#10071: Merge silu_bw to TTNN

tenstorrent · Jul 11, 2024 · 0550df1 · 0550df1
1 parent 3f24708
commit 0550df1
Show file tree

Hide file tree

Showing 11 changed files with 35 additions and 48 deletions.
diff --git a/docs/source/ttnn/ttnn/api.rst b/docs/source/ttnn/ttnn/api.rst
@@ -198,6 +198,7 @@ Pointwise Unary
    ttnn/log_bw
    ttnn/relu6_bw
    ttnn/abs_bw
+   ttnn/silu_bw
 
 Pointwise Binary
 ================

diff --git a/docs/source/ttnn/ttnn/dependencies/tt_lib.rst b/docs/source/ttnn/ttnn/dependencies/tt_lib.rst
@@ -880,8 +880,6 @@ Backward Operations
 
 .. autofunction:: tt_lib.tensor.rpow_bw
 
-.. autofunction:: tt_lib.tensor.silu_bw
-
 .. autofunction:: tt_lib.tensor.selu_bw
 
 .. autofunction:: tt_lib.tensor.square_bw

diff --git a/docs/source/ttnn/ttnn/ttnn/silu_bw.rst b/docs/source/ttnn/ttnn/ttnn/silu_bw.rst
@@ -0,0 +1,6 @@
+.. _ttnn.silu_bw:
+
+  ttnn.silu_bw
+  #############
+
+  .. autofunction:: ttnn.silu_bw
diff --git a/tests/tt_eager/python_api_testing/unit_testing/backward_ops/test_backward_silu.py b/tests/tt_eager/python_api_testing/unit_testing/backward_ops/test_backward_silu.py
@@ -4,7 +4,7 @@
 
 import torch
 import pytest
-import tt_lib
+import ttnn
 from tests.tt_eager.python_api_testing.unit_testing.backward_ops.utility_funcs import data_gen_with_range, compare_pcc
 
 
@@ -22,7 +22,7 @@ def test_bw_silu(input_shapes, device):
 
     pyt_y = torch.nn.functional.silu(in_data)
 
-    tt_output_tensor_on_device = tt_lib.tensor.silu_bw(grad_tensor, input_tensor)
+    tt_output_tensor_on_device = ttnn.silu_bw(grad_tensor, input_tensor)
 
     in_data.retain_grad()
 

diff --git a/tt_eager/tt_dnn/op_library/backward/backward_ops.cpp b/tt_eager/tt_dnn/op_library/backward/backward_ops.cpp
@@ -1001,28 +1001,6 @@ std::vector<Tensor> rpow_bw(
     return operation::decorate_as_composite(__func__, _rpow_bw)(grad, input, exponent, output_mem_config);
 }
 
-// Silu
-// result:  grad * sigmoid_result * (1 + input * (1 - sigmoid_result))
-std::vector<Tensor> _silu_bw(const Tensor& grad, const Tensor& input, const MemoryConfig& output_mem_config) {
-    std::vector<Tensor> grad_tensor;
-    Tensor grad_sigmoid = ttnn::multiply(grad, ttnn::sigmoid(input, output_mem_config), std::nullopt, output_mem_config);
-    Tensor add_sub = ttnn::add(
-        ttnn::multiply(ttnn::subtract(ttnn::full_like(input, 1.0f) , ttnn::sigmoid(input, output_mem_config), std::nullopt, output_mem_config),
-            input,
-            std::nullopt,
-            output_mem_config),
-        1.0f,
-        std::nullopt,
-        output_mem_config);
-    Tensor grad_result = ttnn::multiply(grad_sigmoid, add_sub, std::nullopt, output_mem_config);
-
-    grad_tensor.emplace_back(grad_result);
-    return grad_tensor;
-}
-std::vector<Tensor> silu_bw(const Tensor& grad, const Tensor& input, const MemoryConfig& output_mem_config) {
-    return operation::decorate_as_composite(__func__, _silu_bw)(grad, input, output_mem_config);
-}
-
 // Selu
 // result:  torch.where(input > 0, grad * lambd, grad * lambd * alpha * torch.exp(input))
 std::vector<Tensor> _selu_bw(const Tensor& grad, const Tensor& input, const MemoryConfig& output_mem_config) {

diff --git a/tt_eager/tt_dnn/op_library/backward/backward_ops.hpp b/tt_eager/tt_dnn/op_library/backward/backward_ops.hpp
@@ -285,11 +285,6 @@ std::vector<Tensor> rpow_bw(
     float exponent,
     const MemoryConfig& output_mem_config = operation::DEFAULT_OUTPUT_MEMORY_CONFIG);
 
-std::vector<Tensor> silu_bw(
-    const Tensor& grad,
-    const Tensor& input,
-    const MemoryConfig& output_mem_config = operation::DEFAULT_OUTPUT_MEMORY_CONFIG);
-
 std::vector<Tensor> selu_bw(
     const Tensor& grad,
     const Tensor& input,

diff --git a/tt_eager/tt_lib/csrc/tt_lib_bindings_tensor_backward_ops.cpp b/tt_eager/tt_lib/csrc/tt_lib_bindings_tensor_backward_ops.cpp
@@ -808,23 +808,6 @@ namespace tt::tt_metal::detail{
                 "output_mem_config", "Layout of tensor in TT Accelerator device memory banks", "MemoryConfig", "Default is interleaved in DRAM", "No"
             )doc");
 
