Repository navigation
feat(training): model search harness with a fixed protocol and one logged holdout final - #34
Merged
Merged
Conversation
|
Warning Review limit reachedYou've used all free OSS reviews for now. Wait for the free limit to reset to keep reviewing this public repository. Next included review available in 48 minutes. View limit details
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. |
ng
force-pushed
the
feat/derived-snapshot
branch
from
November 16, 2026 00:22
3f88bf5 to
d9e4227
Compare
ng
force-pushed
the
feat/search-harness
branch
from
November 16, 2026 00:22
9d3d300 to
8a4c301
Compare
6 tasks done
ng
force-pushed
the
feat/search-harness
branch
from
November 16, 2026 00:34
8a4c301 to
d1d4be0
Compare
ng
force-pushed
the
feat/derived-snapshot
branch
from
November 16, 2026 00:37
d9e4227 to
4acfc52
Compare
ng
force-pushed
the
feat/search-harness
branch
from
November 16, 2026 00:37
d1d4be0 to
82130c0
Compare
6 tasks done
…ogged holdout final so candidate comparisons stay honest Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
… and how long 24 candidates take Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…from every candidate so search compares on gated inputs only Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…onical-JSON port Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
ng
force-pushed
the
feat/search-harness
branch
from
November 16, 2026 00:44
82130c0 to
d262b1b
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Plan F (
training-15):scripts/search.py, the model search harness. Based on main, after #27 (derived snapshots and model sets).Candidates. A candidate is one choice per node:
liverows, or one replayed engine and profile, named by hashes or by a profileconfig;none, or amovement_model.pymodel with its cross-fitted epochs;podNative/podContextcontract plus the Python groupscontext/movement/arousal/quality, and the modelbalanced_leaf5orhierarchical.Candidates are named
vitals+movementModel+stage, and the space is capped bybudget.maxCandidates.Caches. Each node's output is cached once per version under
data/cache/:vitals/(replay readings),snapshots/(derived databases),features/, andmovementModel/<model12>/<session>.json.planprints the exactpnpm biometrics:local-fit replay --archive … --compare-storedandderive_snapshot.py ingestcommands, andrunrefuses to start. Missing movement epochs get the same treatment, with themovement_model.py epochscommand.Protocol, fixed in code:
MIN_TRAIN_GROUPS;fusion.movement_rows, and a movement model taught on a locked night is refused;primaryis pre-registered; the winner is printed with the candidate count and labelled exploratory.Gated replays. A replayed night is used only if core's training gate passed it, or, when the vitals choice sets
"allowUngated": true, if core could not judge it. A rejected night is excluded for every candidate, with core's reason. The E2 spec setsallowUngated, because most of its nights stored a different profile; the reason is indocs/search.md.Layer scores:
final.
verify-portable.ts) before any locked night is read.holdout.log_access(purpose='final'), scores the locked nights once, and writesfinal.json,model/candidate.jsonandmodel/set.json(modelset-v1; stage pinned by core's hash, vitalsexpected,training.inputsreplayed|live). It never uploads.final.jsonand the terminal only;report.json, which the dashboard reads, holds only the read count (holdout.accesses()).Provenance.
plan.jsonrecords the CLI command, script SHA-256s, input hashes (snapshot, Watch, readings per night, epochs and model, archive cursor), night lists per split, and the protocol with its SHA-256. Runs underdata/are registered in the run catalog as kindsearch, throughrun_history.register_output. The dashboard mapssearchto its own task, so these runs stay out of the sleep list until B5.Docs.
docs/search.mdcovers the spec reference, protocol, caches and the fusion arm mapping. It also has the prepared E2 spec (docs/search/e2-replayed-vs-live.json) and its command sequence; E2 is not run.Reproduction and cost
docs/search/fusion-xfit.jsonlists thefusion.pyarms as candidates, on the same features, snapshot and cross-fitted movement epochs. On every arm, and onpod_live, the walk-forward pooled macro F1 is identical todata/runs/fusion-20261115-xfit/report.jsonto four decimals (difference 0.0000 against the 0.005 tolerance), on the same 1,165 epochs and 24 night groups. The same input and movement-model hashes were checked.--jobs 4(Apple M4): 8.3 s mean and 12 s max per candidate, with 576 forest fits. Vitals were cached; the replay itself is core's and runs once per profile.Test plan
uv run --locked python -m unittest discover -s scripts -p 'test_*.py'(282 tests). New tests cover:excluding[T]);.github/workflows/validate.yml; the synthetic final writes its own access log).stageModelSha256(stageModelSchema.parse(model)): equal.replay --archivelands and the archive is populated (sequence indocs/search.md).No real health data in git or CI; never contacts a pod, uploads or activates.
🤖 Generated with Claude Code