Windows VS2015 Compile error - 'conversion from 'double' to 'float' r…

…equires a narrowing conversion' (apache#1329)
tqchen · Jun 24, 2018 · 21beebd · 21beebd
1 parent c876126
commit 21beebd
Showing 1 changed file with 1 addition and 1 deletion.
diff --git a/nnvm/include/nnvm/top/nn.h b/nnvm/include/nnvm/top/nn.h
@@ -349,7 +349,7 @@ struct MultiBoxTransformLocParam : public dmlc::Parameter<MultiBoxTransformLocPa
       .describe("Clip out-of-boundary boxes.");
     DMLC_DECLARE_FIELD(threshold).set_default(0.01)
     .describe("Threshold to be a positive prediction.");
-    DMLC_DECLARE_FIELD(variances).set_default(Tuple<float>{0.1, 0.1, 0.2, 0.2})
+    DMLC_DECLARE_FIELD(variances).set_default(Tuple<float>({0.1f, 0.1f, 0.2f, 0.2f}))
     .describe("Variances to be decoded from box regression output.");
   }
 };