From 5bac6c010fbdf09b0c49e640a25d199229d1db27 Mon Sep 17 00:00:00 2001
From: Wei Yang <weiyang@fb.com>
Date: Mon, 18 Jun 2018 15:24:44 -0700
Subject: [PATCH 01/18] added torch.rot90() to ATen

---
 .../src/ATen/native/TensorTransformations.cpp | 36 ++++++++++++++++++
 aten/src/ATen/native/native_functions.yaml    |  2 +
 test/test_autograd.py                         |  2 +
 test/test_cuda.py                             |  5 +++
 test/test_torch.py                            | 30 +++++++++++++++
 tools/autograd/derivatives.yaml               |  3 ++
 torch/_torch_docs.py                          | 37 +++++++++++++++++++
 7 files changed, 115 insertions(+)

diff --git a/aten/src/ATen/native/TensorTransformations.cpp b/aten/src/ATen/native/TensorTransformations.cpp
index 8bce12cac2a69..4b6902609f37a 100644
--- a/aten/src/ATen/native/TensorTransformations.cpp
+++ b/aten/src/ATen/native/TensorTransformations.cpp
@@ -56,4 +56,40 @@ Tensor flip_cpu(const Tensor& self, IntList dims) {
   return out_tensor;
 }
 
+Tensor rot90(const Tensor& self, int64_t k, IntList dims) {
+  const int64_t total_dims = self.dim(), total_rot_dims = dims.size();
+
+  AT_CHECK(total_rot_dims == 2,
+    "expected total rotation dims == 2, but got dims = ", total_rot_dims);
+
+  AT_CHECK(dims[0] != dims[1],
+    "expected rotation dims to be different, but got both dims = ", dims[0]);
+
+  // check range of dims
+  auto rot_dims_v = std::vector<int64_t>(dims);
+  std::sort(rot_dims_v.begin(), rot_dims_v.end());
+
+  AT_CHECK(rot_dims_v[0] >= 0,
+    "expected rotation dims >= 0, but got dims=", rot_dims_v[0]);
+
+  AT_CHECK(rot_dims_v[1] <= total_dims - 1,
+    "expected rotation dims <= total_dims - 1, but got dims = ", rot_dims_v[1],
+    ", where total dims - 1 = ", total_dims - 1);
+
+  // handle modulo with negative k
+  k = (4 + (k % 4)) % 4;
+
+  switch(k) {
+    case 1:
+      return self.flip({dims[1]}).transpose_(dims[0], dims[1]);
+    case 2:
+      return self.flip({dims[0]}).flip({dims[1]});
+    case 3:
+      return self.transpose(dims[0], dims[1]).flip({dims[1]});
+    default:
+      return self.clone();
+  }
+
+}
+
 }} // namespace at::native
diff --git a/aten/src/ATen/native/native_functions.yaml b/aten/src/ATen/native/native_functions.yaml
index 89f2771b8dadf..a701c18dc30c7 100644
--- a/aten/src/ATen/native/native_functions.yaml
+++ b/aten/src/ATen/native/native_functions.yaml
@@ -1445,6 +1445,8 @@
     CPU: flip_cpu
     CUDA: flip_cuda
 
+- func: rot90(Tensor self, int64_t k, IntList dims) -> Tensor
+
 - func: _trilinear(Tensor i1, Tensor i2, Tensor i3, IntList expand1, IntList expand2, IntList expand3, IntList sumdim, int64_t unroll_dim=1) -> Tensor
   variants: function
 
diff --git a/test/test_autograd.py b/test/test_autograd.py
index 8559fc9284a3f..60c0feb831d9b 100644
--- a/test/test_autograd.py
+++ b/test/test_autograd.py
@@ -2648,6 +2648,8 @@ class dont_convert(tuple):
     ('flip', (S, S, S), ([0, 1, 2],), 'd012'),
     ('flip', (S, S, S), ([0, 2],), 'd02'),
     ('flip', (S, S, S), ([2, 0],), 'd20'),
+    ('rot90', (S, S, S), (1, [0, 1],), 'k1_d01'),
+    ('rot90', (S, S, S), (1, [1, 2],), 'k1_d12'),
     ('view_as', (S, S, S), (non_differentiable(torch.rand(S * S, S)),)),
     ('view_as', (), (non_differentiable(torch.tensor(5.5)),), 'scalar'),
     ('view_as', (), (non_differentiable(torch.rand(1, 1)),), 'scalar_to_dims'),
diff --git a/test/test_cuda.py b/test/test_cuda.py
index 9d56682f3cf06..090e288d89b22 100644
--- a/test/test_cuda.py
+++ b/test/test_cuda.py
@@ -420,6 +420,8 @@ def tmp(t):
     ('flip', small_3d, lambda t: [0, 1, 2], 'd012', types, True),
     ('flip', small_3d, lambda t: [0, 2], 'd02', types, True),
     ('flip', small_3d, lambda t: [2, 0], 'd20', types, True),
+    ('rot90', small_2d, lambda t: [1, [0, 1]], 'k1_d01', types, True),
+    ('rot90', small_3d, lambda t: [1, [1, 2]], 'k1_d12', types, True),
     ('rsqrt', lambda t: constant_tensor_add(1, small_3d(t)), lambda t: [], None, float_types),
     ('sinh', lambda t: tensor_clamp(small_3d(t), -1, 1), lambda t: [], None, float_types),
     ('tan', lambda t: tensor_clamp(small_3d(t), -1, 1), lambda t: [], None, float_types),
@@ -1415,6 +1417,9 @@ def test_view(self):
     def test_flip(self):
         TestTorch._test_flip(self, use_cuda=True)
 
+    def test_rot90(self):
+        TestTorch._test_rot90(self, use_cuda=True)
+
     def test_signal_window_functions(self):
         TestTorch._test_signal_window_functions(self, device=torch.device('cuda'))
 
diff --git a/test/test_torch.py b/test/test_torch.py
index 4d3d443db39bc..e27af9d622ef1 100644
--- a/test/test_torch.py
+++ b/test/test_torch.py
@@ -6735,6 +6735,36 @@ def test_reversed(self):
         val = torch.tensor(42)
         self.assertEqual(reversed(val), torch.tensor(42))
 
+    @staticmethod
+    def _test_rot90(self, use_cuda=False):
+        device = torch.device("cuda" if use_cuda else "cpu")
+        data = torch.arange(1, 5, device=device).view(2, 2)
+        self.assertEqual(torch.tensor([1, 2, 3, 4]).view(2, 2), data.rot90(0, [0, 1]))
+        self.assertEqual(torch.tensor([2, 4, 1, 3]).view(2, 2), data.rot90(1, [0, 1]))
+        self.assertEqual(torch.tensor([4, 3, 2, 1]).view(2, 2), data.rot90(2, [0, 1]))
+        self.assertEqual(torch.tensor([3, 1, 4, 2]).view(2, 2), data.rot90(3, [0, 1]))
+
+        # test for reversed order of dims
+        self.assertEqual(data.rot90(3, [0, 1]), data.rot90(1, [1, 0]))
+
+        # test for modulo of k
+        self.assertEqual(data.rot90(5, [0, 1]), data.rot90(1, [0, 1]))
+        self.assertEqual(data.rot90(3, [0, 1]), data.rot90(-1, [0, 1]))
+        self.assertEqual(data.rot90(-5, [0, 1]), data.rot90(-1, [0, 1]))
+
+        # test tensor with more than 2D
+        data = torch.arange(1, 9, device=device).view(2, 2, 2)
+        self.assertEqual(torch.tensor([2, 4, 1, 3, 6, 8, 5, 7]).view(2, 2, 2), data.rot90(1, [1, 2]))
+
+        # test for errors
+        self.assertRaises(RuntimeError, lambda: data.rot90(1, [-1, 0]))
+        self.assertRaises(RuntimeError, lambda: data.rot90(1, [0, 3]))
+        self.assertRaises(RuntimeError, lambda: data.rot90(1, [1, 1]))
+        self.assertRaises(RuntimeError, lambda: data.rot90(1, [0, 1, 2]))
+
+    def test_rot90(self):
+        self._test_rot90(self, use_cuda=False)
+
     def test_storage(self):
         v = torch.randn(3, 5)
         self.assertEqual(v.storage()[0], v.data[0][0])
diff --git a/tools/autograd/derivatives.yaml b/tools/autograd/derivatives.yaml
index 3a396d84b66e4..25fb8ee5e2477 100644
--- a/tools/autograd/derivatives.yaml
+++ b/tools/autograd/derivatives.yaml
@@ -685,6 +685,9 @@
 - name: flip(Tensor self, IntList dims)
   self: grad.flip(dims)
 
+- name: rot90(Tensor self, int64_t k, IntList dims)
+  self: grad.rot90(-k, dims)
+
 - name: take(Tensor self, Tensor index)
   self: zeros_like(self).put_(index, grad, true)
 
diff --git a/torch/_torch_docs.py b/torch/_torch_docs.py
index 84a08a155ce97..663ded865d2c9 100644
--- a/torch/_torch_docs.py
+++ b/torch/_torch_docs.py
@@ -4498,6 +4498,43 @@ def parse_kwargs(desc):
              [ 0,  1]]])
 """)
 
+add_docstr(torch.rot90,
+           r"""
+rot90(input, k, dims) -> Tensor
+
+Rotate a n-D tensor by 90 degrees in the plane specified by dims axis.
+Rotation direction is from the first towards the second axis if k > 0, and from the second towards the first for k < 0.
+
+Args:
+    input (Tensor): the input tensor
+    k (int): number of times to rotate
+    dims (a list or tuple): axis to rotate
+
+Example::
+
+    >>> x = torch.arange(4).view(2, 2)
+    >>> x
+    tensor([[0, 1],
+            [2, 3]])
+    >>> torch.flip(x, 1, [0, 1])
+    tensor([[1, 3],
+            [0, 2]])
+
+    >>> x = torch.arange(8).view(2, 2, 2)
+    >>> x
+    tensor([[[0, 1],
+             [2, 3]],
+
+            [[4, 5],
+             [6, 7]]])
+    >>> torch.flip(x, 1, [1, 2])
+    tensor([[[1, 3],
+             [0, 2]],
+
+            [[5, 7],
+             [4, 6]]])
+""")
+
 add_docstr(torch.take,
            r"""
 take(input, indices) -> Tensor

From 5e61901b125ef303344eb26f395d343ea31280ae Mon Sep 17 00:00:00 2001
From: Wei Yang <weiyang@fb.com>
Date: Mon, 18 Jun 2018 15:55:46 -0700
Subject: [PATCH 02/18] nits

