Add nested squeeze.dim and unsqueeze (#86813)

Pull Request resolved: https://github.com/pytorch/pytorch/pull/86813
Approved by: https://github.com/drisspg

aten/src/ATen/native/native_functions.yaml

@@ -4893,6 +4893,7 @@
   dispatch:
     CompositeExplicitAutograd: squeeze
     QuantizedCPU, QuantizedCUDA: squeeze_quantized
+    NestedTensorCPU, NestedTensorCUDA: squeeze_nested
 
 - func: squeeze.dim(Tensor(a) self, int dim) -> Tensor(a)
   variants: function, method
@@ -4901,6 +4902,7 @@
   dispatch:
     CompositeExplicitAutograd: squeeze
     QuantizedCPU, QuantizedCUDA: squeeze_quantized
+    NestedTensorCPU, NestedTensorCUDA: squeeze_dim_nested
   tags: canonical
 
 - func: squeeze.dimname(Tensor(a) self, Dimname dim) -> Tensor(a)
@@ -5531,6 +5533,7 @@
     CompositeExplicitAutograd: unsqueeze
     SparseCPU, SparseCUDA: unsqueeze_sparse
     QuantizedCPU, QuantizedCUDA: unsqueeze_quantized
+    NestedTensorCPU, NestedTensorCUDA: unsqueeze_nested
   tags: canonical
 
 - func: unsqueeze_(Tensor(a!) self, int dim) -> Tensor(a!)

aten/src/ATen/native/nested/NestedTensorMath.cpp

@@ -1041,6 +1041,73 @@ Tensor transpose_nested(const Tensor& self, int64_t dim0, int64_t dim1) {
       self, sizemat_transposed, stridemat_transposed, std::vector<int64_t>(self_ptr->get_storage_offsets()));
 }
 
+Tensor squeeze_nested(const Tensor& self) {
+  TORCH_CHECK(false,
+  "squeeze(): For nested tensors, squeeze without the dim argument is not supported ",
+  "at the moment, however you can use squeeze(Tensor self, int dim) instead ",
+  "if you need this feature, please open an issue on github describing your use case.");
+  return self;
+}
+
+Tensor squeeze_dim_nested(const Tensor& self, int64_t dim) {
+  auto self_ptr = get_nested_tensor_impl(self);
+  int64_t ndim = self_ptr->dim();
+  int64_t wrapped_dim = at::maybe_wrap_dim(dim, ndim);
+  TORCH_CHECK(wrapped_dim > 0,
+  "squeeze(): For nested tensors, squeezing dimension 0 is not supported at the moment ",
+  "if you need this feature, please open an issue on github describing your use case.");
+  const Tensor& sizemat = self_ptr->get_nested_size_tensor();
+  const Tensor& stridemat = self_ptr->get_nested_stride_tensor();
+  // if tensor.size(dim) != 1 torch.squeeze will return the result, we do the same here
+  c10::optional<int64_t> size_dim = self_ptr->opt_size(dim);
+  if (!(size_dim.has_value() && size_dim.value() == 1)) {
+    // detach to avoid triggering throw_error_if_base_and_tensor_are_same
+    return self.detach();
+  }
+  // if ndim == 2 and we pass the above if statement we should have a
+  // nested tensor of singleton tensors
+  TORCH_CHECK(ndim != 2,
+  "squeeze(): For nested tensors, squeezing a nested tensor of singleton tensors is not ",
+  "supported at the moment, if you need this feature, please open an issue on github",
+  "describing your use case.");
+  auto column_indices = sizemat.new_empty(ndim - 2);
+  int64_t* column_indices_ptr = column_indices.data_ptr<int64_t>();
+  std::iota(column_indices_ptr, column_indices_ptr + wrapped_dim - 1, 0);
+  std::iota(column_indices_ptr + wrapped_dim - 1, column_indices_ptr + ndim - 2, wrapped_dim);
+  auto sizemat_squeezed = at::index_select(sizemat, 1, column_indices);
+  auto stridemat_squeezed = at::index_select(stridemat, 1, column_indices);
+  return create_nested_view_tensor(
+      self, sizemat_squeezed, stridemat_squeezed, std::vector<int64_t>(self_ptr->get_storage_offsets()));
+}
+
+Tensor unsqueeze_nested(const Tensor& self, int64_t dim) {
+  auto self_ptr = get_nested_tensor_impl(self);
+  int64_t ndim = self_ptr->dim();
+  int64_t wrapped_dim = at::maybe_wrap_dim(dim, ndim + 1);
+  TORCH_CHECK(wrapped_dim > 0,
+  "unsqueeze(): For nested tensors, unsqueezing dimension 0 is not supported at the moment ",
+  "if you need this feature, please open an issue on github describing your use case.");
+  const Tensor& sizemat = self_ptr->get_nested_size_tensor();
+  const Tensor& stridemat = self_ptr->get_nested_stride_tensor();
+  auto mat_dim = wrapped_dim - 1;
+  Tensor new_size = sizemat.new_ones({sizemat.size(0), 1});
+  Tensor sizemat_unsqueezed = at::cat({sizemat.slice(1, 0, mat_dim),
+                                       new_size,
+                                       sizemat.slice(1, mat_dim, ndim)}, 1);
+  Tensor new_stride;
+  if (wrapped_dim == ndim) {
+    new_stride = stridemat.new_ones({stridemat.size(0), 1});
+  } else {
+    new_stride = (stridemat.select(1, mat_dim - 1) * sizemat.select(1, mat_dim - 1)).unsqueeze(-1);
+  }
+  Tensor stridemat_unsqueezed = at::cat({stridemat.slice(1, 0, mat_dim),
+                                         new_stride,
+                                         stridemat.slice(1, mat_dim, ndim)}, 1);
+  return create_nested_view_tensor(
+      self, sizemat_unsqueezed, stridemat_unsqueezed, std::vector<int64_t>(self_ptr->get_storage_offsets()));
+}
+
+
 // utilities supporting `view_nested` and `reshape_nested`
 namespace {
 // Args:

test/test_nestedtensor.py

@@ -606,7 +606,6 @@ class TestNestedTensorDeviceType(TestCase):
         self.assertRaises(IndexError, lambda: nt[2])
         self.assertRaises(IndexError, lambda: nt[-3])
         self.assertRaises(NotImplementedError, lambda: nt[:])
-        self.assertRaises(NotImplementedError, lambda: nt[None])
         self.assertRaises(NotImplementedError, lambda: nt[...])
         # tuple of indices: only support integer in the batch dimension
         #                 + all possible indexing in the original tensor dimensions
@@ -1295,6 +1294,48 @@ class TestNestedTensorDeviceType(TestCase):
         ptT = pt.transpose(-1, -2)
         self.assertEqual(ptT, ptT_from_ntT)
 
+    @dtypes(torch.float, torch.float16, torch.double)
+    def test_squeeze_unsqueeze(self, device, dtype):
+        a = torch.arange(6).reshape(2, 3)
+        b = torch.arange(15).reshape(5, 3)
+        nt = torch.nested.nested_tensor([a, b], device=device, dtype=dtype)
+        # error case: squeeze no dimension
+        self.assertRaisesRegex(
+            RuntimeError,
+            "For nested tensors, squeeze without the dim argument",
+            lambda: nt.squeeze()
+        )
+        # error case: squeeze nested dimension
+        self.assertRaisesRegex(
+            RuntimeError,
+            "For nested tensors, squeezing dimension 0",
+            lambda: nt.squeeze(0)
+        )
+        # error case: dimension out of range
+        self.assertRaises(IndexError, lambda: nt.squeeze(3))
+        # error case: squeeze nested tensor of singleton tensors
+        c = torch.ones(1)
+        nt_singleton = torch.nested.nested_tensor([c, c], device=device, dtype=dtype)
+        self.assertRaisesRegex(
+            RuntimeError,
+            "For nested tensors, squeezing a nested tensor of singleton",
+            lambda: nt_singleton.squeeze(1)
+        )
+
+        # squeezing a dim which does not have size 1 should be a no-op
+        nt2 = nt.squeeze(-1)
+        self.assertEqual(nt, nt2)
+
+        # test cases that should work
+        for i in range(-2, 3):
+            if (i == 0):
+                continue
+            nt_unsqueezed = nt.unsqueeze(i)
+            size_idx = i if i < 0 else i - 1
+            self.assertEqual(nt_unsqueezed._nested_tensor_size()[:, size_idx], torch.ones(2, dtype=torch.long))
+            nt_squeezed = nt_unsqueezed.squeeze(i)
+            self.assertEqual(nt_squeezed, nt)
+
     @dtypes(torch.float, torch.float16, torch.double)
     def test_transpose_inference_mode_interaction(self, device, dtype):
         nt = self.random_nt(device, dtype, 4, (4, 4))
@@ -1767,6 +1808,52 @@ class TestNestedTensorAutograd(TestCase):
 