-    m_tensor.def("silu_bw", &tt::tt_metal::silu_bw,
-            py::arg("grad").noconvert(), py::arg("input").noconvert(), py::arg("output_mem_config").noconvert() = operation::DEFAULT_OUTPUT_MEMORY_CONFIG, R"doc(
-            Performs backward operations for silu sin of ``input`` tensors with given ``grad``.
-
-            Input tensors must have BFLOAT16 data type.
-
-            Output tensors will have BFLOAT16 data type.
-
-            .. csv-table::
-                :header: "Argument", "Description", "Data type", "Valid range", "Required"
-
-                "grad", "Gradient tensor", "Tensor", "Tensor of shape [W, Z, Y, X]", "Yes"
-                "input", "Tensor silu_bw is applied to", "Tensor", "Tensor of shape [W, Z, Y, X]", "Yes"
-                "output_mem_config", "Layout of tensor in TT Accelerator device memory banks", "MemoryConfig", "Default is interleaved in DRAM", "No"
-        )doc");
-
-
     m_tensor.def("selu_bw", &tt::tt_metal::selu_bw,
             py::arg("grad").noconvert(), py::arg("input").noconvert(), py::arg("output_mem_config").noconvert() = operation::DEFAULT_OUTPUT_MEMORY_CONFIG, R"doc(
             Performs backward operations for selu sin of ``input`` tensors with given ``grad``.

diff --git a/ttnn/cpp/ttnn/operations/eltwise/unary_backward/device/unary_backward_op.cpp b/ttnn/cpp/ttnn/operations/eltwise/unary_backward/device/unary_backward_op.cpp
@@ -418,6 +418,23 @@ std::vector<Tensor> _abs_bw(const Tensor& grad, const Tensor& input, const Memor
     return grad_tensor;
 }
 
+// Silu
+// result:  grad * sigmoid_result * (1 + input * (1 - sigmoid_result))
+std::vector<Tensor> _silu_bw(const Tensor& grad, const Tensor& input, const MemoryConfig& output_mem_config) {
+    std::vector<Tensor> grad_tensor;
+    Tensor grad_sigmoid = ttnn::multiply(grad, sigmoid(input, output_mem_config), std::nullopt, output_mem_config);
+    Tensor add_sub = add1(
+        ttnn::multiply(sub_unary(1.0f, sigmoid(input, output_mem_config), output_mem_config),
+            input,
+            std::nullopt,
+            output_mem_config),
+        output_mem_config);
+    Tensor grad_result = ttnn::multiply(grad_sigmoid, add_sub, std::nullopt, output_mem_config);
+
+    grad_tensor.emplace_back(grad_result);
+    return grad_tensor;
+}
+
 std::function<std::vector<ttnn::Tensor>(const Tensor&, const Tensor&, const MemoryConfig&)> UnaryBackwardFunction::get_function_type1(UnaryBackwardOpType OpType){
     switch (OpType) {
         case UnaryBackwardOpType::ASSIGN_BW:
@@ -458,6 +475,8 @@ std::function<std::vector<ttnn::Tensor>(const Tensor&, const Tensor&, const Memo
             return _relu6_bw;
         case UnaryBackwardOpType::ABS_BW:
             return _abs_bw;
+        case UnaryBackwardOpType::SILU_BW:
+            return _silu_bw;
         default:
             TT_ASSERT(false && "Undefined op type");
             return 0;

diff --git a/ttnn/cpp/ttnn/operations/eltwise/unary_backward/device/unary_backward_op.hpp b/ttnn/cpp/ttnn/operations/eltwise/unary_backward/device/unary_backward_op.hpp
@@ -45,6 +45,7 @@ enum class UnaryBackwardOpType {
     LOG_BW,
     RELU6_BW,
     ABS_BW,
+    SILU_BW
 };
 
 struct UnaryBackwardFunction{

diff --git a/ttnn/cpp/ttnn/operations/eltwise/unary_backward/unary_backward.hpp b/ttnn/cpp/ttnn/operations/eltwise/unary_backward/unary_backward.hpp
@@ -97,6 +97,7 @@ constexpr auto round_bw = ttnn::register_operation<operations::unary_backward::E
 constexpr auto log_bw = ttnn::register_operation<operations::unary_backward::ExecuteUnaryBackward<operations::unary_backward::UnaryBackwardOpType::LOG_BW>>("ttnn::log_bw");
 constexpr auto relu6_bw = ttnn::register_operation<operations::unary_backward::ExecuteUnaryBackward<operations::unary_backward::UnaryBackwardOpType::RELU6_BW>>("ttnn::relu6_bw");
 constexpr auto abs_bw = ttnn::register_operation<operations::unary_backward::ExecuteUnaryBackward<operations::unary_backward::UnaryBackwardOpType::ABS_BW>>("ttnn::abs_bw");
+constexpr auto silu_bw = ttnn::register_operation<operations::unary_backward::ExecuteUnaryBackward<operations::unary_backward::UnaryBackwardOpType::SILU_BW>>("ttnn::silu_bw");
 
 
 }  // namespace ttnn
diff --git a/ttnn/cpp/ttnn/operations/eltwise/unary_backward/unary_backward_pybind.hpp b/ttnn/cpp/ttnn/operations/eltwise/unary_backward/unary_backward_pybind.hpp
@@ -323,6 +323,11 @@ void py_module(py::module& module) {
         ttnn::abs_bw,
         R"doc(Performs backward operations for abs on :attr:`input_tensor` with given :attr:`grad_tensor`)doc");
 
+    detail::bind_unary_backward(
+        module,
+        ttnn::silu_bw,
+        R"doc(Performs backward operations for silu on :attr:`input_tensor` with given :attr:`grad_tensor`)doc");
+
 }
 
 }  // namespace binary_backward
-Original file line number
+Diff line change
@@ Expand Up / @@ -45,6 +45,7 @@ enum class UnaryBackwardOpType { @@
         LOG_BW,
         RELU6_BW,
         ABS_BW,
+        SILU_BW
     };
     struct UnaryBackwardFunction{
@@ Expand Down @@