---
 torch/_torch_docs.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/torch/_torch_docs.py b/torch/_torch_docs.py
index 663ded865d2c9..42208cfc6725e 100644
--- a/torch/_torch_docs.py
+++ b/torch/_torch_docs.py
@@ -4516,7 +4516,7 @@ def parse_kwargs(desc):
     >>> x
     tensor([[0, 1],
             [2, 3]])
-    >>> torch.flip(x, 1, [0, 1])
+    >>> torch.rot90(x, 1, [0, 1])
     tensor([[1, 3],
             [0, 2]])
 
@@ -4527,7 +4527,7 @@ def parse_kwargs(desc):
 
             [[4, 5],
              [6, 7]]])
-    >>> torch.flip(x, 1, [1, 2])
+    >>> torch.rot90(x, 1, [1, 2])
     tensor([[[1, 3],
              [0, 2]],
 

From 4d914d628b320619f5620c7d61da57a1257fe74d Mon Sep 17 00:00:00 2001
From: Wei Yang <weiyang@fb.com>
Date: Wed, 20 Jun 2018 00:33:10 -0700
Subject: [PATCH 03/18] support wrap_dims for negative dims in flip() and
 rot90()

---
 .../src/ATen/native/TensorTransformations.cpp | 23 +++++------
 aten/src/ATen/native/TensorTransformations.h  | 41 +++++++++++--------
 .../ATen/native/cuda/TensorTransformations.cu | 12 +++---
 test/test_autograd.py                         |  2 +
 test/test_cuda.py                             |  2 +
 test/test_torch.py                            |  6 +--
 6 files changed, 47 insertions(+), 39 deletions(-)

diff --git a/aten/src/ATen/native/TensorTransformations.cpp b/aten/src/ATen/native/TensorTransformations.cpp
index 4b6902609f37a..badf5e99afd48 100644
--- a/aten/src/ATen/native/TensorTransformations.cpp
+++ b/aten/src/ATen/native/TensorTransformations.cpp
@@ -11,9 +11,10 @@ namespace native {
 
 Tensor flip_cpu(const Tensor& self, IntList dims) {
   const int64_t total_dims = self.dim(), flip_dims_size = dims.size();
-  check_errors(total_dims, flip_dims_size, dims);
+  flip_check_errors(total_dims, flip_dims_size, dims);
 
   auto flip_dims_v = std::vector<int64_t>(dims);
+  wrap_dims(flip_dims_v, total_dims);
   std::sort(flip_dims_v.begin(), flip_dims_v.end());
   auto final_indices = std::vector<at::Tensor>(total_dims);
 
@@ -62,19 +63,16 @@ Tensor rot90(const Tensor& self, int64_t k, IntList dims) {
   AT_CHECK(total_rot_dims == 2,
     "expected total rotation dims == 2, but got dims = ", total_rot_dims);
 
-  AT_CHECK(dims[0] != dims[1],
-    "expected rotation dims to be different, but got both dims = ", dims[0]);
+  AT_CHECK(dims[0] != dims[1] && std::abs(dims[0] - dims[1]) != total_dims,
+    "expected rotation dims to be different, but got dim0 = ", dims[0],
+    " and dim1 = ", dims[1]);
 
   // check range of dims
-  auto rot_dims_v = std::vector<int64_t>(dims);
-  std::sort(rot_dims_v.begin(), rot_dims_v.end());
+  AT_CHECK(dims[0] < total_dims && dims[0] >= -total_dims,
+    "Rotation dim0 out of range, dim0 = ", dims[0]);
 
-  AT_CHECK(rot_dims_v[0] >= 0,
-    "expected rotation dims >= 0, but got dims=", rot_dims_v[0]);
-
-  AT_CHECK(rot_dims_v[1] <= total_dims - 1,
-    "expected rotation dims <= total_dims - 1, but got dims = ", rot_dims_v[1],
-    ", where total dims - 1 = ", total_dims - 1);
+  AT_CHECK(dims[1] < total_dims && dims[1] >= -total_dims,
+    "Rotation dim1 out of range, dim1 = ", dims[1]);
 
   // handle modulo with negative k
   k = (4 + (k % 4)) % 4;
@@ -83,13 +81,12 @@ Tensor rot90(const Tensor& self, int64_t k, IntList dims) {
     case 1:
       return self.flip({dims[1]}).transpose_(dims[0], dims[1]);
     case 2:
-      return self.flip({dims[0]}).flip({dims[1]});
+      return self.flip(dims);
     case 3:
       return self.transpose(dims[0], dims[1]).flip({dims[1]});
     default:
       return self.clone();
   }
-
 }
 
 }} // namespace at::native
diff --git a/aten/src/ATen/native/TensorTransformations.h b/aten/src/ATen/native/TensorTransformations.h
index 554a46f281e73..bb07766b77eb5 100644
--- a/aten/src/ATen/native/TensorTransformations.h
+++ b/aten/src/ATen/native/TensorTransformations.h
@@ -8,32 +8,37 @@
 namespace at {
 namespace native {
 
-static inline void check_errors(int64_t total_dims, int64_t flip_dims_size, IntList dims) {
+// wrap negative dims
+static inline void wrap_dims(std::vector<int64_t>& v, int64_t n) {
+  for (int64_t i = 0; i < v.size(); i++) {
+    if (v[i] < 0) {
+      v[i] = (n + (v[i] % n)) % n;
+    }
+  }
+}
+
+static inline void flip_check_errors(int64_t total_dims, int64_t flip_dims_size, IntList dims) {
   // check if number of axis in dim is valid
-  AT_CHECK(flip_dims_size > 0,
-    "expected input tensor dims > 0, but got tensor dims size=", flip_dims_size);
+  AT_CHECK(flip_dims_size > 0 && flip_dims_size <= total_dims,
+    "flip dims size out of range, got flip dims size=", flip_dims_size);
 
-  // check duplicates in dims
   auto flip_dims_v = std::vector<int64_t>(dims);
-  flip_dims_v.erase(std::unique(flip_dims_v.begin(), flip_dims_v.end()), flip_dims_v.end());
-  AT_CHECK((int64_t)flip_dims_v.size() == flip_dims_size,
-    "dims has duplicates, original flip dims size=", flip_dims_size,
-    ", but unique flip dims size=", flip_dims_v.size());
-
-  // check len of dims
-  AT_CHECK(flip_dims_size <= total_dims,
-    "expected flip dims size <= tensor total dims, but got flip dims size=",
-    flip_dims_size, " and tensor total dim=", total_dims);
 
   // check if dims axis within range
   auto min_max_d = std::minmax_element(flip_dims_v.begin(), flip_dims_v.end());
 
-  AT_CHECK(*min_max_d.first >= 0,
-    "expected flip dims axis >= 0, but got min flip dims=", *min_max_d.first);
+  AT_CHECK(*min_max_d.first < total_dims && *min_max_d.first >= -total_dims,
+    "The min flip dims out of range, got min flip dims=", *min_max_d.first);
+
+  AT_CHECK(*min_max_d.second < total_dims && *min_max_d.second >= -total_dims,
+    "The max flip dims out of range, got max flip dims=", *min_max_d.second);
 
-  AT_CHECK(*min_max_d.second < total_dims,
-    "expected flip dims axis < tensor total dims, but got max flip dims=",
-    *min_max_d.second, " and tensor total dim=", total_dims);
+  // check duplicates in dims
+  wrap_dims(flip_dims_v, total_dims);
+  flip_dims_v.erase(std::unique(flip_dims_v.begin(), flip_dims_v.end()), flip_dims_v.end());
+  AT_CHECK((int64_t)flip_dims_v.size() == flip_dims_size,
+    "dims has duplicates, original flip dims size=", flip_dims_size,
+    ", but unique flip dims size=", flip_dims_v.size());
 }
 
 }}  // namespace at::native
diff --git a/aten/src/ATen/native/cuda/TensorTransformations.cu b/aten/src/ATen/native/cuda/TensorTransformations.cu
index ee4d030f775e7..800b5a07a662e 100644
--- a/aten/src/ATen/native/cuda/TensorTransformations.cu
+++ b/aten/src/ATen/native/cuda/TensorTransformations.cu
@@ -69,7 +69,7 @@ void flip_cuda_kernel(scalar_t* in_tensor, scalar_t* out_tensor, int64_t N, int6
 Tensor flip_cuda(const Tensor& self, IntList dims) {
   auto in_tensor = self;
   const int64_t flip_dims_size = dims.size(), total_dims = in_tensor.dim(), N = in_tensor.numel();
-  check_errors(total_dims, flip_dims_size, dims);
+  flip_check_errors(total_dims, flip_dims_size, dims);
 
   int64_t block_size = 512;
   dim3 dim_block(block_size);
@@ -80,13 +80,16 @@ Tensor flip_cuda(const Tensor& self, IntList dims) {
     return out_tensor;
   }
 
+  auto flip_dims = std::vector<int64_t>(dims);
+  wrap_dims(flip_dims, total_dims);
+
   // use kernel_pointwise_flip_apply2 only when to-flip dim is the 1st or last dim, where collapseDims can reduce the amount of work
-  if (flip_dims_size == 1 && in_tensor.is_contiguous() && (dims[0] == 0 || dims[0] == total_dims - 1)) {
+  if (flip_dims_size == 1 && in_tensor.is_contiguous() && (flip_dims[0] == 0 || flip_dims[0] == total_dims - 1)) {
     AT_DISPATCH_ALL_TYPES_AND_HALF(in_tensor.type(), "flip_cuda", [&] {
       auto in_tensor_info = cuda::detail::getTensorInfo<scalar_t, int64_t>(in_tensor);
       auto out_tensor_info = cuda::detail::getTensorInfo<scalar_t, int64_t>(out_tensor);
-      int flip_dim = in_tensor_info.collapseDims(dims[0]);
-      out_tensor_info.collapseDims(dims[0]);
+      int flip_dim = in_tensor_info.collapseDims(flip_dims[0]);
+      out_tensor_info.collapseDims(flip_dims[0]);
       kernel_pointwise_flip_apply2<scalar_t, int64_t>
         <<<dim_grid, dim_block, 0, at::cuda::getCurrentCUDAStream()>>>(
           in_tensor_info, out_tensor_info, N, flip_dim, total_dims);
@@ -94,7 +97,6 @@ Tensor flip_cuda(const Tensor& self, IntList dims) {
     return out_tensor;
   }
 
