How to initalize an empty list in torch jit

Question

I want to use torch jit to construct a list of int. I did not find out how to initialize an empty int List. I can initialize the list with one integer, and I could this initial value at the end, but this is not how it should be done. Here are my attempts:

import torch
from typing import List

@torch.jit.script
def f(n: int) -> List[int]:
    out = [0]
    for i in range(n):
        out.append(i)
    return out

This one is OK but I have an extra first value I do not want.

@torch.jit.script
def g(n: int) -> List[int]:
    out = []
    for i in range(n):
        out.append(i)
    return out

crashes with

aten::append.t(t[](a!) self, t(c -> *) el) -> t[](a!):
Could not match type int to t in argument 'el': Type variable 't' previously matched to type Tensor is matched to type int.
:
  File "<ipython-input-5-f405ae8f8cbb>", line 5
    out = []
    for i in range(n):
        out.append(i)
        ~~~~~~~~~~ <--- HERE
    return out

I understand the compiler needs to understand the data type of the list. So I tried

@torch.jit.script
def h(n: int) -> List[int]:
    out = List[int]
    for i in range(n):
        out.append(i)
    return out

---------------------------------------------------------------------------
RuntimeError                              Traceback (most recent call last)
Cell In[6], line 2
      1 @torch.jit.script
----> 2 def h(n: int) -> List[int]:
      3     out = List[int]
      4     for i in range(n):

File /mnt/sw/nix/store/gpkc8q6zjnp3n3h3w9hbmbj6gjbxs85w-python-3.10.10-view/lib/python3.10/site-packages/torch/jit/_script.py:1341, in script(obj, optimize, _frames_up, _rcb, example_inputs)
   1339 if _rcb is None:
   1340     _rcb = _jit_internal.createResolutionCallbackFromClosure(obj)
-> 1341 fn = torch._C._jit_script_compile(
   1342     qualified_name, ast, _rcb, get_default_args(obj)
   1343 )
   1344 # Forward docstrings
   1345 fn.__doc__ = obj.__doc__

File /mnt/sw/nix/store/gpkc8q6zjnp3n3h3w9hbmbj6gjbxs85w-python-3.10.10-view/lib/python3.10/site-packages/torch/jit/annotations.py:331, in try_ann_to_type(ann, loc)
    329         return TupleType([])
    330     return TupleType([try_ann_to_type(a, loc) for a in ann.__args__])
--> 331 if is_list(ann):
    332     elem_type = try_ann_to_type(ann.__args__[0], loc)
    333     if elem_type:

File /mnt/sw/nix/store/gpkc8q6zjnp3n3h3w9hbmbj6gjbxs85w-python-3.10.10-view/lib/python3.10/site-packages/torch/_jit_internal.py:993, in is_list(ann)
    991 def is_list(ann) -> bool:
    992     if ann is List:
--> 993         raise_error_container_parameter_missing("List")
    995     if not hasattr(ann, "__module__"):
    996         return False

File /mnt/sw/nix/store/gpkc8q6zjnp3n3h3w9hbmbj6gjbxs85w-python-3.10.10-view/lib/python3.10/site-packages/torch/_jit_internal.py:1269, in raise_error_container_parameter_missing(target_type)
   1263 if target_type == "Dict":
   1264     raise RuntimeError(
   1265         "Attempted to use Dict without "
   1266         "contained types. Please add contained type, e.g. "
   1267         "Dict[int, int]"
   1268     )
-> 1269 raise RuntimeError(
   1270     f"Attempted to use {target_type} without a "
   1271     "contained type. Please add a contained type, e.g. "
   1272     f"{target_type}[int]"
   1273 )

RuntimeError: Attempted to use List without a contained type. Please add a contained type, e.g. List[int]

This is the part I do not understand. It should work. Is this a bug? Am I doing something wrong?

torch version 2.0.1.