Refine CPU kernel

1. Allocate memory for output before compute.
2. Rename 'ctc_decode' to 'ctc_align'

paddle/operators/ctc_align_op.cc

@@ -12,20 +12,20 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 See the License for the specific language governing permissions and
 limitations under the License. */
 
-#include "paddle/operators/ctc_decode_op.h"
+#include "paddle/operators/ctc_align_op.h"
 
 namespace paddle {
 namespace operators {
 
-class CTCDecodeOp : public framework::OperatorWithKernel {
+class CTCAlignOp : public framework::OperatorWithKernel {
  public:
   using framework::OperatorWithKernel::OperatorWithKernel;
 
   void InferShape(framework::InferShapeContext* ctx) const override {
     PADDLE_ENFORCE(ctx->HasInput("Input"),
-                   "Input of CTCDecodeOp should not be null.");
+                   "Input of CTCAlignOp should not be null.");
     PADDLE_ENFORCE(ctx->HasOutput("Output"),
-                   "Output of CTCDecodeOp should not be null.");
+                   "Output of CTCAlignOp should not be null.");
 
     auto input_dims = ctx->GetInputDim("Input");
 
@@ -42,14 +42,14 @@ class CTCDecodeOp : public framework::OperatorWithKernel {
   }
 };
 
-class CTCDecodeOpMaker : public framework::OpProtoAndCheckerMaker {
+class CTCAlignOpMaker : public framework::OpProtoAndCheckerMaker {
  public:
-  CTCDecodeOpMaker(OpProto* proto, OpAttrChecker* op_checker)
+  CTCAlignOpMaker(OpProto* proto, OpAttrChecker* op_checker)
       : OpProtoAndCheckerMaker(proto, op_checker) {
     AddInput("Input",
              "(LodTensor, default: LoDTensor<int>), Its shape is "
              "[Lp, 1], where Lp is the sum of all input sequences' length.");
-    AddOutput("Output", "(Tensor, default: Tensor<int>), The decode result.");
+    AddOutput("Output", "(Tensor, default: Tensor<int>), The align result.");
     AddAttr<int>("blank",
                  "(int, default: 0), the blank label setted in Connectionist "
                  "Temporal Classification (CTC) op.")
@@ -59,7 +59,7 @@ class CTCDecodeOpMaker : public framework::OpProtoAndCheckerMaker {
                   "merge repeated elements between two blanks. ")
         .SetDefault(true);
     AddComment(R"DOC(
-CTCDecoder is used to merge repeated elements between two blanks
+CTCAlign op is used to merge repeated elements between two blanks
 and then delete all blanks in sequence.
 
 Given:
@@ -86,7 +86,7 @@ Then:
 }  // namespace paddle
 
 namespace ops = paddle::operators;
-REGISTER_OPERATOR(ctc_decode, ops::CTCDecodeOp, ops::CTCDecodeOpMaker,
+REGISTER_OPERATOR(ctc_align, ops::CTCAlignOp, ops::CTCAlignOpMaker,
                   paddle::framework::EmptyGradOpMaker);
 REGISTER_OP_CPU_KERNEL(
-    ctc_decode, ops::CTCDecodeKernel<paddle::platform::CPUDeviceContext, int>);
+    ctc_align, ops::CTCAlignKernel<paddle::platform::CPUDeviceContext, int>);

paddle/operators/ctc_align_op.cu

@@ -15,7 +15,7 @@ limitations under the License. */
 #include <stdio.h>
 #include <thrust/device_vector.h>
 #include <thrust/host_vector.h>
-#include "paddle/operators/ctc_decode_op.h"
+#include "paddle/operators/ctc_align_op.h"
 
 namespace paddle {
 namespace operators {
@@ -42,7 +42,7 @@ __global__ void MergeAndDelCudaKernel(const int64_t num_token, const T* tokens,
 }
 
 template <typename T>
-class CTCDecodeOpCUDAKernel : public framework::OpKernel<T> {
+class CTCAlignOpCUDAKernel : public framework::OpKernel<T> {
  public:
   void Compute(const framework::ExecutionContext& ctx) const override {
     PADDLE_ENFORCE(platform::is_gpu_place(ctx.GetPlace()),
@@ -87,5 +87,5 @@ class CTCDecodeOpCUDAKernel : public framework::OpKernel<T> {
 }  // namespace operators
 }  // namespace paddle
 