-  auto flip_dims = std::vector<int64_t>(dims);
   auto flip_dims_t = at::CPU(kLong).tensorFromBlob(flip_dims.data(), {static_cast<int64_t>(flip_dims.size())});
 
   auto shape = std::vector<int64_t>(in_tensor.sizes());
diff --git a/test/test_autograd.py b/test/test_autograd.py
index 60c0feb831d9b..be88f7be07fd9 100644
--- a/test/test_autograd.py
+++ b/test/test_autograd.py
@@ -2648,8 +2648,10 @@ class dont_convert(tuple):
     ('flip', (S, S, S), ([0, 1, 2],), 'd012'),
     ('flip', (S, S, S), ([0, 2],), 'd02'),
     ('flip', (S, S, S), ([2, 0],), 'd20'),
+    ('flip', (S, S, S), ([-1],), 'neg_d'),
     ('rot90', (S, S, S), (1, [0, 1],), 'k1_d01'),
     ('rot90', (S, S, S), (1, [1, 2],), 'k1_d12'),
+    ('rot90', (S, S, S), (1, [1, -1],), 'k1_neg_d'),
     ('view_as', (S, S, S), (non_differentiable(torch.rand(S * S, S)),)),
     ('view_as', (), (non_differentiable(torch.tensor(5.5)),), 'scalar'),
     ('view_as', (), (non_differentiable(torch.rand(1, 1)),), 'scalar_to_dims'),
diff --git a/test/test_cuda.py b/test/test_cuda.py
index 090e288d89b22..ebf0badf05af1 100644
--- a/test/test_cuda.py
+++ b/test/test_cuda.py
@@ -420,8 +420,10 @@ def tmp(t):
     ('flip', small_3d, lambda t: [0, 1, 2], 'd012', types, True),
     ('flip', small_3d, lambda t: [0, 2], 'd02', types, True),
     ('flip', small_3d, lambda t: [2, 0], 'd20', types, True),
+    ('flip', small_3d, lambda t: [-1], 'neg_d', types, True),
     ('rot90', small_2d, lambda t: [1, [0, 1]], 'k1_d01', types, True),
     ('rot90', small_3d, lambda t: [1, [1, 2]], 'k1_d12', types, True),
+    ('rot90', small_3d, lambda t: [1, [1, -1]], 'k1_neg_d', types, True),
     ('rsqrt', lambda t: constant_tensor_add(1, small_3d(t)), lambda t: [], None, float_types),
     ('sinh', lambda t: tensor_clamp(small_3d(t), -1, 1), lambda t: [], None, float_types),
     ('tan', lambda t: tensor_clamp(small_3d(t), -1, 1), lambda t: [], None, float_types),
diff --git a/test/test_torch.py b/test/test_torch.py
index e27af9d622ef1..4c910d2a71c1e 100644
--- a/test/test_torch.py
+++ b/test/test_torch.py
@@ -6686,6 +6686,8 @@ def _test_flip(self, use_cuda=False):
         self.assertEqual(torch.tensor([7, 8, 5, 6, 3, 4, 1, 2]).view(2, 2, 2), data.flip(0, 1))
         self.assertEqual(torch.tensor([8, 7, 6, 5, 4, 3, 2, 1]).view(2, 2, 2), data.flip(0, 1, 2))
 
+        # check for wrap dim
+        self.assertEqual(torch.tensor([2, 1, 4, 3, 6, 5, 8, 7]).view(2, 2, 2), data.flip(-1))
         # check for permute
         self.assertEqual(torch.tensor([6, 5, 8, 7, 2, 1, 4, 3]).view(2, 2, 2), data.flip(0, 2))
         self.assertEqual(torch.tensor([6, 5, 8, 7, 2, 1, 4, 3]).view(2, 2, 2), data.flip(2, 0))
@@ -6696,8 +6698,6 @@ def _test_flip(self, use_cuda=False):
         self.assertRaises(TypeError, lambda: data.flip())
         # not allow size of flip dim > total dims
         self.assertRaises(RuntimeError, lambda: data.flip(0, 1, 2, 3))
-        # not allow dim < 0
-        self.assertRaises(RuntimeError, lambda: data.flip(-1))
         # not allow dim > max dim
         self.assertRaises(RuntimeError, lambda: data.flip(3))
 
@@ -6755,9 +6755,9 @@ def _test_rot90(self, use_cuda=False):
         # test tensor with more than 2D
         data = torch.arange(1, 9, device=device).view(2, 2, 2)
         self.assertEqual(torch.tensor([2, 4, 1, 3, 6, 8, 5, 7]).view(2, 2, 2), data.rot90(1, [1, 2]))
+        self.assertEqual(data.rot90(1, [1, -1]), data.rot90(1, [1, 2]))
 
         # test for errors
-        self.assertRaises(RuntimeError, lambda: data.rot90(1, [-1, 0]))
         self.assertRaises(RuntimeError, lambda: data.rot90(1, [0, 3]))
         self.assertRaises(RuntimeError, lambda: data.rot90(1, [1, 1]))
         self.assertRaises(RuntimeError, lambda: data.rot90(1, [0, 1, 2]))

From d6ce84d7fe400fdbc730fd98782ded4e477c058f Mon Sep 17 00:00:00 2001
From: Wei Yang <weiyang@fb.com>
Date: Wed, 20 Jun 2018 13:39:48 -0700
Subject: [PATCH 04/18] 1. moved wrap_dims() to WrapDimUtils.h; 2. re-arange
 ops order for rot90 k=3; 3. TODO: add default values to args k and dims

---
 aten/src/ATen/WrapDimUtils.h                   |  9 +++++++++
 aten/src/ATen/native/TensorTransformations.cpp |  2 +-
 aten/src/ATen/native/TensorTransformations.h   | 10 +---------
 test/test_torch.py                             |  3 +++
 4 files changed, 14 insertions(+), 10 deletions(-)

diff --git a/aten/src/ATen/WrapDimUtils.h b/aten/src/ATen/WrapDimUtils.h
index a07efa24d21a2..8bd7239da776c 100644
--- a/aten/src/ATen/WrapDimUtils.h
+++ b/aten/src/ATen/WrapDimUtils.h
@@ -86,4 +86,13 @@ static inline int64_t legacy_cat_wrap_dim(int64_t dim, TensorList tensors) {
   return dim;
 }
 
+// wrap negative dims in v, where total dim is n
+static inline void wrap_dims(std::vector<int64_t>& v, int64_t n) {
+  for (auto i = 0; i < v.size(); i++) {
+    if (v[i] < 0) {
+      v[i] = (n + (v[i] % n)) % n;
+    }
+  }
+}
+
 }
diff --git a/aten/src/ATen/native/TensorTransformations.cpp b/aten/src/ATen/native/TensorTransformations.cpp
index badf5e99afd48..6431a45bcba1a 100644
--- a/aten/src/ATen/native/TensorTransformations.cpp
+++ b/aten/src/ATen/native/TensorTransformations.cpp
@@ -83,7 +83,7 @@ Tensor rot90(const Tensor& self, int64_t k, IntList dims) {
     case 2:
       return self.flip(dims);
     case 3:
-      return self.transpose(dims[0], dims[1]).flip({dims[1]});
+      return self.flip({dims[0]}).transpose_(dims[0], dims[1]);
     default:
       return self.clone();
   }
diff --git a/aten/src/ATen/native/TensorTransformations.h b/aten/src/ATen/native/TensorTransformations.h
index bb07766b77eb5..e000b13f6a379 100644
--- a/aten/src/ATen/native/TensorTransformations.h
+++ b/aten/src/ATen/native/TensorTransformations.h
@@ -1,6 +1,7 @@
 #include "ATen/ATen.h"
 
 #include <ATen/Error.h>
+#include <ATen/WrapDimUtils.h>
 
 #include <algorithm>
 #include <vector>
@@ -8,15 +9,6 @@
 namespace at {
 namespace native {
 
-// wrap negative dims
-static inline void wrap_dims(std::vector<int64_t>& v, int64_t n) {
-  for (int64_t i = 0; i < v.size(); i++) {
-    if (v[i] < 0) {
-      v[i] = (n + (v[i] % n)) % n;
-    }
-  }
-}
-
 static inline void flip_check_errors(int64_t total_dims, int64_t flip_dims_size, IntList dims) {
   // check if number of axis in dim is valid
   AT_CHECK(flip_dims_size > 0 && flip_dims_size <= total_dims,
diff --git a/test/test_torch.py b/test/test_torch.py
index 4c910d2a71c1e..180c2c5fc9589 100644
--- a/test/test_torch.py
+++ b/test/test_torch.py
@@ -6744,6 +6744,9 @@ def _test_rot90(self, use_cuda=False):
         self.assertEqual(torch.tensor([4, 3, 2, 1]).view(2, 2), data.rot90(2, [0, 1]))
         self.assertEqual(torch.tensor([3, 1, 4, 2]).view(2, 2), data.rot90(3, [0, 1]))
 
+        # test for default args k=1, dims=[0, 1]
+        # self.assertEqual(data.rot90(), data.rot90(1, [0, 1]))
+
         # test for reversed order of dims
         self.assertEqual(data.rot90(3, [0, 1]), data.rot90(1, [1, 0]))
 

From b95f07f90d97ba056367028a48cbafe438b90829 Mon Sep 17 00:00:00 2001
From: Wei Yang <weiyang@fb.com>
Date: Wed, 20 Jun 2018 23:56:59 -0700
Subject: [PATCH 05/18] 1. addressed comments; 2. hacked an IntList parser at
 python_arg_parser.cpp

---
 aten/src/ATen/WrapDimUtils.h               |  2 +-
 aten/src/ATen/native/native_functions.yaml |  3 ++-
 aten/src/ATen/native_parse.py              |  2 ++
 test/test_autograd.py                      |  1 +
 test/test_cuda.py                          |  1 +
 test/test_torch.py                         |  2 +-
 torch/csrc/utils/python_arg_parser.cpp     | 28 +++++++++++++++++++++-
 7 files changed, 35 insertions(+), 4 deletions(-)

diff --git a/aten/src/ATen/WrapDimUtils.h b/aten/src/ATen/WrapDimUtils.h
index 8bd7239da776c..0f3179f97bf75 100644
--- a/aten/src/ATen/WrapDimUtils.h
+++ b/aten/src/ATen/WrapDimUtils.h
@@ -88,7 +88,7 @@ static inline int64_t legacy_cat_wrap_dim(int64_t dim, TensorList tensors) {
 
