Repository navigation
feat(dashboard): model-search jobs and comparison view - #37
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 43 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/dashboard-search
branch
from
November 16, 2026 00:47
2763e1d to
ecd078c
Compare
…ce and compare their reports without ever showing holdout scores Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
ng
force-pushed
the
feat/dashboard-search
branch
from
November 16, 2026 00:49
ecd078c to
345ef9f
Compare
ng
marked this pull request as ready for review
November 16, 2026 00:49
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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 B5 (training-11): model-search jobs and a comparison view.
On main d857949 (after #34 and #35).
Jobs. There are three new allow-listed actions, each with the usual schema,
data/path checks and argv-only execution:search-plan:search.py plan SPEC --out NEW [--cache DIR]search-run:search.py run RUN [--jobs N]search-report:search.py report RUNThe run folder must hold a
plan.json, and the run is linked into the run list when the job finishes. There is nofinalaction.finalreads the locked holdout once and exportsmodel/set.json, so it stays a deliberate terminal step. Tests assert that:final--candidate,--access-logor--holdout-finalsearch.pyhas no pod-write routesView. Search in the sidebar lists runs of kind
search. A run shows:With more than one search, an "All searches" table compares their primary results and winners.
Holdout. The view uses only
report.json, which counts holdout reads but holds no holdout scores.final.jsonis never read or served. A test plants a sentinel score infinal.jsonand checks it appears in no/api/runs*response. The view shows only "holdout reads so far" and whether this run has had its final read. Per-night rows aren't displayed at all: the vitals layer shows pooled MAE and ρ only.Test plan
uv run --locked python scripts/validate_knowledge.py,check-vendor.pyuv run --locked python -m unittest discover -s scripts -p 'test_*.py': 290 tests, including the search action schemas and argv, theplan.jsonrequirement, nofinal, and nofinal.jsonin served payloadsnpm run typecheck,node --check scripts/ui/dashboard.js,node --test scripts/test-dashboard.mjs: 32 tests, including row ranking, comparison text, cross-search summary, and a render for scored, final, ready and blocked reports that never shows a planted final scoremake-synthetic-inputs.py+train.pysmoke runsmoke-pod.pysynthetic HTTP pod rundocs/search/synthetic.json, 14 nights), all throughPOST /api/jobs:search-plan,search-runandsearch-reporteach completed and linked the run. Nofinal.jsonwas written. Earlier, against the same inputs, in headless Chrome the Search view showed the primary comparison, the 4-candidate table, the layer scores, "Holdout reads so far: 0" and the limitations, and Report started a job and opened its log.🤖 Generated with Claude Code