From 0fca8b28c013ea2cda4e3ade539414518e50be97 Mon Sep 17 00:00:00 2001
From: Manfei <41607353+ManfeiBai@users.noreply.github.com>
Date: Fri, 15 Dec 2023 09:28:40 -0800
Subject: [PATCH] Promote int to float for tanh operation (consistent with
 Pytorch) (#6166)

---
 test/cpp/test_aten_xla_tensor_2.cpp     | 14 ++++++++++++++
 test/test_core_aten_ops.py              |  1 -
 torch_xla/csrc/ops/ops_lower_fn.cpp     |  5 +++++
 torch_xla/csrc/ops/ops_xla_shape_fn.cpp |  6 +++++-
 4 files changed, 24 insertions(+), 2 deletions(-)

diff --git a/test/cpp/test_aten_xla_tensor_2.cpp b/test/cpp/test_aten_xla_tensor_2.cpp
index 4a86537d824..a9102217096 100644
--- a/test/cpp/test_aten_xla_tensor_2.cpp
+++ b/test/cpp/test_aten_xla_tensor_2.cpp
@@ -2425,6 +2425,20 @@ TEST_F(AtenXlaTensorTest, TestTanh) {
   });
 }
 
+// In torch, tanh works with integer inputs. The same should be true for
+// torch_xla
+TEST_F(AtenXlaTensorTest, TestTanhWithInt) {
+  torch::Tensor a = torch::rand({2, 2});
+  torch::Tensor b = torch::tanh(a);
+  ForEachDevice([&](const torch::Device& device) {
+    torch::Tensor xla_a = CopyToDevice(a, device);
+    torch::Tensor xla_b = torch::tanh(xla_a);
+    AllClose(b, xla_b, /*rtol=*/1e-3, /*atol=*/1e-5);
+  });
+  ExpectCounterNotChanged("aten::.*", cpp_test::GetIgnoredCounters());
+  ExpectCounterChanged("xla::tanh", cpp_test::GetIgnoredCounters());
+}
+
 TEST_F(AtenXlaTensorTest, TestClampMinMax) {
   torch::Tensor a = torch::rand({2, 2}, torch::TensorOptions(torch::kFloat));
   torch::Scalar min_val(0.311);
diff --git a/test/test_core_aten_ops.py b/test/test_core_aten_ops.py
index 449e54f08cc..15c7a12b64c 100644
--- a/test/test_core_aten_ops.py
+++ b/test/test_core_aten_ops.py
@@ -4413,7 +4413,6 @@ def test_aten_tanh_1(self):
     kwargs = dict()
     run_export_and_compare(self, torch.ops.aten.tanh, args, kwargs)
 
-  @unittest.skip
   def test_aten_tanh_2(self):
     args = (torch.randint(0, 10, (10, 10)).to(torch.int32),)
     kwargs = dict()
diff --git a/torch_xla/csrc/ops/ops_lower_fn.cpp b/torch_xla/csrc/ops/ops_lower_fn.cpp
index 1f62d388959..6aa4857e334 100644
--- a/torch_xla/csrc/ops/ops_lower_fn.cpp
+++ b/torch_xla/csrc/ops/ops_lower_fn.cpp
@@ -718,6 +718,11 @@ torch_xla::XlaOpVector Tan::Lower(LoweringContext* loctx) const {
 
 torch_xla::XlaOpVector Tanh::Lower(LoweringContext* loctx) const {
   xla::XlaOp xla_input = loctx->GetOutputOp(operand(0));
+  if (xla::primitive_util::IsIntegralType(XlaHelpers::TypeOfXlaOp(xla_input))) {
+    xla::PrimitiveType input_type = XlaHelpers::TypeOfXlaOp(xla_input);
+    xla_input = ConvertTo(xla_input, input_type, xla::PrimitiveType::F32,
+                          /*device=*/nullptr);
+  }
   return ReturnOp(xla::Tanh(xla_input), loctx);
 }
 
diff --git a/torch_xla/csrc/ops/ops_xla_shape_fn.cpp b/torch_xla/csrc/ops/ops_xla_shape_fn.cpp
index 21c6e0e8dac..825c9b59cf8 100644
--- a/torch_xla/csrc/ops/ops_xla_shape_fn.cpp
+++ b/torch_xla/csrc/ops/ops_xla_shape_fn.cpp
@@ -803,7 +803,11 @@ xla::Shape TakeOutputShape(const torch::lazy::Value& input,
 }
 
 xla::Shape TanhOutputShape(const torch::lazy::Value& input) {
-  return GetXlaShape(input);
+  xla::Shape result_shape = GetXlaShape(input);
+  if (xla::primitive_util::IsIntegralType(result_shape.element_type())) {
+    result_shape.set_element_type(xla::PrimitiveType::F32);
+  }
+  return result_shape;
 }
 
 xla::Shape TrilOutputShape(const torch::lazy::Value& input) {