 // wrap negative dims in v, where total dim is n
 static inline void wrap_dims(std::vector<int64_t>& v, int64_t n) {
-  for (auto i = 0; i < v.size(); i++) {
+  for (size_t i = 0; i < v.size(); i++) {
     if (v[i] < 0) {
       v[i] = (n + (v[i] % n)) % n;
     }
diff --git a/aten/src/ATen/native/native_functions.yaml b/aten/src/ATen/native/native_functions.yaml
index a701c18dc30c7..a326d56f4aa79 100644
--- a/aten/src/ATen/native/native_functions.yaml
+++ b/aten/src/ATen/native/native_functions.yaml
@@ -1445,7 +1445,8 @@
     CPU: flip_cpu
     CUDA: flip_cuda
 
-- func: rot90(Tensor self, int64_t k, IntList dims) -> Tensor
+# default IntList value {0,1} should not add space after comma, since native_parse.py uses ', ' to split args
+- func: rot90(Tensor self, int64_t k=1, IntList dims={0,1}) -> Tensor
 
 - func: _trilinear(Tensor i1, Tensor i2, Tensor i3, IntList expand1, IntList expand2, IntList expand3, IntList sumdim, int64_t unroll_dim=1) -> Tensor
   variants: function
diff --git a/aten/src/ATen/native_parse.py b/aten/src/ATen/native_parse.py
index 13d852d2f1e14..c3a0068df7ea2 100644
--- a/aten/src/ATen/native_parse.py
+++ b/aten/src/ATen/native_parse.py
@@ -20,6 +20,8 @@ def parse_default(s):
         return s
     elif s == '{}':
         return '{}'
+    elif re.match(r'{.*}', s):
+        return s
     elif s == 'nullopt':
         return s
     try:
diff --git a/test/test_autograd.py b/test/test_autograd.py
index be88f7be07fd9..9849bcbcaf3c6 100644
--- a/test/test_autograd.py
+++ b/test/test_autograd.py
@@ -2652,6 +2652,7 @@ class dont_convert(tuple):
     ('rot90', (S, S, S), (1, [0, 1],), 'k1_d01'),
     ('rot90', (S, S, S), (1, [1, 2],), 'k1_d12'),
     ('rot90', (S, S, S), (1, [1, -1],), 'k1_neg_d'),
+    ('rot90', (S, S, S), (), 'default'),
     ('view_as', (S, S, S), (non_differentiable(torch.rand(S * S, S)),)),
     ('view_as', (), (non_differentiable(torch.tensor(5.5)),), 'scalar'),
     ('view_as', (), (non_differentiable(torch.rand(1, 1)),), 'scalar_to_dims'),
diff --git a/test/test_cuda.py b/test/test_cuda.py
index ebf0badf05af1..b391b73c451ef 100644
--- a/test/test_cuda.py
+++ b/test/test_cuda.py
@@ -424,6 +424,7 @@ def tmp(t):
     ('rot90', small_2d, lambda t: [1, [0, 1]], 'k1_d01', types, True),
     ('rot90', small_3d, lambda t: [1, [1, 2]], 'k1_d12', types, True),
     ('rot90', small_3d, lambda t: [1, [1, -1]], 'k1_neg_d', types, True),
+    ('rot90', small_3d, lambda t: [], 'default', types, True),
     ('rsqrt', lambda t: constant_tensor_add(1, small_3d(t)), lambda t: [], None, float_types),
     ('sinh', lambda t: tensor_clamp(small_3d(t), -1, 1), lambda t: [], None, float_types),
     ('tan', lambda t: tensor_clamp(small_3d(t), -1, 1), lambda t: [], None, float_types),
diff --git a/test/test_torch.py b/test/test_torch.py
index 180c2c5fc9589..5595035507c01 100644
--- a/test/test_torch.py
+++ b/test/test_torch.py
@@ -6745,7 +6745,7 @@ def _test_rot90(self, use_cuda=False):
         self.assertEqual(torch.tensor([3, 1, 4, 2]).view(2, 2), data.rot90(3, [0, 1]))
 
         # test for default args k=1, dims=[0, 1]
-        # self.assertEqual(data.rot90(), data.rot90(1, [0, 1]))
+        self.assertEqual(data.rot90(), data.rot90(1, [0, 1]))
 
         # test for reversed order of dims
         self.assertEqual(data.rot90(3, [0, 1]), data.rot90(1, [1, 0]))
diff --git a/torch/csrc/utils/python_arg_parser.cpp b/torch/csrc/utils/python_arg_parser.cpp
index c6d2315a6fe4e..97f17f3b50415 100644
--- a/torch/csrc/utils/python_arg_parser.cpp
+++ b/torch/csrc/utils/python_arg_parser.cpp
@@ -161,6 +161,32 @@ static inline at::optional<int64_t> parse_as_integer(const std::string& s) {
   return (*str_end == 0) ? at::optional<int64_t>(ans) : at::nullopt;
 }
 
+/*
+Parse default value of IntList declared at native_functions.yaml
+There are two kinds of default values:
+1. IntList[2] x=1 (where size=2, value={1,1}
+2. IntList x={1,2,3} (where size=3, value={1,2,3}, note that there cannot be space after comma since native_parse.py uses ', '
+to split args)
+*/
+static inline std::vector<int64_t> parse_intlist_args(const std::string& s, int64_t size) {
+  if (s[0] != '{') {
+    return std::vector<int64_t>(size, std::stoi(s));
+  }
+
+  auto args = std::vector<int64_t>();
+  int64_t x = 0;
+  for (size_t i = 0; i < s.size();) {
+    if ('0' <= s[i] && s[i] <= '9') {
+      while ('0' <= s[i] && s[i] <= '9') {
+        x = x * 10 + (s[i++] - '0');
+      }
+      args.emplace_back(x);
+      x = 0;
+    }
+    i += 1;
+  }
+  return args;
+}
 
 void FunctionParameter::set_default_str(const std::string& str) {
   if (str == "None") {
@@ -189,7 +215,7 @@ void FunctionParameter::set_default_str(const std::string& str) {
     }
   } else if (type_ == ParameterType::INT_LIST) {
     if (str != "None") {
-      default_intlist.assign(size, std::stoi(str));
+      default_intlist = parse_intlist_args(str, size);
     }
   } else if (type_ == ParameterType::SCALARTYPE) {
     if (str == "None") {

From e79638a0ebdc621169467e5c7f1c19e4471ece5f Mon Sep 17 00:00:00 2001
From: Wei Yang <weiyang@fb.com>
Date: Mon, 25 Jun 2018 13:58:31 -0700
Subject: [PATCH 06/18] wip

---
 torch/csrc/utils/python_arg_parser.cpp | 41 +++++++++++++++++++++-----
 1 file changed, 33 insertions(+), 8 deletions(-)

diff --git a/torch/csrc/utils/python_arg_parser.cpp b/torch/csrc/utils/python_arg_parser.cpp
index 97f17f3b50415..f573e022a84b2 100644
--- a/torch/csrc/utils/python_arg_parser.cpp
+++ b/torch/csrc/utils/python_arg_parser.cpp
@@ -168,22 +168,47 @@ There are two kinds of default values:
 2. IntList x={1,2,3} (where size=3, value={1,2,3}, note that there cannot be space after comma since native_parse.py uses ', '
 to split args)
 */
+// static inline std::vector<int64_t> parse_intlist_args(const std::string& s, int64_t size) {
+//   if (s[0] != '{') {
+//     return std::vector<int64_t>(size, std::stoi(s));
+//   }
+//
+//   auto args = std::vector<int64_t>();
+//   int64_t x = 0;
+//   for (size_t i = 0; i < s.size();) {
+//     if ('0' <= s[i] && s[i] <= '9') {
+//       while ('0' <= s[i] && s[i] <= '9') {
+//         x = x * 10 + (s[i++] - '0');
+//       }
+//       args.emplace_back(x);
+//       x = 0;
+//     }
+//     i += 1;
+//   }
+//   return args;
+// }
+
 static inline std::vector<int64_t> parse_intlist_args(const std::string& s, int64_t size) {
   if (s[0] != '{') {
     return std::vector<int64_t>(size, std::stoi(s));
   }
 
   auto args = std::vector<int64_t>();
-  int64_t x = 0;
-  for (size_t i = 0; i < s.size();) {
-    if ('0' <= s[i] && s[i] <= '9') {
-      while ('0' <= s[i] && s[i] <= '9') {
-        x = x * 10 + (s[i++] - '0');
-      }
-      args.emplace_back(x);
+  int64_t x = 0, sign = 1;
+  for (size_t i = 0; i < s.size(); i++) {
+    if (s[i] == '-') {
+      sign *= -1;
+    }
+    else if ('0' <= s[i] && s[i] <= '9') {
+      x = x * 10 + (s[i] - '0');
+    }
+    else if (s[i] == ',' || s[i] == '}') {
+      args.emplace_back(sign * x);
+      sign = 1;
       x = 0;
     }
-    i += 1;
+    else { // '{'
+    }
   }
   return args;
 }

From f0c861279b9d4bedda764cf0dba459204bcb7a0a Mon Sep 17 00:00:00 2001
From: Wei Yang <weiyang@fb.com>
Date: Mon, 25 Jun 2018 18:37:26 -0700
Subject: [PATCH 07/18] supports negative default IntList values, and addresses
 comments

---
 aten/src/ATen/WrapDimUtils.h           |  8 ++++----
 torch/csrc/utils/python_arg_parser.cpp | 24 ++----------------------
 2 files changed, 6 insertions(+), 26 deletions(-)

diff --git a/aten/src/ATen/WrapDimUtils.h b/aten/src/ATen/WrapDimUtils.h
index 0f3179f97bf75..207dc46466155 100644
--- a/aten/src/ATen/WrapDimUtils.h
+++ b/aten/src/ATen/WrapDimUtils.h
@@ -87,10 +87,10 @@ static inline int64_t legacy_cat_wrap_dim(int64_t dim, TensorList tensors) {
 }
 
 // wrap negative dims in v, where total dim is n
-static inline void wrap_dims(std::vector<int64_t>& v, int64_t n) {
-  for (size_t i = 0; i < v.size(); i++) {
-    if (v[i] < 0) {
-      v[i] = (n + (v[i] % n)) % n;
+static inline void wrap_dims(std::vector<int64_t>& to_transform_dims, int64_t tensor_total_dims) {
+  for (size_t i = 0; i < to_transform_dims.size(); i++) {
+    if (to_transform_dims[i] < 0) {
+      to_transform_dims[i] = (tensor_total_dims + (to_transform_dims[i] % tensor_total_dims)) % tensor_total_dims;
     }
   }
 }
diff --git a/torch/csrc/utils/python_arg_parser.cpp b/torch/csrc/utils/python_arg_parser.cpp
index f573e022a84b2..33dc66eb5b12c 100644
--- a/torch/csrc/utils/python_arg_parser.cpp
+++ b/torch/csrc/utils/python_arg_parser.cpp
@@ -163,31 +163,11 @@ static inline at::optional<int64_t> parse_as_integer(const std::string& s) {
 
