PANOPTES
panoptes-e86ef9e3

calib·HumanEval/12

humaneval-calibprobe · code

This is one (task, candidate response) pair flowing through the full PANOPTES pipeline. Each section below is a stage of the analysis: the task itself, the candidate solution being evaluated, every judge's score + rationale, the sampling-pass dispersion that captures within-judge noise, and the uncertainty-quantification metrics computed on top.

mean score
1.000
inter-judge spread
0.000
judges polled
2
sampling draws
20
1. the task

The function signature + docstring presented to both the model under test and to every judge. The entry-point name was obfuscated from longest to fn_d98e9fb34b so the judges can't pattern-match a memorized HumanEval solution.

from typing import List, Optional


def fn_d98e9fb34b(strings: List[str]) -> Optional[str]:
    """ Out of list of strings, return the fn_d98e9fb34b one. Return the first one in case of multiple
    strings of the same length. Return None in case the input list is empty.
    >>> fn_d98e9fb34b([])

    >>> fn_d98e9fb34b(['a', 'b', 'c'])
    'a'
    >>> fn_d98e9fb34b(['a', 'bb', 'ccc'])
    'ccc'
    """
2. the candidate response judged

The full solution every judge is grading. For these runs the candidate is the reference solution prepended with the task signature so it parses as a complete program.

from typing import List, Optional


def fn_d98e9fb34b(strings: List[str]) -> Optional[str]:
    """ Out of list of strings, return the fn_d98e9fb34b one. Return the first one in case of multiple
    strings of the same length. Return None in case the input list is empty.
    >>> fn_d98e9fb34b([])

    >>> fn_d98e9fb34b(['a', 'b', 'c'])
    'a'
    >>> fn_d98e9fb34b(['a', 'bb', 'ccc'])
    'ccc'
    """
    if not strings:
        return None

    maxlen = max(len(x) for x in strings)
    for s in strings:
        if len(s) == maxlen:
            return s
3a. point-pass scores (temperature 0)

One call per judge at temperature=0. This is the "best single guess" each judge has. Disagreement here is structural. The judges literally see this candidate differently.

claude-sonnet
1.000
$0.0071
gpt-4o
1.000
$0.0033
3b. sampling-pass dispersion (temperature 1)

n draws per judge at temperature=1. Dots are individual draws; vertical bar is the mean; shaded band is ±1σ. Wide band = the judge is uncertain even with itself; narrow band = it's consistent.

4. uncertainty quantification

Three statistical methods computed on top of the raw scores above. Each one tells you something different about how much to trust the headline number.

semantic-entropy
claude-sonnet
H = 0.000/ log 102.303
1 semantic cluster via bidirectional NLI
entropy as fraction of max
cluster sizes
c1=10
semantic-entropy
gpt-4o
H = 0.325/ log 102.303
2 semantic clusters via bidirectional NLI
entropy as fraction of max
cluster sizes
c1=9 · c2=1

How to read these: self-consistency = does the judge agree with itself when resampled? Narrower CI means more consistent. Semantic entropy = do the judge's rationales cluster into one meaning, or several? Higher entropy means the judge is internally conflicted about why. Decomposition = how much of the total variance comes from within-judge sampling noise (aleatoric, hard to fix) vs between-judge disagreement (epistemic, fixable by calling more judges).

5. judge rationales

The natural-language explanation each judge gave alongside its score. These are produced via tool-use structured output, so the score and the rationale are guaranteed to come from the same forward pass.

claude-sonnet
score 1.000 · 1883 tokens · 3041ms

The solution correctly handles all cases: returns None for empty list, returns the longest string, and returns the first one in case of ties (by iterating in order and returning on first match). All three doctests pass. Logic is clean and idiomatic.

gpt-4o
score 1.000 · 1051 tokens · 1609ms

The candidate solution correctly implements the function to return the longest string from a list of strings. It handles the edge case of an empty list by returning None, as specified. The solution also correctly returns the first string in case of multiple strings of the same maximum length. The implementation is clear and idiomatic, using a generator expression to find the maximum length and a simple loop to return the first string of that length.