Use training for batch_stats

fastai · Mar 23, 2019 · 042f845 · 042f845
1 parent 1e6e4a1
commit 042f845
Show file tree

Hide file tree

Showing 2 changed files with 2 additions and 2 deletions.
diff --git a/CHANGES.md b/CHANGES.md
@@ -18,6 +18,7 @@ of that change.
 
 ### Fixed:
 
+- Default to using training set for `batch_stats` instead of validation
 
 
 ## 1.0.50 (2019-03-19)

diff --git a/fastai/vision/data.py b/fastai/vision/data.py
@@ -164,10 +164,9 @@ def single_from_classes(path:Union[Path, str], classes:Collection[str], ds_tfms:
         sd = ImageList([], path=path, ignore_empty=True).split_none()
         return sd.label_const(0, label_cls=CategoryList, classes=classes).transform(ds_tfms, **kwargs).databunch()
 
-    def batch_stats(self, funcs:Collection[Callable]=None)->Tensor:
+    def batch_stats(self, funcs:Collection[Callable]=None, ds_type:DatasetType=DatasetType.Train)->Tensor:
         "Grab a batch of data and call reduction function `func` per channel"
         funcs = ifnone(funcs, [torch.mean,torch.std])
-        ds_type = DatasetType.Valid if self.valid_dl else DatasetType.Train
         x = self.one_batch(ds_type=ds_type, denorm=False)[0].cpu()
         return [func(channel_view(x), 1) for func in funcs]