-REGISTER_OP_CUDA_KERNEL(ctc_decode,
-                        paddle::operators::CTCDecodeOpCUDAKernel<int>);
+REGISTER_OP_CUDA_KERNEL(ctc_align,
+                        paddle::operators::CTCAlignOpCUDAKernel<int>);

paddle/operators/ctc_align_op.h

@@ -23,7 +23,7 @@ using Tensor = framework::Tensor;
 using LoDTensor = framework::LoDTensor;
 
 template <typename DeviceContext, typename T>
-class CTCDecodeKernel : public framework::OpKernel<T> {
+class CTCAlignKernel : public framework::OpKernel<T> {
  public:
   void Compute(const framework::ExecutionContext& ctx) const override {
     auto* input = ctx.Input<LoDTensor>("Input");
@@ -43,7 +43,8 @@ class CTCDecodeKernel : public framework::OpKernel<T> {
     bool merge_repeated = ctx.Attr<bool>("merge_repeated");
 
     // merge repeated tokens and delete blank
-    std::vector<std::vector<int>> pathes(num_sequences);
+    T* output_data = output->mutable_data<T>(ctx.GetPlace());
+    size_t output_idx = 0;
     std::vector<size_t> output_lod0(1, 0);
     const T* input_data = input->data<T>();
     for (size_t seq_idx = 0; seq_idx < num_sequences; ++seq_idx) {
@@ -52,11 +53,12 @@ class CTCDecodeKernel : public framework::OpKernel<T> {
            i < input_lod[level][seq_idx + 1]; ++i) {
         if (input_data[i] != blank &&
             !(merge_repeated && input_data[i] == prev_token)) {
-          pathes[seq_idx].push_back(input_data[i]);
+          output_data[output_idx] = input_data[i];
+          ++output_idx;
         }
         prev_token = input_data[i];
       }
-      output_lod0.push_back(output_lod0.back() + pathes[seq_idx].size());
+      output_lod0.push_back(output_idx);
     }
 
     // set output lod
@@ -65,14 +67,7 @@ class CTCDecodeKernel : public framework::OpKernel<T> {
     output->set_lod(output_lod);
 
     // resize output dims
-    T* output_data = output->mutable_data<T>(
-        {static_cast<int64_t>(output_lod0.back()), 1}, ctx.GetPlace());
-
-    // copy result to output
-    for (int i = 0; i < num_sequences; ++i) {
-      memcpy(output_data + output_lod0[i], pathes[i].data(),
-             sizeof(int) * pathes[i].size());
-    }
+    output->Resize({static_cast<int64_t>(output_lod0.back()), 1});
   }
 };
 

python/paddle/v2/fluid/tests/test_ctc_align.py

@@ -5,7 +5,7 @@ from op_test import OpTest
 from test_softmax_op import stable_softmax
 
 
-def CTCDecode(input, lod, blank, merge_repeated):
+def CTCAlign(input, lod, blank, merge_repeated):
     lod0 = lod[0]
     result = []
     for i in range(len(lod0) - 1):
@@ -20,9 +20,9 @@ def CTCDecode(input, lod, blank, merge_repeated):
     return result
 
 
-class TestCTCDecodeOp(OpTest):
+class TestCTCAlignOp(OpTest):
     def config(self):
-        self.op_type = "ctc_decode"
+        self.op_type = "ctc_align"
         self.input_lod = [[0, 11, 18]]
         self.blank = 0
         self.merge_repeated = False
@@ -32,8 +32,8 @@ class TestCTCDecodeOp(OpTest):
 
     def setUp(self):
         self.config()
-        output = CTCDecode(self.input, self.input_lod, self.blank,
-                           self.merge_repeated)
+        output = CTCAlign(self.input, self.input_lod, self.blank,
+                          self.merge_repeated)
 
         self.inputs = {"Input": (self.input, self.input_lod), }
         self.outputs = {"Output": output}
@@ -47,9 +47,9 @@ class TestCTCDecodeOp(OpTest):
         pass
 
 
-class TestCTCDecodeOpCase1(TestCTCDecodeOp):
+class TestCTCAlignOpCase1(TestCTCAlignOp):
     def config(self):
-        self.op_type = "ctc_decode"
+        self.op_type = "ctc_align"
         self.input_lod = [[0, 11, 19]]
         self.blank = 0
         self.merge_repeated = True