[0-size Tensor No.172] Add 0-size Tensor support for gelu

paddle/phi/kernels/cpu/gelu_grad_kernel.cc

@@ -133,6 +133,9 @@ void GeluGradKernel(const Context& dev_ctx,
                     bool approximate,
                     DenseTensor* x_grad) {
   dev_ctx.template Alloc<T>(x_grad);
+  if (x_grad && x_grad->numel() == 0) {
+    return;
+  }
   auto eigen_x = EigenVector<T>::Flatten(x);
   auto eigen_out_grad = EigenVector<T>::Flatten(out_grad);
   auto eigen_x_grad = EigenVector<T>::Flatten(*x_grad);

paddle/phi/kernels/cpu/gelu_kernel.cc

@@ -93,6 +93,9 @@ void GeluKernel(const Context& dev_ctx,
                 bool approximate,
                 DenseTensor* out) {
   dev_ctx.template Alloc<T>(out);
+  if (out && out->numel() == 0) {
+    return;
+  }
   auto eigen_out = EigenVector<T>::Flatten(*out);
   auto eigen_x = EigenVector<T>::Flatten(x);
   auto& dev = *dev_ctx.eigen_device();

paddle/phi/kernels/gpu/gelu_grad_kernel.cu

@@ -66,6 +66,9 @@ void GeluGradKernel(const Context& dev_ctx,
                     bool approximate,
                     DenseTensor* x_grad) {
   dev_ctx.template Alloc<T>(x_grad);
+  if (x_grad && x_grad->numel() == 0) {
+    return;
+  }
   std::vector<const DenseTensor*> ins = {&x, &out_grad};
   std::vector<DenseTensor*> outs = {x_grad};
   if (approximate) {

paddle/phi/kernels/gpu/gelu_kernel.cu

@@ -61,6 +61,9 @@ void GeluKernel(const Context& dev_ctx,
                 bool approximate,
                 DenseTensor* out) {
   dev_ctx.template Alloc<T>(out);
+  if (out && out->numel() == 0) {
+    return;
+  }
   std::vector<const DenseTensor*> ins = {&x};
   std::vector<DenseTensor*> outs = {out};
   if (approximate) {

paddle/phi/kernels/xpu/gelu_grad_kernel.cc

@@ -29,6 +29,9 @@ void GeluGradKernel(const Context& dev_ctx,
                     DenseTensor* x_grad) {
   using XPUType = typename XPUTypeTrait<T>::Type;
   dev_ctx.template Alloc<T>(x_grad);
+  if (x_grad && x_grad->numel() == 0) {
+    return;
+  }
   int r = xpu::gelu_grad<XPUType>(
       dev_ctx.x_context(),
       reinterpret_cast<const XPUType*>(x.data<T>()),

paddle/phi/kernels/xpu/gelu_kernel.cc

@@ -29,6 +29,9 @@ void GeluKernel(const Context& dev_ctx,
                 DenseTensor* out) {
   using XPUType = typename XPUTypeTrait<T>::Type;
   dev_ctx.template Alloc<T>(out);
+  if (out && out->numel() == 0) {
+    return;
+  }
   int r = xpu::gelu<XPUType>(dev_ctx.x_context(),
                              reinterpret_cast<const XPUType*>(x.data<T>()),
                              reinterpret_cast<XPUType*>(out->data<T>()),

test/legacy_test/test_gelu_op.py

@@ -98,5 +98,45 @@ def run_gelu_op(approximate):
         )
 
 
+class TestGeluOp_ZeroSize(unittest.TestCase):
+    def _test_case1_cpu(self, approximate):
+        x = np.random.uniform(-1, 1, size=(0, 17)).astype(np.float32)
+        y_ref = gelu(x, approximate)
+
+        place = base.CPUPlace()
+        with dg.guard(place) as g:
+            x_var = paddle.to_tensor(x)
+            x_var.stop_gradient = False
+            y_var = F.gelu(x_var, approximate)
+            y_test = y_var.numpy()
+
+            loss = paddle.sum(y_var)
+            loss.backward()
+        np.testing.assert_allclose(y_ref, y_test, rtol=1e-05, atol=1e-08)
+        np.testing.assert_allclose(x_var.grad.shape, x_var.shape)
+
+    def _test_case1_gpu(self, approximate):
+        x = np.random.uniform(-1, 1, size=(0, 17)).astype(np.float32)
+        y_ref = gelu(x, approximate)
+
+        place = base.CUDAPlace(0)
+        with dg.guard(place) as g:
+            x_var = paddle.to_tensor(x)
+            x_var.stop_gradient = False
+            y_var = F.gelu(x_var, approximate)
+            y_test = y_var.numpy()
+
+            loss = paddle.sum(y_var)
+            loss.backward()
+        np.testing.assert_allclose(y_ref, y_test, rtol=1e-05, atol=1e-08)
+        np.testing.assert_allclose(x_var.grad.shape, x_var.shape)
+
+    def test_cases(self):
+        for approximate in [True, False]:
+            self._test_case1_cpu(approximate)
+            if base.is_compiled_with_cuda():
+                self._test_case1_gpu(approximate)
+
+
 if __name__ == '__main__':
     unittest.main()