations=10.
This knob can be used to control the total memory usage.
Returns:
A tensor `t` with shape [b, y_1, ..., y_n, x1, ..., x_m] where `t[i, ...]`