File size: 6,403 Bytes
f791e67
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
#!/usr/bin/env python3
"""Standalone fail-closed checker for the Space's retained Claim 1–3 evidence.

This intentionally uses only Python's standard library and files reachable
from the Space entrypoint. It recomputes the row-level ONNX and independent
CodeNet correlations, and integrity-checks the retained APPS/KBSS summary.
"""
from __future__ import annotations

import csv
import hashlib
import json
import math
import statistics
from collections import defaultdict
from pathlib import Path


ROOT = Path(__file__).resolve().parents[1]
EVIDENCE = ROOT / "evidence/cumulative"
EXPECTED_HASHES = {
    "full_n512.csv":
        "3016dfb245deb56b3d0d3d06f902471bc415f80196d61ecf4d49eb773f1edd76",
    "full_gpu_n200.csv":
        "7843f75c38b9ee0bf3f15d796f151808fc01a526c704f429049016d57538f390",
    "table3_results.json":
        "173dbce239ca51269d3b231dd851b7fb299dffc82d2b0eede75f61ae6a835dc2",
}
EXPECTED_ACCURACY = {
    "NASBench101": 0.4065993610622034,
    "ENAS": 0.24946117751369043,
    "NASNet": 0.20673752343866375,
}
EXPECTED_LANGUAGES = {
    "C++", "Python", "Java", "C", "Ruby", "C#", "Rust", "Go", "Haskell",
    "Kotlin", "JavaScript", "PHP", "D", "Scala", "OCaml", "Perl", "Fortran",
}


def sha256(path: Path) -> str:
    return hashlib.sha256(path.read_bytes()).hexdigest()


def ranks(values: list[float]) -> list[float]:
    """Average ranks for ties, equivalent to scipy.stats.rankdata."""
    order = sorted(range(len(values)), key=values.__getitem__)
    output = [0.0] * len(values)
    start = 0
    while start < len(order):
        end = start + 1
        while end < len(order) and values[order[end]] == values[order[start]]:
            end += 1
        rank = (start + 1 + end) / 2
        for index in order[start:end]:
            output[index] = rank
        start = end
    return output


def spearman(xs: list[float], ys: list[float]) -> float:
    rx, ry = ranks(xs), ranks(ys)
    mx, my = statistics.fmean(rx), statistics.fmean(ry)
    numerator = sum((x - mx) * (y - my) for x, y in zip(rx, ry))
    denominator = math.sqrt(
        sum((x - mx) ** 2 for x in rx) * sum((y - my) ** 2 for y in ry)
    )
    if denominator == 0:
        raise AssertionError("undefined Spearman correlation")
    return numerator / denominator


def check_hashes() -> dict[str, str]:
    observed = {}
    for filename, expected in EXPECTED_HASHES.items():
        path = EVIDENCE / filename
        if not path.is_file():
            raise AssertionError(f"missing retained artifact: {filename}")
        observed[filename] = sha256(path)
        if observed[filename] != expected:
            raise AssertionError(f"SHA-256 mismatch: {filename}")
    return observed


def check_accuracy() -> dict:
    grouped: dict[str, list[tuple[float, float]]] = defaultdict(list)
    identifiers = set()
    with (EVIDENCE / "full_n512.csv").open(newline="") as handle:
        for row in csv.DictReader(handle):
            identifier = row["identifier"]
            if identifier in identifiers:
                raise AssertionError(f"duplicate identifier: {identifier}")
            identifiers.add(identifier)
            draws = [float(row[f"draw_{index}"]) for index in range(8)]
            prediction = float(row["prediction"])
            if not math.isclose(
                prediction, statistics.median(draws), abs_tol=1e-12
            ):
                raise AssertionError(f"prediction is not median: {identifier}")
            grouped[row["space"]].append((float(row["target"]), prediction))
    if set(grouped) != set(EXPECTED_ACCURACY):
        raise AssertionError("accuracy spaces changed")
    observed = {}
    for space, expected in EXPECTED_ACCURACY.items():
        pairs = grouped[space]
        if len(pairs) != 512:
            raise AssertionError(f"{space}: expected 512 rows")
        xs, ys = map(list, zip(*pairs))
        rho = spearman(xs, ys)
        if not math.isclose(rho, expected, abs_tol=1e-12):
            raise AssertionError(f"{space}: Spearman changed: {rho}")
        observed[space] = rho
    return {"rows": len(identifiers), "raw_draws": len(identifiers) * 8,
            "spearman": observed}


def check_codenet() -> dict:
    grouped: dict[str, list[tuple[float, float]]] = defaultdict(list)
    with (EVIDENCE / "full_gpu_n200.csv").open(newline="") as handle:
        for row in csv.DictReader(handle):
            grouped[row["language"]].append(
                (float(row["y_true"]), float(row["y_pred"]))
            )
    if set(grouped) != EXPECTED_LANGUAGES:
        raise AssertionError("CodeNet language set changed")
    observed = {}
    for language, pairs in grouped.items():
        if len(pairs) != 200:
            raise AssertionError(f"{language}: expected 200 rows")
        xs, ys = map(list, zip(*pairs))
        observed[language] = spearman(xs, ys)
    average = statistics.fmean(observed.values())
    if not math.isclose(average, 0.5234034026121069, abs_tol=1e-12):
        raise AssertionError(f"CodeNet mean changed: {average}")
    return {"languages": len(observed), "rows_per_language": 200,
            "average_spearman": average, "per_language": observed}


def check_table3() -> dict:
    summary = json.loads((EVIDENCE / "table3_results.json").read_text())["table3"]
    apps, kbss = summary["APPS"], summary["KBSS"]
    if apps["n"] != 512 or not math.isclose(
        apps["spearman"], 0.9268067718469594, abs_tol=1e-15
    ):
        raise AssertionError("APPS retained summary changed")
    if kbss["n"] != 512 or not math.isclose(
        kbss["spearman"], 0.5352789637599933, abs_tol=1e-15
    ):
        raise AssertionError("KBSS retained summary changed")
    return {"APPS": apps, "KBSS": kbss}


def verify() -> dict:
    return {
        "status": "PASS",
        "scope": (
            "row-level recomputation for ONNX accuracy and independent CodeNet; "
            "hash and summary integrity for APPS/KBSS"
        ),
        "sha256": check_hashes(),
        "claim_1": check_accuracy(),
        "claim_2": check_table3(),
        "claim_3": check_codenet(),
        "limitation": (
            "The judged repository did not retain paper-scale APPS/KBSS rows; "
            "this checker cannot independently recompute those two correlations."
        ),
    }


if __name__ == "__main__":
    print("SPACE_CUMULATIVE_RESULT " + json.dumps(verify(), sort_keys=True))