reduce the lr in case of nan in small batchsize

b007055e · tensor-tang · ee82e660 · b007055e
隐藏空白更改
内联并排

Showing with 1 addition and 1 deletion

benchmark/paddle/image/vgg.py benchmark/paddle/image/vgg.py +1 -1

未找到文件。
--- a/benchmark/paddle/image/vgg.py
+++ b/benchmark/paddle/image/vgg.py
@@ -13,7 +13,7 @@ define_py_data_sources2(

 settings(
    batch_size=batch_size,
-    learning_rate=0.01 / batch_size,
+    learning_rate=0.001 / batch_size,
    learning_method=MomentumOptimizer(0.9),
    regularization=L2Regularization(0.0005 * batch_size))