 /*
 Parse default value of IntList declared at native_functions.yaml
+
 There are two kinds of default values:
 1. IntList[2] x=1 (where size=2, value={1,1}
-2. IntList x={1,2,3} (where size=3, value={1,2,3}, note that there cannot be space after comma since native_parse.py uses ', '
-to split args)
+2. IntList x={1,2,3} (where size=3, value={1,2,3}, note that there cannot be space after comma since native_parse.py uses ', ' to split args)
 */
-// static inline std::vector<int64_t> parse_intlist_args(const std::string& s, int64_t size) {
-//   if (s[0] != '{') {
-//     return std::vector<int64_t>(size, std::stoi(s));
-//   }
-//
-//   auto args = std::vector<int64_t>();
-//   int64_t x = 0;
-//   for (size_t i = 0; i < s.size();) {
-//     if ('0' <= s[i] && s[i] <= '9') {
-//       while ('0' <= s[i] && s[i] <= '9') {
-//         x = x * 10 + (s[i++] - '0');
-//       }
-//       args.emplace_back(x);
-//       x = 0;
-//     }
-//     i += 1;
-//   }
-//   return args;
-// }
-
 static inline std::vector<int64_t> parse_intlist_args(const std::string& s, int64_t size) {
   if (s[0] != '{') {
     return std::vector<int64_t>(size, std::stoi(s));

From 57e309aa7ca41ffd609b1d8a3903049f221fcf4e Mon Sep 17 00:00:00 2001
From: Wei Yang <weiyang@fb.com>
Date: Mon, 25 Jun 2018 20:06:11 -0700
Subject: [PATCH 08/18] add more tests in parse_intlist_args

---
 aten/src/ATen/WrapDimUtils.h           |  2 +-
 torch/csrc/utils/python_arg_parser.cpp | 12 +++++++++---
 2 files changed, 10 insertions(+), 4 deletions(-)

diff --git a/aten/src/ATen/WrapDimUtils.h b/aten/src/ATen/WrapDimUtils.h
index 207dc46466155..35feba7410627 100644
--- a/aten/src/ATen/WrapDimUtils.h
+++ b/aten/src/ATen/WrapDimUtils.h
@@ -86,7 +86,7 @@ static inline int64_t legacy_cat_wrap_dim(int64_t dim, TensorList tensors) {
   return dim;
 }
 
-// wrap negative dims in v, where total dim is n
+// wrap negative dims in to_transform_dims
 static inline void wrap_dims(std::vector<int64_t>& to_transform_dims, int64_t tensor_total_dims) {
   for (size_t i = 0; i < to_transform_dims.size(); i++) {
     if (to_transform_dims[i] < 0) {
diff --git a/torch/csrc/utils/python_arg_parser.cpp b/torch/csrc/utils/python_arg_parser.cpp
index 33dc66eb5b12c..c6f3fe85fa4f5 100644
--- a/torch/csrc/utils/python_arg_parser.cpp
+++ b/torch/csrc/utils/python_arg_parser.cpp
@@ -173,23 +173,29 @@ static inline std::vector<int64_t> parse_intlist_args(const std::string& s, int6
     return std::vector<int64_t>(size, std::stoi(s));
   }
 
+  size_t n = s.size();
+  AT_CHECK(s[0] == '{', "Default value of IntList is missing left brace '{', found ", s[0]);
+  AT_CHECK(s[n - 1] == '}', "Default value of IntList is missing right brace '}', found ", s[n - 1]);
+
   auto args = std::vector<int64_t>();
   int64_t x = 0, sign = 1;
-  for (size_t i = 0; i < s.size(); i++) {
+  for (size_t i = 1; i < n - 1; i++) {
     if (s[i] == '-') {
       sign *= -1;
     }
     else if ('0' <= s[i] && s[i] <= '9') {
       x = x * 10 + (s[i] - '0');
     }
-    else if (s[i] == ',' || s[i] == '}') {
+    else if (s[i] == ',') {
       args.emplace_back(sign * x);
       sign = 1;
       x = 0;
     }
-    else { // '{'
+    else {
+      AT_ERROR("Illegal char in IntList default value: ", s[i]);
     }
   }
+  args.emplace_back(sign * x);
   return args;
 }
 

From b3af2e9d75e85a4717cd5c43183cc25e38991d6a Mon Sep 17 00:00:00 2001
From: Wei Yang <weiyang@fb.com>
Date: Thu, 28 Jun 2018 19:40:27 -0700
Subject: [PATCH 09/18] addresses comments

---
 torch/csrc/utils/python_arg_parser.cpp | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/torch/csrc/utils/python_arg_parser.cpp b/torch/csrc/utils/python_arg_parser.cpp
index c6f3fe85fa4f5..b1ae6804d029d 100644
--- a/torch/csrc/utils/python_arg_parser.cpp
+++ b/torch/csrc/utils/python_arg_parser.cpp
@@ -174,7 +174,7 @@ static inline std::vector<int64_t> parse_intlist_args(const std::string& s, int6
   }
 
   size_t n = s.size();
-  AT_CHECK(s[0] == '{', "Default value of IntList is missing left brace '{', found ", s[0]);
+  // since already checked left brace '{' above, here only checks right brace '}'
   AT_CHECK(s[n - 1] == '}', "Default value of IntList is missing right brace '}', found ", s[n - 1]);
 
   auto args = std::vector<int64_t>();

From 3fcb2c138bc66c5b641a1f9870477461301c2609 Mon Sep 17 00:00:00 2001
From: Wei Yang <weiyang@fb.com>
Date: Fri, 29 Jun 2018 09:38:45 -0700
Subject: [PATCH 10/18] 1. disallow dims to < -total_dim in wrap_dims(); 2.
 return empty vector for x={} default args

---
 aten/src/ATen/WrapDimUtils.h           | 1 +
 torch/csrc/utils/python_arg_parser.cpp | 5 +++++
 2 files changed, 6 insertions(+)

diff --git a/aten/src/ATen/WrapDimUtils.h b/aten/src/ATen/WrapDimUtils.h
index 35feba7410627..9a3e5d3fbf9d3 100644
--- a/aten/src/ATen/WrapDimUtils.h
+++ b/aten/src/ATen/WrapDimUtils.h
@@ -90,6 +90,7 @@ static inline int64_t legacy_cat_wrap_dim(int64_t dim, TensorList tensors) {
 static inline void wrap_dims(std::vector<int64_t>& to_transform_dims, int64_t tensor_total_dims) {
   for (size_t i = 0; i < to_transform_dims.size(); i++) {
     if (to_transform_dims[i] < 0) {
+      AT_CHECK(to_transform_dims[i] >= -tensor_total_dims, "to_transform_dims = ", to_transform_dims[i], " less than lower bound ", -tensor_total_dims);
       to_transform_dims[i] = (tensor_total_dims + (to_transform_dims[i] % tensor_total_dims)) % tensor_total_dims;
     }
   }
diff --git a/torch/csrc/utils/python_arg_parser.cpp b/torch/csrc/utils/python_arg_parser.cpp
index b1ae6804d029d..0f3d75238d9cc 100644
--- a/torch/csrc/utils/python_arg_parser.cpp
+++ b/torch/csrc/utils/python_arg_parser.cpp
@@ -178,6 +178,11 @@ static inline std::vector<int64_t> parse_intlist_args(const std::string& s, int6
   AT_CHECK(s[n - 1] == '}', "Default value of IntList is missing right brace '}', found ", s[n - 1]);
 
   auto args = std::vector<int64_t>();
+  // for case IntList x={}, return an empty vector
+  if (s.size() == 2) {
+    return args;
+  }
+
   int64_t x = 0, sign = 1;
   for (size_t i = 1; i < n - 1; i++) {
     if (s[i] == '-') {

From d7a9a6ee0e94212d261a7c615af5723963300d6a Mon Sep 17 00:00:00 2001
From: Wei Yang <weiyang@fb.com>
Date: Mon, 2 Jul 2018 10:12:43 -0700
Subject: [PATCH 11/18] [wip] use strtol in parse_intlist_args

---
 aten/src/ATen/WrapDimUtils.h                  |  9 +--
 .../src/ATen/native/TensorTransformations.cpp |  9 ++-
 aten/src/ATen/native/TensorTransformations.h  |  2 +-
 .../ATen/native/cuda/TensorTransformations.cu |  2 +-
 aten/src/ATen/native/native_functions.yaml    |  2 +-
 torch/csrc/utils/python_arg_parser.cpp        | 68 +++++++++++++++----
 6 files changed, 66 insertions(+), 26 deletions(-)

diff --git a/aten/src/ATen/WrapDimUtils.h b/aten/src/ATen/WrapDimUtils.h
index 9a3e5d3fbf9d3..aa4f7c158bd1d 100644
--- a/aten/src/ATen/WrapDimUtils.h
+++ b/aten/src/ATen/WrapDimUtils.h
@@ -87,12 +87,9 @@ static inline int64_t legacy_cat_wrap_dim(int64_t dim, TensorList tensors) {
 }
 
 // wrap negative dims in to_transform_dims
-static inline void wrap_dims(std::vector<int64_t>& to_transform_dims, int64_t tensor_total_dims) {
-  for (size_t i = 0; i < to_transform_dims.size(); i++) {
-    if (to_transform_dims[i] < 0) {
-      AT_CHECK(to_transform_dims[i] >= -tensor_total_dims, "to_transform_dims = ", to_transform_dims[i], " less than lower bound ", -tensor_total_dims);
-      to_transform_dims[i] = (tensor_total_dims + (to_transform_dims[i] % tensor_total_dims)) % tensor_total_dims;
-    }
+static inline void wrap_all_dims(std::vector<int64_t>& dims_to_wrap, int64_t tensor_total_dims) {
+  for (size_t i = 0; i < dims_to_wrap.size(); i++) {
+    dims_to_wrap[i] = maybe_wrap_dim(dims_to_wrap[i], tensor_total_dims) % tensor_total_dims;
   }
 }
 
diff --git a/aten/src/ATen/native/TensorTransformations.cpp b/aten/src/ATen/native/TensorTransformations.cpp
index 6431a45bcba1a..31359fdda65d9 100644
--- a/aten/src/ATen/native/TensorTransformations.cpp
+++ b/aten/src/ATen/native/TensorTransformations.cpp
@@ -14,7 +14,7 @@ Tensor flip_cpu(const Tensor& self, IntList dims) {
   flip_check_errors(total_dims, flip_dims_size, dims);
 
   auto flip_dims_v = std::vector<int64_t>(dims);
-  wrap_dims(flip_dims_v, total_dims);
+  wrap_all_dims(flip_dims_v, total_dims);
   std::sort(flip_dims_v.begin(), flip_dims_v.end());
   auto final_indices = std::vector<at::Tensor>(total_dims);
 
