Files
AareLC_train/data/gold/unseen_core.json
T
GotthardGandClaude Fable 5.1 831367d35d noise-aware gate, frozen core of unseen frames, and a cycle script for the Spark
A paired bootstrap over the 149 unseen gold frames (2026-10-02) showed that a
difference between two models carries about +-0.023 of noise in F1, +-0.018 in
the mask score and +-0.10 in crystal F1, while the gate's tolerance was 0.005.

- tools/gate_stats.py: per-frame counts and a paired bootstrap (pure Python).
- train_run.py: when the baseline's gold predictions exist, the gate requires
  the candidate to beat it on F1 or the mask score in 90 % of the draws and
  not to lose the other in more than 90 %; the point rule is the fallback and
  is always reported (pass_point_rule).
- tools/compare_unseen.py replaces the Spark's ~/fair_gate_all.py: all models
  on the frames none trained on, on the frozen core (data/gold/unseen_core.json,
  149 frames) and on all unseen frames once the gold set grows, with the
  interval and win share against the active model.
- measure_weights.py keeps a version's gate verdict when it re-measures it.
- tools/spark_cycle.sh: gold growth, export, re-measure, train, gate, compare,
  push, with what cycles 3 and 4 taught built in.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
2026-10-02 13:25:45 +02:00

211 lines
3.1 KiB
JSON

{
"frozen": "2026-10-02",
"note": "core = gold frames no model had trained on when the set was frozen; seen = gold frames the April beamline model trained on (or a dataset used)",
"core": [
"tpi_10130",
"tpi_10697",
"tpi_1119",
"tpi_14517",
"tpi_14688",
"tpi_14689",
"tpi_14698",
"tpi_14709",
"tpi_14711",
"tpi_14715",
"tpi_14718",
"tpi_15087",
"tpi_16223",
"tpi_16235",
"tpi_16237",
"tpi_16252",
"tpi_16278",
"tpi_16292",
"tpi_16319",
"tpi_16321",
"tpi_16325",
"tpi_16357",
"tpi_16361",
"tpi_16747",
"tpi_16753",
"tpi_16766",
"tpi_16770",
"tpi_16956",
"tpi_16963",
"tpi_17889",
"tpi_20221",
"tpi_21693",
"tpi_31669",
"tpi_31928",
"tpi_31929",
"tpi_32762",
"tpi_33849",
"tpi_34038",
"tpi_34389",
"tpi_36563",
"tpi_4053",
"tpi_45649",
"tpi_52402",
"tpi_55303",
"tpi_55371",
"tpi_55383",
"tpi_55408",
"tpi_55416",
"tpi_55565",
"tpi_55896",
"tpi_55912",
"tpi_55941",
"tpi_55958",
"tpi_56012",
"tpi_56096",
"tpi_56104",
"tpi_56715",
"tpi_56717",
"tpi_56840",
"tpi_57070",
"tpi_57242",
"tpi_57300",
"tpi_57342",
"tpi_57654",
"tpi_57684",
"tpi_57702",
"tpi_57991",
"tpi_58097",
"tpi_58229",
"tpi_58417",
"tpi_58588",
"tpi_58595",
"tpi_58842",
"tpi_59004",
"tpi_59011",
"tpi_59120",
"tpi_59182",
"tpi_59312",
"tpi_59412",
"tpi_59496",
"tpi_59564",
"tpi_59570",
"tpi_59911",
"tpi_59988",
"tpi_60052",
"tpi_60625",
"tpi_61066",
"tpi_61069",
"tpi_61087",
"tpi_61317",
"tpi_61327",
"tpi_61334",
"tpi_61340",
"tpi_61343",
"tpi_61344",
"tpi_61350",
"tpi_61358",
"tpi_61361",
"tpi_61364",
"tpi_61372",
"tpi_61374",
"tpi_61379",
"tpi_61380",
"tpi_61381",
"tpi_61385",
"tpi_61392",
"tpi_61411",
"tpi_61419",
"tpi_61428",
"tpi_61430",
"tpi_61436",
"tpi_61437",
"tpi_61439",
"tpi_61441",
"tpi_61442",
"tpi_61443",
"tpi_61445",
"tpi_61447",
"tpi_61449",
"tpi_61450",
"tpi_61452",
"tpi_61457",
"tpi_61565",
"tpi_61572",
"tpi_61579",
"tpi_61581",
"tpi_61594",
"tpi_61610",
"tpi_61613",
"tpi_61618",
"tpi_61627",
"tpi_61629",
"tpi_61633",
"tpi_61635",
"tpi_61651",
"tpi_61653",
"tpi_61658",
"tpi_61670",
"tpi_61686",
"tpi_61702",
"tpi_61769",
"tpi_61834",
"tpi_61839",
"tpi_61865",
"tpi_6203",
"tpi_7667",
"tpi_8507",
"tpi_8744",
"tpi_9752"
],
"seen": [
"tpi_10036",
"tpi_10948",
"tpi_11106",
"tpi_11133",
"tpi_11523",
"tpi_12381",
"tpi_12548",
"tpi_12567",
"tpi_12580",
"tpi_1265",
"tpi_12804",
"tpi_14167",
"tpi_14524",
"tpi_14530",
"tpi_14552",
"tpi_14642",
"tpi_14863",
"tpi_2014",
"tpi_2640",
"tpi_3077",
"tpi_3466",
"tpi_3550",
"tpi_4296",
"tpi_4313",
"tpi_4528",
"tpi_5066",
"tpi_5178",
"tpi_5225",
"tpi_5387",
"tpi_5425",
"tpi_5725",
"tpi_5956",
"tpi_6202",
"tpi_6209",
"tpi_6730",
"tpi_6737",
"tpi_6787",
"tpi_7072",
"tpi_7302",
"tpi_7348",
"tpi_7532",
"tpi_7658",
"tpi_7683",
"tpi_7751",
"tpi_7810",
"tpi_7836",
"tpi_7879",
"tpi_8142",
"tpi_8196",
"tpi_8348",
"tpi_8400",
"tpi_8715",
"tpi_8855",
"tpi_9805"
]
}