fix pytorch featureset, total train sample is incorrect. (intel-analy…

…tics#2421) * fix pytorch featureset * fix style check
dding3 · Jul 2, 2020 · 0895aee · 0895aee
1 parent 58c86cb
commit 0895aee
Showing 1 changed file with 2 additions and 1 deletion.
diff --git a/python/dllib/src/bigdl/dllib/feature/common.py b/python/dllib/src/bigdl/dllib/feature/common.py
@@ -375,7 +375,8 @@ def pytorch_dataloader(cls, dataloader, bigdl_type="float"):
         if dataloader.batch_size % node_num != 0:
             true_bs = math.ceil(dataloader.batch_size / node_num) * node_num
             warning_msg = "Detect dataloader's batch_size is not divisible by node number(" + \
-                          node_num + "), will adjust batch_size to " + true_bs + " automatically"
+                          str(node_num) + "), will adjust batch_size to " + str(true_bs) + \
+                          " automatically"
             warnings.warn(warning_msg)
 
         bys = CloudPickleSerializer.dumps(CloudPickleSerializer, dataloader)