diff options
author | 2018-10-03 15:30:54 -0700 | |
---|---|---|
committer | 2018-10-03 15:31:05 -0700 | |
commit | d0882f34a46f616b3cf5854ebecef1f21e161ef0 (patch) | |
tree | 7748b3acf2635c755ebb546166e22ba50688d8ac /tensorflow/python/training | |
parent | efbee1ab2cac59f511cc0850d84414e711bbda3b (diff) | |
parent | d0690d46466bf0393ad65544d1e8c55e948df133 (diff) |
Merge pull request #22591 from EFanZh:fix-docs
PiperOrigin-RevId: 215639962
Diffstat (limited to 'tensorflow/python/training')
-rw-r--r-- | tensorflow/python/training/distribute.py | 6 |
1 files changed, 3 insertions, 3 deletions
diff --git a/tensorflow/python/training/distribute.py b/tensorflow/python/training/distribute.py index b3f3c29b2f..144b167170 100644 --- a/tensorflow/python/training/distribute.py +++ b/tensorflow/python/training/distribute.py @@ -633,7 +633,7 @@ class DistributionStrategy(object): Args: fn: function to run using this distribution strategy. The function must - have the following signature: def fn(context, *inputs). + have the following signature: `def fn(context, *inputs)`. `context` is an instance of `MultiStepContext` that will be passed when `fn` is run. `context` can be used to specify the outputs to be returned from `fn` by calling `context.set_last_step_output`. It can also be used @@ -799,9 +799,9 @@ class DistributionStrategy(object): return merged(results) ``` - Otherwise this returns `fn(var, *args, **kwargs)` colocated with `var`.' + Otherwise this returns `fn(var, *args, **kwargs)` colocated with `var`. - Neither *args nor **kwargs may contain per-device values. + Neither `*args` nor `**kwargs` may contain per-device values. If they contain mirrored values, they will be unwrapped before calling `fn`. |