1fabd4a246
The model layer of the project, built honestly:
- tools/dataset_validate.py — full-sweep validator over the receiver
store (sha256, schema, monotonic labels, telemetry-row gate). On the
current corpus: 64,798 accepted + 8,154 degraded + 3,701 rejected +
7 errored across 76,660 shipped episodes. data/processed/validation_v1.parquet
is committed as the per-episode acceptance index.
- training/_features.py — channel registry (46 channels across
proc/guest/qmp/netflow), summary-stat windowing AND channel×time
tensor extraction at 10s/5s windowing. Time alignment uses t_wall_ns
(Unix ns) — tested fix for a real netflow-vs-host clock-base
inconsistency that was silently dropping every netflow channel.
- training/_split.py — three held-out recipes (host / sample / time)
with profile-stratification assertions. held_out_host carries
untested_profiles for cases like scan-and-dial absent from the test
host (5 of 6 profiles tested cross-device, never silently averaged).
- training/models/ — 6 architectures behind a common BaseModel
interface: gbt (XGBoost), mlp, cnn, gru, lstm, transformer. Each
trained twice (realistic / oracle) per the deployment threat model.
Schema-hashed checkpoints refuse to load if _features.py changed
since training (silent-input-drift protection, tested).
- training/trainer/ — unified training loop: class-weighted CE, LR
warmup + cosine, gradient clipping, mixed precision when CUDA,
early stopping on val macro F1, best-on-val checkpoint. Same loop
runs MLP/CNN/GRU/LSTM/Transformer; GBT uses XGBoost
early_stopping_rounds on val mlogloss.
- training/eval_/ — bootstrap 95% CIs on macro F1, per-class F1,
per-profile and per-host breakdown, paired-bootstrap significance
for model-vs-model gap. Confusion matrix uses union of seen labels.
- training/dashboard/producers/ — replay/metrics/perf/profiles
emitting the six event types the dashboard's awaiting scenes
consume; on-demand tensor extraction so the Pi can run live
inference without 65 GB of shards.
- 17 unit tests (split coverage, features round-trip, schema mismatch,
determinism, time-base alignment regression).
End-to-end smoke-trained all six on a 567-episode subset; held-out
test macro F1 reported with paired-bootstrap significance. The
methodology now reports honest cross-device generalization, not
in-distribution validation.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
39 lines
1.3 KiB
Python
39 lines
1.3 KiB
Python
"""CLI dispatcher for dashboard producers.
|
|
|
|
python -m training.dashboard.producers replay --episode … --host-id …
|
|
python -m training.dashboard.producers metrics --window … --schema …
|
|
python -m training.dashboard.producers perf --window … --schema …
|
|
python -m training.dashboard.producers profiles --validation … --store …
|
|
|
|
Each subcommand forwards remaining argv to the matching module's main().
|
|
"""
|
|
from __future__ import annotations
|
|
|
|
import sys
|
|
|
|
|
|
SUBCOMMANDS = {
|
|
"replay": "training.dashboard.producers.replay",
|
|
"metrics": "training.dashboard.producers.metrics",
|
|
"perf": "training.dashboard.producers.perf",
|
|
"profiles": "training.dashboard.producers.profiles",
|
|
}
|
|
|
|
|
|
def main() -> int:
|
|
if len(sys.argv) < 2 or sys.argv[1] in {"-h", "--help"}:
|
|
print("usage: python -m training.dashboard.producers "
|
|
"<replay|metrics|perf|profiles> [args]", file=sys.stderr)
|
|
return 2
|
|
sub = sys.argv[1]
|
|
if sub not in SUBCOMMANDS:
|
|
print(f"unknown subcommand: {sub}", file=sys.stderr)
|
|
return 2
|
|
import importlib
|
|
mod = importlib.import_module(SUBCOMMANDS[sub])
|
|
sys.argv = [f"{sys.argv[0]} {sub}"] + sys.argv[2:]
|
|
return int(mod.main() or 0)
|
|
|
|
|
|
if __name__ == "__main__":
|
|
raise SystemExit(main())
|