feat(learning-harness): fix dates and the post-cutoff slice

The catalog generator fills `fix_date` from the recipe at the precision the
recipe has -- a full ISO day when one was recorded, otherwise the year.
Every harvested recipe carries a year and almost none carries a day, and
writing `-01-01` would turn a year into a claim about a month. A row counts
as post-cutoff only when its earliest possible date is after the cutoff, so
a coarse date can under-claim the clean slice but never over-claim it.

Each family now reports the same numbers over the rows fixed after the
cutoff, beside the whole number and never instead of it, with the undated
count on its own line. Without a configured cutoff every scoreboard says so
rather than inventing one: guessing a model's training cutoff and then
publishing a contamination number from the guess would be worse than
reporting none.

Co-Authored-By: Claude Opus 5 (1M context) <[email protected]>
Claude-Session: https://claude.ai/code/session_01T7XD7x5KonMdmGRkp63EQG
This commit is contained in:
Marius Ciepluch
2026-09-06 13:15:20 +02:00
co-authored by Claude Opus 5
parent d1faa04771
commit 3e71e9e88b
12 changed files with 1195 additions and 360 deletions
+59
View File
@@ -25,6 +25,7 @@ fix_policy = "silent"
provenance = "agent"
split = "train"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-16"
@@ -56,6 +57,7 @@ fix_policy = "silent"
provenance = "agent"
split = "holdout"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-89"
@@ -85,6 +87,7 @@ fix_policy = "silent"
provenance = "agent"
split = "holdout"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-918"
@@ -114,6 +117,7 @@ fix_policy = "silent"
provenance = "agent"
split = "holdout"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-918"
@@ -145,6 +149,7 @@ fix_policy = "silent"
provenance = "agent"
split = "train"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-601"
@@ -176,6 +181,7 @@ fix_policy = "silent"
provenance = "agent"
split = "holdout"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-16"
@@ -208,6 +214,7 @@ fix_policy = "silent"
provenance = "agent"
split = "train"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-862"
@@ -239,6 +246,7 @@ fix_policy = "silent"
provenance = "agent"
split = "holdout"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-16"
@@ -268,6 +276,7 @@ fix_policy = "silent"
provenance = "agent"
split = "holdout"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-16"
@@ -299,6 +308,7 @@ fix_policy = "silent"
provenance = "agent"
split = "train"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-639"
@@ -330,6 +340,7 @@ fix_policy = "silent"
provenance = "agent"
split = "holdout"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-16"
@@ -361,6 +372,7 @@ fix_policy = "silent"
provenance = "agent"
split = "train"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-16"
@@ -390,6 +402,7 @@ fix_policy = "silent"
provenance = "agent"
split = "holdout"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-22"
@@ -421,6 +434,7 @@ fix_policy = "silent"
provenance = "agent"
split = "holdout"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-862"
@@ -450,6 +464,7 @@ fix_policy = "silent"
provenance = "agent"
split = "train"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-1321"
@@ -481,6 +496,7 @@ fix_policy = "silent"
provenance = "agent"
split = "train"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-22"
@@ -513,6 +529,7 @@ fix_policy = "silent"
provenance = "agent"
split = "holdout"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-89"
@@ -542,6 +559,7 @@ fix_policy = "silent"
provenance = "agent"
split = "train"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-1321"
@@ -571,6 +589,7 @@ fix_policy = "silent"
provenance = "agent"
split = "holdout"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-1321"
@@ -602,6 +621,7 @@ fix_policy = "silent"
provenance = "agent"
split = "train"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-601"
@@ -633,6 +653,7 @@ fix_policy = "silent"
provenance = "agent"
split = "holdout"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-16"
@@ -664,6 +685,7 @@ fix_policy = "silent"
provenance = "agent"
split = "train"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-601"
@@ -695,6 +717,7 @@ fix_policy = "silent"
provenance = "agent"
split = "holdout"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-798"
@@ -726,6 +749,7 @@ fix_policy = "silent"
provenance = "agent"
split = "train"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-22"
@@ -758,6 +782,7 @@ fix_policy = "silent"
provenance = "agent"
split = "holdout"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-22"
@@ -789,6 +814,7 @@ fix_policy = "silent"
provenance = "agent"