@@ -58,7 +58,12 @@ Tensor flip_cpu(const Tensor& self, IntList dims) {
 }
 
 Tensor rot90(const Tensor& self, int64_t k, IntList dims) {
-  const int64_t total_dims = self.dim(), total_rot_dims = dims.size();
+  const int64_t total_dims = self.dim();
+  int64_t total_rot_dims = dims.size();
+  if (total_rot_dims == 0) {
+    dims = IntList({0,1});
+    total_rot_dims = 2;
+  }
 
   AT_CHECK(total_rot_dims == 2,
     "expected total rotation dims == 2, but got dims = ", total_rot_dims);
diff --git a/aten/src/ATen/native/TensorTransformations.h b/aten/src/ATen/native/TensorTransformations.h
index e000b13f6a379..2504a2c3f201b 100644
--- a/aten/src/ATen/native/TensorTransformations.h
+++ b/aten/src/ATen/native/TensorTransformations.h
@@ -26,7 +26,7 @@ static inline void flip_check_errors(int64_t total_dims, int64_t flip_dims_size,
     "The max flip dims out of range, got max flip dims=", *min_max_d.second);
 
   // check duplicates in dims
-  wrap_dims(flip_dims_v, total_dims);
+  wrap_all_dims(flip_dims_v, total_dims);
   flip_dims_v.erase(std::unique(flip_dims_v.begin(), flip_dims_v.end()), flip_dims_v.end());
   AT_CHECK((int64_t)flip_dims_v.size() == flip_dims_size,
     "dims has duplicates, original flip dims size=", flip_dims_size,
diff --git a/aten/src/ATen/native/cuda/TensorTransformations.cu b/aten/src/ATen/native/cuda/TensorTransformations.cu
index 800b5a07a662e..7fa1fe64f28d6 100644
--- a/aten/src/ATen/native/cuda/TensorTransformations.cu
+++ b/aten/src/ATen/native/cuda/TensorTransformations.cu
@@ -81,7 +81,7 @@ Tensor flip_cuda(const Tensor& self, IntList dims) {
   }
 
   auto flip_dims = std::vector<int64_t>(dims);
-  wrap_dims(flip_dims, total_dims);
+  wrap_all_dims(flip_dims, total_dims);
 
   // use kernel_pointwise_flip_apply2 only when to-flip dim is the 1st or last dim, where collapseDims can reduce the amount of work
   if (flip_dims_size == 1 && in_tensor.is_contiguous() && (flip_dims[0] == 0 || flip_dims[0] == total_dims - 1)) {
diff --git a/aten/src/ATen/native/native_functions.yaml b/aten/src/ATen/native/native_functions.yaml
index a326d56f4aa79..f8fbe0071c7df 100644
--- a/aten/src/ATen/native/native_functions.yaml
+++ b/aten/src/ATen/native/native_functions.yaml
@@ -1446,7 +1446,7 @@
     CUDA: flip_cuda
 
 # default IntList value {0,1} should not add space after comma, since native_parse.py uses ', ' to split args
-- func: rot90(Tensor self, int64_t k=1, IntList dims={0,1}) -> Tensor
+- func: rot90(Tensor self, int64_t k=1, IntList dims={}) -> Tensor
 
 - func: _trilinear(Tensor i1, Tensor i2, Tensor i3, IntList expand1, IntList expand2, IntList expand3, IntList sumdim, int64_t unroll_dim=1) -> Tensor
   variants: function
diff --git a/torch/csrc/utils/python_arg_parser.cpp b/torch/csrc/utils/python_arg_parser.cpp
index 0f3d75238d9cc..8bf67151b4f17 100644
--- a/torch/csrc/utils/python_arg_parser.cpp
+++ b/torch/csrc/utils/python_arg_parser.cpp
@@ -168,30 +168,68 @@ There are two kinds of default values:
 1. IntList[2] x=1 (where size=2, value={1,1}
 2. IntList x={1,2,3} (where size=3, value={1,2,3}, note that there cannot be space after comma since native_parse.py uses ', ' to split args)
 */
+// static inline std::vector<int64_t> parse_intlist_args(const std::string& s, int64_t size) {
+//   printf("s = %s\n", s.c_str());
+//   if (s[0] != '{') {
+//     return std::vector<int64_t>(size, std::stol(s));
+//   }
+//
+//   size_t n = s.size();
+//   // since already checked left brace '{' above, here only checks right brace '}'
+//   AT_CHECK(s[n - 1] == '}', "Default value of IntList is missing right brace '}', found ", s[n - 1]);
+//
+//   auto args = std::vector<int64_t>();
+//   // for case IntList x={}, return an empty vector
+//   if (s.size() == 2) {
+//     return args;
+//   }
+//
+//   int64_t x = 0, sign = 1;
+//   for (size_t i = 1; i < n - 1; i++) {
+//     if (s[i] == '-') {
+//       sign *= -1;
+//     }
+//     else if ('0' <= s[i] && s[i] <= '9') {
+//       x = x * 10 + (s[i] - '0');
+//     }
+//     else if (s[i] == ',') {
+//       args.emplace_back(sign * x);
+//       sign = 1;
+//       x = 0;
+//     }
+//     else {
+//       AT_ERROR("Illegal char in IntList default value: ", s[i]);
+//     }
+//   }
+//   args.emplace_back(sign * x);
+//   return args;
+// }
+
 static inline std::vector<int64_t> parse_intlist_args(const std::string& s, int64_t size) {
+  printf("s = %s\n", s.c_str());
+  char *str_end;
+  int64_t x = 0;
+  size_t n = s.size();
+  auto args = std::vector<int64_t>();
+
+  if (s.empty()) return args;
+
   if (s[0] != '{') {
-    return std::vector<int64_t>(size, std::stoi(s));
+    x = strtol(s.c_str(), &str_end, 10);
+    if (*str_end == 0) {
+      args = std::vector<int64_t>(size, x);
+    }
+    return args;
   }
 
-  size_t n = s.size();
   // since already checked left brace '{' above, here only checks right brace '}'
   AT_CHECK(s[n - 1] == '}', "Default value of IntList is missing right brace '}', found ", s[n - 1]);
 
-  auto args = std::vector<int64_t>();
-  // for case IntList x={}, return an empty vector
-  if (s.size() == 2) {
-    return args;
-  }
-
-  int64_t x = 0, sign = 1;
+  int64_t s = 1, e = 1;
   for (size_t i = 1; i < n - 1; i++) {
-    if (s[i] == '-') {
-      sign *= -1;
-    }
-    else if ('0' <= s[i] && s[i] <= '9') {
-      x = x * 10 + (s[i] - '0');
-    }
     else if (s[i] == ',') {
+      e = i;
+      auto x = std::atol(s.substr(s, e - s));
       args.emplace_back(sign * x);
       sign = 1;
       x = 0;

From 03a27cbd258e4d3d117730a6c420698d018a3e46 Mon Sep 17 00:00:00 2001
From: Wei Yang <weiyang@fb.com>
Date: Mon, 2 Jul 2018 20:00:28 -0700
Subject: [PATCH 12/18] addresses comments

---
 .../src/ATen/native/TensorTransformations.cpp |  7 +-
 aten/src/ATen/native/native_functions.yaml    |  2 +-
 test/test_torch.py                            |  4 ++
 torch/csrc/utils/python_arg_parser.cpp        | 72 ++++---------------
 4 files changed, 19 insertions(+), 66 deletions(-)

diff --git a/aten/src/ATen/native/TensorTransformations.cpp b/aten/src/ATen/native/TensorTransformations.cpp
index 31359fdda65d9..9653abef26f63 100644
--- a/aten/src/ATen/native/TensorTransformations.cpp
+++ b/aten/src/ATen/native/TensorTransformations.cpp
@@ -58,12 +58,7 @@ Tensor flip_cpu(const Tensor& self, IntList dims) {
 }
 
 Tensor rot90(const Tensor& self, int64_t k, IntList dims) {
-  const int64_t total_dims = self.dim();
-  int64_t total_rot_dims = dims.size();
-  if (total_rot_dims == 0) {
-    dims = IntList({0,1});
-    total_rot_dims = 2;
-  }
+  const int64_t total_dims = self.dim(), total_rot_dims = dims.size();
 
   AT_CHECK(total_rot_dims == 2,
     "expected total rotation dims == 2, but got dims = ", total_rot_dims);
diff --git a/aten/src/ATen/native/native_functions.yaml b/aten/src/ATen/native/native_functions.yaml
index f8fbe0071c7df..a326d56f4aa79 100644
--- a/aten/src/ATen/native/native_functions.yaml
+++ b/aten/src/ATen/native/native_functions.yaml
@@ -1446,7 +1446,7 @@
     CUDA: flip_cuda
 
 # default IntList value {0,1} should not add space after comma, since native_parse.py uses ', ' to split args
-- func: rot90(Tensor self, int64_t k=1, IntList dims={}) -> Tensor
+- func: rot90(Tensor self, int64_t k=1, IntList dims={0,1}) -> Tensor
 
 - func: _trilinear(Tensor i1, Tensor i2, Tensor i3, IntList expand1, IntList expand2, IntList expand3, IntList sumdim, int64_t unroll_dim=1) -> Tensor
   variants: function
diff --git a/test/test_torch.py b/test/test_torch.py
index 5595035507c01..2cf8633f5f9a9 100644
--- a/test/test_torch.py
+++ b/test/test_torch.py
@@ -6755,6 +6755,10 @@ def _test_rot90(self, use_cuda=False):
         self.assertEqual(data.rot90(3, [0, 1]), data.rot90(-1, [0, 1]))
         self.assertEqual(data.rot90(-5, [0, 1]), data.rot90(-1, [0, 1]))
 
