The operator's call: "there should be one default learning rate and one
default max iterations and they should both be configurable at runtime."
There were SIX copies, not the four the audit described, and they did not
agree:
nisps/ml/mlp.hpp no-arg train() hardcoding 1.f / 1000u / 0.001f —
and firmware's ONLY training path calls exactly
this, so firmware had no runtime knob at all
wasm-iml.ts train() and trainAsync() TS default params (x2)
engine-api.ts learningRate ?? 1.0, with no maxIterations knob
vcv/src/iml.hpp 200 / 0.1 / 0.00001 — silently divergent
external_synth_midi.hpp its own kDefaultLearningRate/kDefaultMaxIterations
schemas/modes/*.json x9, identical, read by nobody at runtime
Now: schemas/ml_defaults.json is the single declaration (validated against a
sibling meta-schema, matching the midi_device.schema.json convention), codegen
emits it to C++ and TS in the same run, and MLPCore carries a TrainConfig whose
default member initialisers read the generated constant.
set_train_config()/nisps_ml_set_train_config() make it runtime-overridable on
every target; the explicit-argument train() overload is untouched. min_error
joins the tuple — it was duplicated identically and belongs with the other two.
The per-mode ml block loses default_learning_rate/default_max_iterations.
default_spread stays (genuinely wired on both targets) and input_channels stays
(codegen-time validated, real information for sound_analysis_midi).
VCV BEHAVIOUR CHANGE, deliberate: MEMLNaut.cpp constructs IML positionally and
relies on those defaults, so the module moves to 1000/1.0/0.001 — 5x the max
iterations, 10x the learning rate, and a 100x looser early-stop threshold. The
old values were never justified anywhere; they arrived with fbc68eb alongside
an unrelated module rewrite and no tuning rationale. Firmware and WASM have
shipped 1.0/1000 all along. It is now runtime-settable if this turns out worse.
The generated header lands in nisps/ml/generated/, not nisps/modes/generated/
where the rest of codegen output lives: training hyperparameters are an ML
fact, and nisps/ml sits below nisps/modes, so emitting them there would make
mlp.hpp include upward. The agent that built this flagged the directory-crossing
rather than hiding it; this is the fix. CI's generated-freshness gate learns the
new directory.
Gates: run-all-tests.sh ALL GREEN — 4/4 ctest, parity PASS (max delta 2.38e-7),
lint clean, manifold typecheck + 17 unit + 33 e2e (which exercise train() and
trainAsync() through a real browser).
Phase 3 (L39, ST13, ST9, ST10).
- L39: deleted codegen/seed-midi-devices.ts. It regenerated — and therefore
silently CLOBBERED — the hand-tuned device schemas it had originally created.
Its purpose is served and git retains it. Confirmed nothing invokes it (the
root package.json that once might have was deleted in Phase 1).
- ST13: moved the provenance artifact out of the repo root. AUDIT CORRECTION:
the finding's proposed destination, schemas/midi_devices/synth-midi-cc.sources.json,
would have BROKEN codegen and CI — generate-midi-devices.ts treats every *.json
directly under that directory as a device template and ajv-validates it, and
the sources file has no device_id/midi/params, so the generator would exit
non-zero. It went to schemas/midi_devices/sources/ instead.
- ST9: schemas/schema.json's own description stated the direction of truth
BACKWARDS. Rewritten: schemas are canonical, codegen emits C++ and TypeScript.
(The audit cited params_notes.md; the real path is schemas/modes/params_notes.md.)
- ST10: codegen/README.md rewritten against current reality — it described
retired playground paths and a phantom temp dir.
Descriptions are never emitted, so this is behaviour-inert; proven by
regenerating all four generated directories byte-identically.