split = "train"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-601"
@@ -820,6 +846,7 @@ fix_policy = "silent"
provenance = "agent"
split = "holdout"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-89"
@@ -852,6 +879,7 @@ fix_policy = "silent"
provenance = "agent"
split = "train"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-601"
@@ -882,6 +910,7 @@ provenance = "agent"
split = "holdout"
review_tier = "title"
known_limit = "client_side_caller"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-862"
@@ -912,6 +941,7 @@ fix_policy = "silent"
provenance = "agent"
split = "train"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-601"
@@ -941,6 +971,7 @@ fix_policy = "silent"
provenance = "agent"
split = "holdout"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-16"
@@ -970,6 +1001,7 @@ fix_policy = "silent"
provenance = "agent"
split = "train"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-16"
@@ -999,6 +1031,7 @@ fix_policy = "silent"
provenance = "agent"
split = "holdout"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-1321"
@@ -1028,6 +1061,7 @@ fix_policy = "silent"
provenance = "agent"
split = "train"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-1321"
@@ -1059,6 +1093,7 @@ fix_policy = "silent"
provenance = "agent"
split = "holdout"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-16"
@@ -1088,6 +1123,7 @@ fix_policy = "silent"
provenance = "agent"
split = "holdout"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-16"
@@ -1117,6 +1153,7 @@ fix_policy = "silent"
provenance = "agent"
split = "train"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-918"
@@ -1148,6 +1185,7 @@ fix_policy = "silent"
provenance = "agent"
split = "train"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-798"
@@ -1179,6 +1217,7 @@ fix_policy = "silent"
provenance = "agent"
split = "holdout"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-22"
@@ -1209,6 +1248,7 @@ fix_policy = "silent"
provenance = "agent"
split = "train"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-639"
@@ -1239,6 +1279,7 @@ fix_policy = "silent"
provenance = "agent"
split = "holdout"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-639"
@@ -1269,6 +1310,7 @@ fix_policy = "silent"
provenance = "agent"
split = "train"
review_tier = "title"
fix_date = "2026"
[[pair.context]]
relpath = "services/web/app/src/router.mjs"
@@ -1304,6 +1346,7 @@ fix_policy = "silent"
provenance = "agent"
split = "holdout"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-1321"
@@ -1333,6 +1376,7 @@ fix_policy = "silent"
provenance = "agent"
split = "train"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-1321"
@@ -1363,6 +1407,7 @@ fix_policy = "silent"
provenance = "agent"
split = "holdout"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-16"
@@ -1394,6 +1439,7 @@ fix_policy = "silent"
provenance = "agent"
split = "train"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-89"
@@ -1426,6 +1472,7 @@ fix_policy = "silent"
provenance = "agent"
split = "holdout"
review_tier = "title"
fix_date = "2025"
[[pair.expected]]
cwe = "CWE-601"
@@ -1457,6 +1504,7 @@ fix_policy = "silent"
provenance = "agent"
split = "train"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-918"
@@ -1486,6 +1534,7 @@ fix_policy = "silent"
provenance = "agent"
split = "holdout"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-639"
@@ -1516,6 +1565,7 @@ fix_policy = "silent"
provenance = "agent"
split = "train"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-639"
@@ -1548,6 +1598,7 @@ fix_policy = "silent"
provenance = "agent"
split = "holdout"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-601"
@@ -1577,6 +1628,7 @@ fix_policy = "silent"
provenance = "agent"
split = "holdout"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-78"
@@ -1608,6 +1660,7 @@ fix_policy = "silent"
provenance = "agent"
split = "train"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-639"
@@ -1637,6 +1690,7 @@ fix_policy = "silent"
provenance = "agent"
split = "holdout"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-22"
@@ -1668,6 +1722,7 @@ fix_policy = "silent"
provenance = "agent"
split = "holdout"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-16"
@@ -1697,6 +1752,7 @@ fix_policy = "silent"
provenance = "agent"
split = "train"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-22"
@@ -1726,6 +1782,7 @@ fix_policy = "silent"
provenance = "agent"
split = "train"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-918"
@@ -1755,6 +1812,7 @@ fix_policy = "silent"
provenance = "agent"
split = "holdout"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-89"
@@ -1784,6 +1842,7 @@ fix_policy = "silent"
provenance = "agent"
split = "train"
review_tier = "title"
fix_date = "2026"
[[pair.expected]]
cwe = "CWE-16"