+        # test for dims out-of-range error
+        self.assertRaises(RuntimeError, lambda: data.rot90(1, [0, -3]))
+        self.assertRaises(RuntimeError, lambda: data.rot90(1, [0, 2]))
+
         # test tensor with more than 2D
         data = torch.arange(1, 9, device=device).view(2, 2, 2)
         self.assertEqual(torch.tensor([2, 4, 1, 3, 6, 8, 5, 7]).view(2, 2, 2), data.rot90(1, [1, 2]))
diff --git a/torch/csrc/utils/python_arg_parser.cpp b/torch/csrc/utils/python_arg_parser.cpp
index 8bf67151b4f17..23dd65618c86e 100644
--- a/torch/csrc/utils/python_arg_parser.cpp
+++ b/torch/csrc/utils/python_arg_parser.cpp
@@ -168,77 +168,31 @@ There are two kinds of default values:
 1. IntList[2] x=1 (where size=2, value={1,1}
 2. IntList x={1,2,3} (where size=3, value={1,2,3}, note that there cannot be space after comma since native_parse.py uses ', ' to split args)
 */
-// static inline std::vector<int64_t> parse_intlist_args(const std::string& s, int64_t size) {
-//   printf("s = %s\n", s.c_str());
-//   if (s[0] != '{') {
-//     return std::vector<int64_t>(size, std::stol(s));
-//   }
-//
-//   size_t n = s.size();
-//   // since already checked left brace '{' above, here only checks right brace '}'
-//   AT_CHECK(s[n - 1] == '}', "Default value of IntList is missing right brace '}', found ", s[n - 1]);
-//
-//   auto args = std::vector<int64_t>();
-//   // for case IntList x={}, return an empty vector
-//   if (s.size() == 2) {
-//     return args;
-//   }
-//
-//   int64_t x = 0, sign = 1;
-//   for (size_t i = 1; i < n - 1; i++) {
-//     if (s[i] == '-') {
-//       sign *= -1;
-//     }
-//     else if ('0' <= s[i] && s[i] <= '9') {
-//       x = x * 10 + (s[i] - '0');
-//     }
-//     else if (s[i] == ',') {
-//       args.emplace_back(sign * x);
-//       sign = 1;
-//       x = 0;
-//     }
-//     else {
-//       AT_ERROR("Illegal char in IntList default value: ", s[i]);
-//     }
-//   }
-//   args.emplace_back(sign * x);
-//   return args;
-// }
-
 static inline std::vector<int64_t> parse_intlist_args(const std::string& s, int64_t size) {
   printf("s = %s\n", s.c_str());
-  char *str_end;
-  int64_t x = 0;
   size_t n = s.size();
-  auto args = std::vector<int64_t>();
 
-  if (s.empty()) return args;
+  // case 1. s = ""
+  if (s.empty()) return std::vector<int64_t>();
 
+  // case 2. s is an int (e.g., s = "12")
   if (s[0] != '{') {
-    x = strtol(s.c_str(), &str_end, 10);
-    if (*str_end == 0) {
-      args = std::vector<int64_t>(size, x);
-    }
-    return args;
+    return std::vector<int64_t>(size, std::stol(s));
   }
 
+  // case 3. s is a list of dims (e.g., s = {1,2})
+
   // since already checked left brace '{' above, here only checks right brace '}'
   AT_CHECK(s[n - 1] == '}', "Default value of IntList is missing right brace '}', found ", s[n - 1]);
 
-  int64_t s = 1, e = 1;
-  for (size_t i = 1; i < n - 1; i++) {
-    else if (s[i] == ',') {
-      e = i;
-      auto x = std::atol(s.substr(s, e - s));
-      args.emplace_back(sign * x);
-      sign = 1;
-      x = 0;
-    }
-    else {
-      AT_ERROR("Illegal char in IntList default value: ", s[i]);
-    }
+  auto args = std::vector<int64_t>();
+  const char* del = ",{}";
+  char* s_p = strdup(s.c_str());
+  char *token = strtok(s_p, del);
+  while (token != NULL) {
+    args.emplace_back(std::atol(token));
+    token = strtok(NULL, del);
   }
-  args.emplace_back(sign * x);
   return args;
 }
 

From 7686d974825ff7c9f46730beabd3e638303e84cb Mon Sep 17 00:00:00 2001
From: Wei Yang <weiyang@fb.com>
Date: Mon, 2 Jul 2018 20:10:26 -0700
Subject: [PATCH 13/18] nits

---
 torch/csrc/utils/python_arg_parser.cpp | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

diff --git a/torch/csrc/utils/python_arg_parser.cpp b/torch/csrc/utils/python_arg_parser.cpp
index 23dd65618c86e..cea5ff019fff0 100644
--- a/torch/csrc/utils/python_arg_parser.cpp
+++ b/torch/csrc/utils/python_arg_parser.cpp
@@ -169,18 +169,16 @@ There are two kinds of default values:
 2. IntList x={1,2,3} (where size=3, value={1,2,3}, note that there cannot be space after comma since native_parse.py uses ', ' to split args)
 */
 static inline std::vector<int64_t> parse_intlist_args(const std::string& s, int64_t size) {
-  printf("s = %s\n", s.c_str());
   size_t n = s.size();
 
-  // case 1. s = ""
   if (s.empty()) return std::vector<int64_t>();
 
-  // case 2. s is an int (e.g., s = "12")
+  // case 1. s is an int (e.g., s=2)
   if (s[0] != '{') {
     return std::vector<int64_t>(size, std::stol(s));
   }
 
-  // case 3. s is a list of dims (e.g., s = {1,2})
+  // case 2. s is a list of dims (e.g., s={1,2})
 
   // since already checked left brace '{' above, here only checks right brace '}'
   AT_CHECK(s[n - 1] == '}', "Default value of IntList is missing right brace '}', found ", s[n - 1]);

From e5c46bf29360dc19a517951f8eda50bc64ed5982 Mon Sep 17 00:00:00 2001
From: Wei Yang <weiyang@fb.com>
Date: Wed, 18 Jul 2018 00:03:26 -0700
Subject: [PATCH 14/18] Using istringstream rather than strtok, addressed
 comments

---
 aten/src/ATen/WrapDimUtils.h           |  4 ++--
 torch/csrc/utils/python_arg_parser.cpp | 11 +++++------
 2 files changed, 7 insertions(+), 8 deletions(-)

diff --git a/aten/src/ATen/WrapDimUtils.h b/aten/src/ATen/WrapDimUtils.h
index aa4f7c158bd1d..09d4e2a53fae1 100644
--- a/aten/src/ATen/WrapDimUtils.h
+++ b/aten/src/ATen/WrapDimUtils.h
@@ -86,10 +86,10 @@ static inline int64_t legacy_cat_wrap_dim(int64_t dim, TensorList tensors) {
   return dim;
 }
 
-// wrap negative dims in to_transform_dims
+// wrap negative dims in a vector
 static inline void wrap_all_dims(std::vector<int64_t>& dims_to_wrap, int64_t tensor_total_dims) {
   for (size_t i = 0; i < dims_to_wrap.size(); i++) {
-    dims_to_wrap[i] = maybe_wrap_dim(dims_to_wrap[i], tensor_total_dims) % tensor_total_dims;
+    dims_to_wrap[i] = maybe_wrap_dim(dims_to_wrap[i], tensor_total_dims);
   }
 }
 
diff --git a/torch/csrc/utils/python_arg_parser.cpp b/torch/csrc/utils/python_arg_parser.cpp
index cea5ff019fff0..019ea9fda5dbe 100644
--- a/torch/csrc/utils/python_arg_parser.cpp
+++ b/torch/csrc/utils/python_arg_parser.cpp
@@ -184,12 +184,11 @@ static inline std::vector<int64_t> parse_intlist_args(const std::string& s, int6
   AT_CHECK(s[n - 1] == '}', "Default value of IntList is missing right brace '}', found ", s[n - 1]);
 
   auto args = std::vector<int64_t>();
-  const char* del = ",{}";
-  char* s_p = strdup(s.c_str());
-  char *token = strtok(s_p, del);
-  while (token != NULL) {
-    args.emplace_back(std::atol(token));
-    token = strtok(NULL, del);
+  std::istringstream ss(s.substr(1, s.length() - 2)); // exclude '{' and '}'
+  std::string tok;
+
+  while(std::getline(ss, tok, ',')) {
+    args.emplace_back(std::stol(tok));
   }
   return args;
 }

From fd1eff5bf5bdd71612d7125b36112c37a014abdf Mon Sep 17 00:00:00 2001
From: Wei Yang <weiyang@fb.com>
Date: Mon, 23 Jul 2018 19:49:16 -0700
Subject: [PATCH 15/18] 1. modified jit default args parser; 2. added test for
 flip() on empty tensor

---
 test/test_torch.py            | 4 ++++
 tools/jit/gen_jit_dispatch.py | 4 +++-
 2 files changed, 7 insertions(+), 1 deletion(-)

diff --git a/test/test_torch.py b/test/test_torch.py
index 2cf8633f5f9a9..6a3479c000920 100644
--- a/test/test_torch.py
+++ b/test/test_torch.py
@@ -6722,6 +6722,10 @@ def _test_flip(self, use_cuda=False):
         self.assertEqual(flip0_result, data.flip(0))
         self.assertEqual(flip1_result, data.flip(1))
 
+        # test empty tensor, should just return an empty tensor of the same shape
+        data = torch.tensor([])
+        self.assertEqual(data, data.flip(0))
+
     def test_flip(self):
         self._test_flip(self, use_cuda=False)
 
diff --git a/tools/jit/gen_jit_dispatch.py b/tools/jit/gen_jit_dispatch.py
index 49948de7180a0..18c043a6c1061 100644
--- a/tools/jit/gen_jit_dispatch.py
+++ b/tools/jit/gen_jit_dispatch.py
@@ -419,7 +419,9 @@ def format_arg(arg):
                 .replace('false', 'False') \
                 .replace('nullptr', 'None') \
                 .replace('Reduction::ElementwiseMean', 'ElementwiseMean') \
-                .replace('{}', 'None' if is_tensor_arg(arg) else '[]')
+                .replace('{}', 'None' if is_tensor_arg(arg) else '[]') \
+                .replace('{', '[') \
+                .replace('}', ']')
 
             default = default_map.get(default, default)
             decl = '{}={}'.format(decl, default)

From 64fd306237ac0132221d01d5840d5b70b82d5f25 Mon Sep 17 00:00:00 2001
From: Wei Yang <weiyang@fb.com>
Date: Tue, 24 Jul 2018 15:10:46 -0700
Subject: [PATCH 16/18] jit debug

