numpy/numpy · error · TypeError

threshold must be numeric

Error message

threshold must be numeric

What it means

Raised as TypeError by set_printoptions when threshold is not an instance of numbers.Number. threshold controls how many elements trigger summarization; this guard (added per gh-12351) rejects a common bad value such as a string that was historically suggested online.

Source

Thrown at numpy/_core/arrayprint.py:107

    elif legacy == '1.21':
        options['legacy'] = 121
    elif legacy == '1.25':
        options['legacy'] = 125
    elif legacy == '2.1':
        options['legacy'] = 201
    elif legacy == '2.2':
        options['legacy'] = 202
    elif legacy is None:
        pass  # OK, do nothing.
    else:
        warnings.warn(
            "legacy printing option can currently only be '1.13', '1.21', "
            "'1.25', '2.1', '2.2' or `False`", stacklevel=3)

    if threshold is not None:
        # forbid the bad threshold arg suggested by stack overflow, gh-12351
        if not isinstance(threshold, numbers.Number):
            raise TypeError("threshold must be numeric")
        if np.isnan(threshold):
            raise ValueError("threshold must be non-NAN, try "
                             "sys.maxsize for untruncated representation")

    if precision is not None:
        # forbid the bad precision arg as suggested by issue #18254
        try:
            options['precision'] = operator.index(precision)
        except TypeError as e:
            raise TypeError('precision must be an integer') from e

    return options


@set_module('numpy')
def set_printoptions(precision=None, threshold=None, edgeitems=None,
                     linewidth=None, suppress=None, nanstr=None,
                     infstr=None, formatter=None, sign=None, floatmode=None,

View on GitHub (pinned to e117b3ca4e)

Solutions

1. Pass an int (e.g. 1000) or sys.maxsize to disable summarization
2. Coerce external input: threshold = int(threshold) before calling
3. Ensure threshold is a numbers.Number (int, float that is not NaN)

Example fix

# before
np.set_printoptions(threshold='all')
# after
import sys
np.set_printoptions(threshold=sys.maxsize)

import numbers
def safe_threshold(thr):
    if not isinstance(thr, numbers.Number):
        raise TypeError(f"threshold must be numeric, got {type(thr).__name__}")
    return int(thr)
# np.set_printoptions(threshold=safe_threshold(user_thr))

import numbers
def is_numeric_threshold(thr) -> bool:
    return isinstance(thr, numbers.Number) and not isinstance(thr, bool)

try:
    np.set_printoptions(threshold=thr)
except TypeError as e:
    if 'threshold must be numeric' in str(e):
        import sys
        np.set_printoptions(threshold=sys.maxsize)
    else:
        raise

When it happens

Trigger: np.set_printoptions(threshold='all'); threshold=np.inf supplied as a string; passing a non-numeric object.

Common situations: Copying outdated Stack Overflow advice; deserializing threshold from JSON as a string; UI controls returning text.

Related errors

• threshold must be non-NAN, try sys.maxsize for untruncated r
• precision must be an integer
• floatmode option must be one of "fixed", "unique", "maxprec"
• sign option must be one of ' ', '+', or '-'
• unsupported order value: {order}

AI-assisted analysis of numpy/numpy@e117b3ca4e (2026-08-07). Data as JSON: /api/errors/f5e7b74ca323c907. Report an issue: GitHub.