diff --git a/paddle/fluid/inference/tensorrt/convert/fc_op.cc b/paddle/fluid/inference/tensorrt/convert/fc_op.cc
index bd05608d7620ee4d917b30f919fba70f6aeff17f..45b079559754a8f5c3fe39781b5700a75f425e99 100644
--- a/paddle/fluid/inference/tensorrt/convert/fc_op.cc
+++ b/paddle/fluid/inference/tensorrt/convert/fc_op.cc
@@ -59,7 +59,7 @@ class FcOpConverter : public OpConverter {
                   const framework::Scope& scope) override {
     VLOG(4) << "convert a fluid fc op to tensorrt fc layer without bias";
 
-    framework::OpDesc op_desc(op, nullptr, nullptr);
+    framework::OpDesc op_desc(op, nullptr);
     PADDLE_ENFORCE_EQ(op_desc.Input("X").size(), 1);
     PADDLE_ENFORCE_EQ(op_desc.Input("Y").size(), 1);  // Y is a weight
     PADDLE_ENFORCE_EQ(op_desc.Output("Out").size(), 1);
diff --git a/paddle/fluid/inference/tensorrt/convert/op_converter.h b/paddle/fluid/inference/tensorrt/convert/op_converter.h
index 4d21e241c0fe0abd9d454aa4f5f5ffeda747bed9..3beafeefd06f24ec50b0e61c1fabe13d7e53f242 100644
--- a/paddle/fluid/inference/tensorrt/convert/op_converter.h
+++ b/paddle/fluid/inference/tensorrt/convert/op_converter.h
@@ -40,7 +40,7 @@ class OpConverter {
   void ConvertOp(const framework::proto::OpDesc& op,
                  const std::unordered_set<std::string>& parameters,
                  const framework::Scope& scope, TensorRTEngine* engine) {
-    framework::OpDesc op_desc(op, nullptr, nullptr);
+    framework::OpDesc op_desc(op, nullptr);
 
     OpConverter* it{nullptr};
 
diff --git a/paddle/fluid/platform/profiler.cc b/paddle/fluid/platform/profiler.cc
index 3d8d64e4c2758675067834810ebb9aee1e88fdb9..01de9d7041bf3eb40884e2a6295027cccfaebd2a 100644
--- a/paddle/fluid/platform/profiler.cc
+++ b/paddle/fluid/platform/profiler.cc
@@ -127,6 +127,7 @@ double Event::CpuElapsedMs(const Event& e) const {
 
 double Event::CudaElapsedMs(const Event& e) const {
 #ifdef PADDLE_WITH_CUDA
+  if (!has_cuda_) return 0.0;
   PADDLE_ENFORCE(e.has_cuda() && has_cuda());
   PADDLE_ENFORCE(e.device() == device());
   PADDLE_ENFORCE(cudaEventSynchronize(event_));
diff --git a/python/paddle/batch.py b/python/paddle/batch.py
index 317cf037c69f8639e3760fbfce20565127794fcb..d48c54fcbb66487617b1946bc69724870c8f879c 100644
--- a/python/paddle/batch.py
+++ b/python/paddle/batch.py
@@ -15,7 +15,7 @@
 __all__ = ['batch']
 
 
-def batch(reader, batch_size):
+def batch(reader, batch_size, drop_last=False):
     """
     Create a batched reader.
 
@@ -23,6 +23,8 @@ def batch(reader, batch_size):
     :type reader: callable
     :param batch_size: size of each mini-batch
     :type batch_size: int
+    :param drop_last: drop the last batch, if the size of last batch is not equal to batch_size.
+    :type drop_last: bool
     :return: the batched reader.
     :rtype: callable
     """
@@ -35,7 +37,7 @@ def batch(reader, batch_size):
             if len(b) == batch_size:
                 yield b
                 b = []
-        if b:
+        if drop_last == False and len(b) != 0:
             yield b
 
     return batch_reader
diff --git a/python/paddle/v2/minibatch.py b/python/paddle/v2/minibatch.py
index 317cf037c69f8639e3760fbfce20565127794fcb..d48c54fcbb66487617b1946bc69724870c8f879c 100644
--- a/python/paddle/v2/minibatch.py
+++ b/python/paddle/v2/minibatch.py
@@ -15,7 +15,7 @@
 __all__ = ['batch']
 
 
-def batch(reader, batch_size):
+def batch(reader, batch_size, drop_last=False):
     """
     Create a batched reader.
 
@@ -23,6 +23,8 @@ def batch(reader, batch_size):
     :type reader: callable
     :param batch_size: size of each mini-batch
     :type batch_size: int
+    :param drop_last: drop the last batch, if the size of last batch is not equal to batch_size.
+    :type drop_last: bool
     :return: the batched reader.
     :rtype: callable
     """
@@ -35,7 +37,7 @@ def batch(reader, batch_size):
             if len(b) == batch_size:
                 yield b
                 b = []
-        if b:
+        if drop_last == False and len(b) != 0:
             yield b
 
     return batch_reader