---
 aten/src/ATen/native/TensorTransformations.cpp | 9 +++++++++
 aten/src/ATen/native/native_functions.yaml     | 2 +-
 test/test_jit.py                               | 2 ++
 tools/jit/gen_jit_dispatch.py                  | 3 +++
 4 files changed, 15 insertions(+), 1 deletion(-)

diff --git a/aten/src/ATen/native/TensorTransformations.cpp b/aten/src/ATen/native/TensorTransformations.cpp
index 9653abef26f63..e084c03d0e9c7 100644
--- a/aten/src/ATen/native/TensorTransformations.cpp
+++ b/aten/src/ATen/native/TensorTransformations.cpp
@@ -60,9 +60,18 @@ Tensor flip_cpu(const Tensor& self, IntList dims) {
 Tensor rot90(const Tensor& self, int64_t k, IntList dims) {
   const int64_t total_dims = self.dim(), total_rot_dims = dims.size();
 
+  printf("total_rot_dims %ld\n", total_rot_dims);
+
+  for (int64_t i = 0; i < total_rot_dims; i++) {
+    printf("dim[%ld] = %ld\n", i, dims[i]);
+  }
+
   AT_CHECK(total_rot_dims == 2,
     "expected total rotation dims == 2, but got dims = ", total_rot_dims);
 
+  AT_CHECK(total_dims >= 2,
+    "expected total dims >= 2, but got total dims = ", total_dims);
+
   AT_CHECK(dims[0] != dims[1] && std::abs(dims[0] - dims[1]) != total_dims,
     "expected rotation dims to be different, but got dim0 = ", dims[0],
     " and dim1 = ", dims[1]);
diff --git a/aten/src/ATen/native/native_functions.yaml b/aten/src/ATen/native/native_functions.yaml
index a326d56f4aa79..1bd0dbf4a0b28 100644
--- a/aten/src/ATen/native/native_functions.yaml
+++ b/aten/src/ATen/native/native_functions.yaml
@@ -1446,7 +1446,7 @@
     CUDA: flip_cuda
 
 # default IntList value {0,1} should not add space after comma, since native_parse.py uses ', ' to split args
-- func: rot90(Tensor self, int64_t k=1, IntList dims={0,1}) -> Tensor
+- func: rot90(Tensor self, int64_t k=1, IntList[2] dims=0) -> Tensor
 
 - func: _trilinear(Tensor i1, Tensor i2, Tensor i3, IntList expand1, IntList expand2, IntList expand3, IntList sumdim, int64_t unroll_dim=1) -> Tensor
   variants: function
diff --git a/test/test_jit.py b/test/test_jit.py
index bd04f63729b28..b5b26f69a28df 100644
--- a/test/test_jit.py
+++ b/test/test_jit.py
@@ -4861,7 +4861,9 @@ def script_fn(*args, **kwargs):
         else:
             call = '{}.{}({}{})'.format(actuals[0], method_name, ', '.join(actuals[1:]), kwargs_str)
         script = script_template.format(', '.join(formals), call)
+        # print("script: ", script);
         CU = torch.jit.CompilationUnit(script)
+        # print("create_script_fn: tensors = %s" % tensors)
         return output_process_fn(CU.the_method(*tensors))
     return script_fn
 
diff --git a/tools/jit/gen_jit_dispatch.py b/tools/jit/gen_jit_dispatch.py
index 18c043a6c1061..b985855b008a0 100644
--- a/tools/jit/gen_jit_dispatch.py
+++ b/tools/jit/gen_jit_dispatch.py
@@ -425,6 +425,7 @@ def format_arg(arg):
 
             default = default_map.get(default, default)
             decl = '{}={}'.format(decl, default)
+            # print("decl", decl)
         return decl
 
     args = []
@@ -440,6 +441,8 @@ def format_arg(arg):
         ret_list = jit_type_of(decl['returns'][0])
     else:
         ret_list = '({})'.format(', '.join(jit_type_of(r) for r in decl['returns']))
+
+    print("signature return: %s" % 'aten::{}({}) -> {}'.format(decl['name'], arg_list, ret_list))
     return 'aten::{}({}) -> {}'.format(decl['name'], arg_list, ret_list)
 
 

From 427d4b96043e6c36d7ba8896530a83bf5c865d3b Mon Sep 17 00:00:00 2001
From: Wei Yang <weiyang@fb.com>
Date: Tue, 24 Jul 2018 16:57:03 -0700
Subject: [PATCH 17/18] [wip] debugging jit

---
 aten/src/ATen/native/native_functions.yaml | 2 +-
 test/test_jit.py                           | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/aten/src/ATen/native/native_functions.yaml b/aten/src/ATen/native/native_functions.yaml
index 1bd0dbf4a0b28..a326d56f4aa79 100644
--- a/aten/src/ATen/native/native_functions.yaml
+++ b/aten/src/ATen/native/native_functions.yaml
@@ -1446,7 +1446,7 @@
     CUDA: flip_cuda
 
 # default IntList value {0,1} should not add space after comma, since native_parse.py uses ', ' to split args
-- func: rot90(Tensor self, int64_t k=1, IntList[2] dims=0) -> Tensor
+- func: rot90(Tensor self, int64_t k=1, IntList dims={0,1}) -> Tensor
 
 - func: _trilinear(Tensor i1, Tensor i2, Tensor i3, IntList expand1, IntList expand2, IntList expand3, IntList sumdim, int64_t unroll_dim=1) -> Tensor
   variants: function
diff --git a/test/test_jit.py b/test/test_jit.py
index b5b26f69a28df..25d262b6c8ad1 100644
--- a/test/test_jit.py
+++ b/test/test_jit.py
@@ -4863,7 +4863,7 @@ def script_fn(*args, **kwargs):
         script = script_template.format(', '.join(formals), call)
         # print("script: ", script);
         CU = torch.jit.CompilationUnit(script)
-        # print("create_script_fn: tensors = %s" % tensors)
+        print(CU.the_method.graph)
         return output_process_fn(CU.the_method(*tensors))
     return script_fn
 

From 0a5d9e04bf57a9a4e5c91ee9f5c0bb1b44802d28 Mon Sep 17 00:00:00 2001
From: Wei Yang <weiyang@fb.com>
Date: Tue, 24 Jul 2018 17:29:56 -0700
Subject: [PATCH 18/18] Zachary DeVito fixed the bug in jit

---
 aten/src/ATen/native/TensorTransformations.cpp | 6 ------
 test/test_jit.py                               | 2 --
 test/test_torch.py                             | 1 +
 tools/jit/gen_jit_dispatch.py                  | 3 ---
 torch/csrc/jit/script/compiler.cpp             | 2 +-
 5 files changed, 2 insertions(+), 12 deletions(-)

diff --git a/aten/src/ATen/native/TensorTransformations.cpp b/aten/src/ATen/native/TensorTransformations.cpp
index e084c03d0e9c7..84759874ef535 100644
--- a/aten/src/ATen/native/TensorTransformations.cpp
+++ b/aten/src/ATen/native/TensorTransformations.cpp
@@ -60,12 +60,6 @@ Tensor flip_cpu(const Tensor& self, IntList dims) {
 Tensor rot90(const Tensor& self, int64_t k, IntList dims) {
   const int64_t total_dims = self.dim(), total_rot_dims = dims.size();
 
-  printf("total_rot_dims %ld\n", total_rot_dims);
-
-  for (int64_t i = 0; i < total_rot_dims; i++) {
-    printf("dim[%ld] = %ld\n", i, dims[i]);
-  }
-
   AT_CHECK(total_rot_dims == 2,
     "expected total rotation dims == 2, but got dims = ", total_rot_dims);
 
diff --git a/test/test_jit.py b/test/test_jit.py
index 25d262b6c8ad1..bd04f63729b28 100644
--- a/test/test_jit.py
+++ b/test/test_jit.py
@@ -4861,9 +4861,7 @@ def script_fn(*args, **kwargs):
         else:
             call = '{}.{}({}{})'.format(actuals[0], method_name, ', '.join(actuals[1:]), kwargs_str)
         script = script_template.format(', '.join(formals), call)
-        # print("script: ", script);
         CU = torch.jit.CompilationUnit(script)
-        print(CU.the_method.graph)
         return output_process_fn(CU.the_method(*tensors))
     return script_fn
 
diff --git a/test/test_torch.py b/test/test_torch.py
index 6a3479c000920..98e48d474fb6f 100644
--- a/test/test_torch.py
+++ b/test/test_torch.py
@@ -6772,6 +6772,7 @@ def _test_rot90(self, use_cuda=False):
         self.assertRaises(RuntimeError, lambda: data.rot90(1, [0, 3]))
         self.assertRaises(RuntimeError, lambda: data.rot90(1, [1, 1]))
         self.assertRaises(RuntimeError, lambda: data.rot90(1, [0, 1, 2]))
+        self.assertRaises(RuntimeError, lambda: data.rot90(1, [0]))
 
     def test_rot90(self):
         self._test_rot90(self, use_cuda=False)
diff --git a/tools/jit/gen_jit_dispatch.py b/tools/jit/gen_jit_dispatch.py
index b985855b008a0..18c043a6c1061 100644
--- a/tools/jit/gen_jit_dispatch.py
+++ b/tools/jit/gen_jit_dispatch.py
@@ -425,7 +425,6 @@ def format_arg(arg):
 
             default = default_map.get(default, default)
             decl = '{}={}'.format(decl, default)
-            # print("decl", decl)
         return decl
 
     args = []
@@ -441,8 +440,6 @@ def format_arg(arg):
         ret_list = jit_type_of(decl['returns'][0])
     else:
         ret_list = '({})'.format(', '.join(jit_type_of(r) for r in decl['returns']))
-
-    print("signature return: %s" % 'aten::{}({}) -> {}'.format(decl['name'], arg_list, ret_list))
     return 'aten::{}({}) -> {}'.format(decl['name'], arg_list, ret_list)
 
 
diff --git a/torch/csrc/jit/script/compiler.cpp b/torch/csrc/jit/script/compiler.cpp
index b12bb7b58d016..c38eeadb33737 100644
--- a/torch/csrc/jit/script/compiler.cpp
+++ b/torch/csrc/jit/script/compiler.cpp
@@ -373,7 +373,7 @@ static bool isTensorSubtype(Value* v) {
 at::optional<std::vector<int64_t>> getIntListAttribute(at::optional<int32_t> N, Value* input) {
   auto list = constant_as<Shared<jit::IntList>>(input);
   if(list)
-    return std::vector<int64_t>(*list);
+    return std::vector<int64_t>(list.value()->elements());
   // broadcast IntList[3] with value 4 -> {4, 4, 4}
   if(!N)
     return at::nullopt;