reference/generated/tensordict.assert_close

Run in Google Colab

Colab

Download Notebook

Notebook

View on GitHub

GitHub

assert_close

tensordict.assert_close(actual: T, expected: T, rtol: float | None = None, atol: float | None = None, equal_nan: bool = True, intersection: bool = False, msg: str = '', prefix: NestedKey = ()) → bool

Asserts that two tensordicts, actual and expected, are element-wise equal within a tolerance for all entries.

This function checks if the elements of the actual tensor are close to the corresponding elements of the expected tensordict, within a relative tolerance (rtol) and an absolute tolerance (atol).

It is similar to the assert_close() function in PyTorch, but with tensordicts inputs.

Parameters:

• actual (T) – The tensordict containing actual values.

• expected (T) – The tensordict containing expected values.

• rtol (float | None, optional) – The relative tolerance parameter. Default is None.

• atol (float | None, optional) – The absolute tolerance parameter. Default is None.

• equal_nan (bool, optional) – If True, NaNs will be considered equal to NaNs. Default is True.

• intersection (bool, optional) – If True, only the intersection of the two tensordicts will be compared. Default is False.

• msg (str, optional) – An optional message to include in the assertion error if the check fails.

• prefix (NestedKey, optional) – a prefix to add to the key for error messages.

Returns:

True if the tensors are close within the specified tolerances, raise an exception otherwise.

Return type:

bool

Raises:

AssertionError – If the tensordicts are not close within the specified tolerances.