diff options
Diffstat (limited to 'tensorflow/g3doc/api_docs/python/contrib.layers.md')
| -rw-r--r-- | tensorflow/g3doc/api_docs/python/contrib.layers.md | 5 |
1 files changed, 4 insertions, 1 deletions
diff --git a/tensorflow/g3doc/api_docs/python/contrib.layers.md b/tensorflow/g3doc/api_docs/python/contrib.layers.md index 8407b703a8..e3c965e704 100644 --- a/tensorflow/g3doc/api_docs/python/contrib.layers.md +++ b/tensorflow/g3doc/api_docs/python/contrib.layers.md @@ -78,7 +78,10 @@ can have speed penalty, specially in distributed settings. `batch_size`. The normalization is over all but the last dimension if `data_format` is `NHWC` and the second dimension if `data_format` is `NCHW`. -* <b>`decay`</b>: decay for the moving average. +* <b>`decay`</b>: decay for the moving average. Reasonable values for `decay` are close + to 1.0, typically in the multiple-nines range: 0.999, 0.99, 0.9, etc. Lower + `decay` value (recommend trying `decay`=0.9) if model experiences reasonably + good training performance but poor validation and/or test performance. * <b>`center`</b>: If True, subtract `beta`. If False, `beta` is ignored. * <b>`scale`</b>: If True, multiply by `gamma`. If False, `gamma` is not used. When the next layer is linear (also e.g. `nn.relu`), this can be |