…_fpn` (#455)

Closes #264 

The error encountered in the linked issues was due to a subtle
difference in type promotions when calling `torch.div` with the argument
`rounding_mode='floor'`. Specifically, if both of the two operands are
of the integer type, then the output would still be integer type. This
is different from my original implementation, which first calls
`truediv` and then `ops.floor`, which will make the output datatype
`float32`.

After fixing this issue, another error was encountered:

```
File "/home/bolin/Desktop/hidet/python/hidet/graph/frontend/torch/interpreter.py", line 70, in __call__
    return self.forward(*args)
           ^^^^^^^^^^^^^^^^^^^
  File "/home/bolin/Desktop/hidet/python/hidet/graph/frontend/torch/interpreter.py", line 237, in forward
    self._raise_exception(e, node.target, exec_func, hidet_args, hidet_kwargs)
  File "/home/bolin/Desktop/hidet/python/hidet/graph/frontend/torch/interpreter.py", line 186, in _raise_exception
    raise RuntimeError('\n'.join(msg))
torch._dynamo.exc.BackendCompilerFailed: backend='hidet' raised:
RuntimeError: Can not interpret torch.nn.functional.batch_norm given arguments:
  torch.nn.functional.batch_norm(tensor(...), tensor(...), tensor(...), tensor(...), tensor(...), training=False, eps=1e-05)
Possible candidates are:
  batch_norm(x: hidet.Tensor, running_mean: Optional[hidet.Tensor], running_var: Optional[hidet.Tensor], weight: Optional[hidet.Tensor], bias: Optional[hidet.Tensor], training: bool, momentum: float, eps: float)
    File "/home/bolin/Desktop/hidet/python/hidet/graph/frontend/torch/register_functions.py", line 302
```

This PR also fixes this error by adding default values to some
parameters of the `batch_norm` function in
`torch.nn.functional.batch_norm`, to match the signature as in the[
PyTorch
documents](https://pytorch.org/docs/stable/generated/torch.nn.functional.batch_norm.html).

…_fpn` (#455)

Closes #264 

The error encountered in the linked issues was due to a subtle
difference in type promotions when calling `torch.div` with the argument
`rounding_mode='floor'`. Specifically, if both of the two operands are
of the integer type, then the output would still be integer type. This
is different from my original implementation, which first calls
`truediv` and then `ops.floor`, which will make the output datatype
`float32`.

After fixing this issue, another error was encountered:

```
File "/home/bolin/Desktop/hidet/python/hidet/graph/frontend/torch/interpreter.py", line 70, in __call__
    return self.forward(*args)
           ^^^^^^^^^^^^^^^^^^^
  File "/home/bolin/Desktop/hidet/python/hidet/graph/frontend/torch/interpreter.py", line 237, in forward
    self._raise_exception(e, node.target, exec_func, hidet_args, hidet_kwargs)
  File "/home/bolin/Desktop/hidet/python/hidet/graph/frontend/torch/interpreter.py", line 186, in _raise_exception
    raise RuntimeError('\n'.join(msg))
torch._dynamo.exc.BackendCompilerFailed: backend='hidet' raised:
RuntimeError: Can not interpret torch.nn.functional.batch_norm given arguments:
  torch.nn.functional.batch_norm(tensor(...), tensor(...), tensor(...), tensor(...), tensor(...), training=False, eps=1e-05)
Possible candidates are:
  batch_norm(x: hidet.Tensor, running_mean: Optional[hidet.Tensor], running_var: Optional[hidet.Tensor], weight: Optional[hidet.Tensor], bias: Optional[hidet.Tensor], training: bool, momentum: float, eps: float)
    File "/home/bolin/Desktop/hidet/python/hidet/graph/frontend/torch/register_functions.py", line 302
```

This PR also fixes this error by adding default values to some
parameters of the `batch_norm` function in
`torch.nn.functional.batch_norm`, to match the signature as in the[
PyTorch
documents](https://pytorch.org/docs/stable/generated/torch.nn.functional.batch_norm.html).

…_fpn` (#455)

Closes #264 

The error encountered in the linked issues was due to a subtle
difference in type promotions when calling `torch.div` with the argument
`rounding_mode='floor'`. Specifically, if both of the two operands are
of the integer type, then the output would still be integer type. This
is different from my original implementation, which first calls
`truediv` and then `ops.floor`, which will make the output datatype
`float32`.

After fixing this issue, another error was encountered:

```
File "/home/bolin/Desktop/hidet/python/hidet/graph/frontend/torch/interpreter.py", line 70, in __call__
    return self.forward(*args)
           ^^^^^^^^^^^^^^^^^^^
  File "/home/bolin/Desktop/hidet/python/hidet/graph/frontend/torch/interpreter.py", line 237, in forward
    self._raise_exception(e, node.target, exec_func, hidet_args, hidet_kwargs)
  File "/home/bolin/Desktop/hidet/python/hidet/graph/frontend/torch/interpreter.py", line 186, in _raise_exception
    raise RuntimeError('\n'.join(msg))
torch._dynamo.exc.BackendCompilerFailed: backend='hidet' raised:
RuntimeError: Can not interpret torch.nn.functional.batch_norm given arguments:
  torch.nn.functional.batch_norm(tensor(...), tensor(...), tensor(...), tensor(...), tensor(...), training=False, eps=1e-05)
Possible candidates are:
  batch_norm(x: hidet.Tensor, running_mean: Optional[hidet.Tensor], running_var: Optional[hidet.Tensor], weight: Optional[hidet.Tensor], bias: Optional[hidet.Tensor], training: bool, momentum: float, eps: float)
    File "/home/bolin/Desktop/hidet/python/hidet/graph/frontend/torch/register_functions.py", line 302
```

This PR also fixes this error by adding default values to some
parameters of the `batch_norm` function in
`torch.nn.functional.batch_norm`, to match the signature as in the[
PyTorch
documents](https://pytorch.org/docs/stable/generated/torch.nn.functional.batch_norm.html).