jax-ml/jax · error · TypeError

len() of unsized object

Error message

len() of unsized object

What it means

RefRValue._len converts an IndexError from shape[0] into the numpy-style TypeError 'len() of unsized object', mirroring NumPy's behavior for 0-d arrays. It means the ref has a shape but it's empty (scalar-like).

Source

Thrown at jax/_src/state/types.py:523

    return self.update(inner_aval=self.inner_aval.update_weak_type(weak_type))

  def update_manual_axis_type(self, mat):
    return self.update(inner_aval=self.inner_aval.update_manual_axis_type(mat))

  def update(self, inner_aval=None, memory_space=None, kind=None):  # pyrefly: ignore[bad-override]
    inner_aval = self.inner_aval if inner_aval is None else inner_aval
    memory_space = self.memory_space if memory_space is None else memory_space
    kind = self.kind if kind is None else kind
    return AbstractRef(inner_aval, memory_space, kind)

  ndim = property(lambda self: len(self.shape))
  size = property(lambda self: math.prod(self.shape))

  def _len(self, ignored_tracer) -> int:
    try:
      return self.shape[0]
    except IndexError as err:
      raise TypeError("len() of unsized object") from err  # same as numpy error

  @property
  def shape(self):
    try:
      return self.inner_aval.shape  # pyrefly: ignore[missing-attribute]
    except AttributeError:
      raise AttributeError(
          f"{self!r} has no `shape`."
      ) from None

  @property
  def dtype(self):
    try:
      return self.inner_aval.dtype  # pyrefly: ignore[missing-attribute]
    except AttributeError:
      raise AttributeError(
          f"{self!r} has no `dtype`."
      ) from None

View on GitHub (pinned to 1e1c6a8fc0)

Solutions

1. Check ref.ndim > 0 before len()
2. Use ref.size to detect emptiness instead
3. Reshape the underlying buffer to at least 1D

Example fix

# before
n = len(ref)
# after
n = ref.shape[0] if ref.ndim > 0 else 1

if ref.ndim == 0:
    raise TypeError("ref is unsized")

def is_sized(ref):
    return getattr(ref, "ndim", 0) > 0

When it happens

Trigger: Calling len(ref) on a ref whose shape is () (0-dimensional / unsized).

Common situations: Generic code doing len(x) on containers that may hold scalars; refs created from scalar buffers.

Related errors

• len() of unsized object
• len() of unsized object
• numpy masked arrays are not supported as direct inputs to JA
• Python int {value} too large to convert to int64
• Python int {value} too large to convert to int32

AI-assisted analysis of jax-ml/jax@1e1c6a8fc0 (2026-08-27). Data as JSON: /api/errors/cff9dc86af85e49e. Report an issue: GitHub.