         self.assertEqual(torch.nested.to_padded_tensor(nt.grad, 0.0), pt.grad)
 
+    def test_nested_tensor_squeeze_backward(self):
+        nt = torch.nested.nested_tensor([torch.randn((2, 6, 1)), torch.randn((3, 6, 1))], requires_grad=True)
+        with torch.no_grad():
+            pt = torch.nested.to_padded_tensor(nt, 0.0).requires_grad_(True)
+
+        ynt = nt.squeeze(-1)
+        ypt = pt.squeeze(-1)
+        ynt.backward(ynt.clone())
+        ypt.backward(ypt.clone())
+
+        self.assertEqual(torch.nested.to_padded_tensor(nt.grad, 0.0), pt.grad)
+
+    def test_nested_tensor_squeeze_gradcheck(self):
+        a = torch.randn((2, 6, 1), dtype=torch.float64, requires_grad=True)
+        b = torch.randn((3, 6, 1), dtype=torch.float64, requires_grad=True)
+
+        def grad_test_func(a, b):
+            nt = torch.nested.as_nested_tensor([a, b])
+            result = nt.squeeze(-1)
+            return torch.nested.to_padded_tensor(result, 0.0)
+
+        assert torch.autograd.gradcheck(grad_test_func, inputs=(a, b), eps=1e-3)
+
+    def test_nested_tensor_unsqueeze_backward(self):
+        nt = torch.nested.nested_tensor([torch.randn((2, 6)), torch.randn((3, 6))], requires_grad=True)
+        with torch.no_grad():
+            pt = torch.nested.to_padded_tensor(nt, 0.0).requires_grad_(True)
+
+        ynt = nt.unsqueeze(2)
+        ypt = pt.unsqueeze(2)
+        ynt.backward(ynt.clone())
+        ypt.backward(ypt.clone())
+
+        self.assertEqual(torch.nested.to_padded_tensor(nt.grad, 0.0), pt.grad)
+
+    def test_nested_tensor_unsqueeze_gradcheck(self):
+        a = torch.randn((2, 6), dtype=torch.float64, requires_grad=True)
+        b = torch.randn((3, 6), dtype=torch.float64, requires_grad=True)
+
+        def grad_test_func(a, b):
+            nt = torch.nested.as_nested_tensor([a, b])
+            result = nt.unsqueeze(-1)
+            return torch.nested.to_padded_tensor(result, 0.0)
+
+        assert torch.autograd.gradcheck(grad_test_func, inputs=(a, b), eps=1e-3)
+
     def test_nested_tensor_linear(self):
 
         a = torch.randn(1, 2, requires_grad=True, dtype=torch.float64)

tools/autograd/derivatives.yaml

@@ -1497,8 +1497,12 @@
   result: auto_linear
 
 - name: squeeze.dim(Tensor(a) self, int dim) -> Tensor(a)
-  self: unsqueeze_to(grad, dim, self.sym_sizes())
-  result: auto_linear
+  dispatch:
+    Default:
+      self: unsqueeze_to(grad, dim, self.sym_sizes())
+      result: auto_linear
+    AutogradNestedTensor:
+      self: grad.unsqueeze(dim)
 
 - name: squeeze_(Tensor(a!) self) -> Tensor(a!)
   self: unsqueeze_to(grad, self.sym_sizes())