def zeros(*size, **kwargs):
r"""Return a tensor filled with zeros.
.. math:: \text{out} \leftarrow 0
Parameters
----------
size : int...
The size(s) indicating the out shape.
out : dragon.vm.torch.Tensor, optional
The output tensor.
dtype : str, optional, default='float32'
The optional data type.
device : dragon.vm.torch.device, optional
The optional device of returned tensor.
requires_grad : bool, optional, default=False
**True** to record gradient for returned tensor.
Returns
-------
dragon.vm.torch.Tensor
The output tensor.
"""
out = kwargs.get('out', utils.new_leaf(size, kwargs))
return fill(out, size, 0)
def randperm(n, out=None, dtype='int64', device=None, requires_grad=False):
"""Return a tensor with value in the permuted range.
Specify ``n`` to determine an interval :math:`[0, n)`:
```python
print(torch.randperm(4))
```
Parameters
----------
n: number
The end of interval.
out : dragon.vm.torch.Tensor, optional
The output tensor.
dtype : str, optional, default='int64'
The optional data type.
device : dragon.vm.torch.device, optional
The optional device of returned tensor.
requires_grad : bool, optional, default=False
**True** to record gradient for returned tensor.
Returns
-------
dragon.Tensor
The output tensor.
"""
out = out if out else utils.new_leaf([n], locals())
return _functions.Permutation \
.instantiate(out.device, dtype=out.dtype) \
.apply(n, out)
def eye(
n,
m=None,
out=None,
dtype='float32',
device=None,
requires_grad=False,
):
r"""Return a tensor constructed as the identity matrix.
.. math:: \text{out} \leftarrow \text{diag}(1, 1, ..., 1)
The rows and cols of matrix are determined by ``n`` and ``m``:
```python
print(torch.eye(2)) # [[1., 0.], [0., 1.]]
print(torch.eye(2, 3)) # [[1., 0., 0.], [0., 1., 0.]]
```
Parameters
----------
n : int
The number output rows.
m : int, optional
The number output cols.
out : dragon.vm.torch.Tensor, optional
The output tensor.
dtype : str, optional, default='float32'
The optional data type.
device : dragon.vm.torch.device, optional
The optional device of returned tensor.
requires_grad : bool, optional, default=False
**True** to record gradient for returned tensor.
Returns
-------
dragon.vm.torch.Tensor
The output tensor.
"""
m = n if m is None else m
out = utils.new_leaf([n, m], locals()) if out is None else out
return _functions.Eye \
.instantiate(out.device, ndim=2, dtype=out.dtype) \
.apply(out, [n, m])
def full_like(
input,
fill_value,
out=None,
dtype='int64',
device=None,
requires_grad=False,
):
"""Return a tensor filled with a scalar with size as input.
Examples:
```python
print(torch.full_like(torch.zeros(1, 2), 1)) # [[1, 1]]
```
Parameters
----------
input : dragon.vm.torch.Tensor
The tensor for indicating shape.
fill_value : number
The scalar to fill.
out : dragon.vm.torch.Tensor, optional
The output tensor.
dtype : str, optional, default='int64'
The optional data type.
device : dragon.vm.torch.device, optional
The optional device of returned tensor.
requires_grad : bool, optional, default=False
**True** to record gradient for returned tensor.
Returns
-------
dragon.vm.torch.Tensor
The output tensor.
"""
out = utils.new_leaf(input.shape, locals())
return fill_like(out, input, fill_value)
def full(
size,
fill_value,
out=None,
dtype='int64',
device=None,
requires_grad=False,
):
"""Return a tensor filled with a scalar.
Examples:
```python
print(torch.full((1, 2), 1)) # [[1, 1]]
```
Parameters
----------
size : Sequence[int]
The output shape.
fill_value : number
The scalar to fill.
out : dragon.vm.torch.Tensor, optional
The output tensor.
dtype : str, optional, default='int64'
The optional data type.
device : dragon.vm.torch.device, optional
The optional device of returned tensor.
requires_grad : bool, optional, default=False
**True** to record gradient for returned tensor.
Returns
-------
dragon.vm.torch.Tensor
The output tensor.
"""
out = out if out else utils.new_leaf(size, locals())
return fill(out, size, fill_value)
def zeros_like(input, **kwargs):
r"""Return a tensor of zeros with shape as the other.
.. math:: \text{out} \leftarrow 0
Parameters
----------
input : dragon.vm.torch.Tensor
The tensor for indicating shape.
dtype : str, optional, default='float32'
The optional data type.
device : dragon.vm.torch.device, optional
The optional device of returned tensor.
requires_grad : bool, optional, default=False
**True** to record gradient for returned tensor.
Returns
-------
dragon.vm.torch.Tensor
The output tensor.
"""
out = utils.new_leaf(input.shape, kwargs)
return fill_like(out, input, 0)
def randn(*size, **kwargs):
"""Return a tensor from the normal distribution of N(0, 1).
Parameters
----------
size : int...
The size(s) indicating the out shape.
out : dragon.vm.torch.Tensor, optional
The output tensor.
dtype : str, optional, default='float32'
The optional data type.
device : dragon.vm.torch.device, optional
The optional device of returned tensor.
requires_grad : bool, optional, default=False
**True** to record gradient for returned tensor.
Returns
-------
dragon.vm.torch.Tensor
The output tensor.
"""
out = kwargs.get('out', utils.new_leaf(size, kwargs))
return normal_fill(out, 0, 1)