Add claude GitHub actions 1772771924093#5
Merged
Conversation
YigitTurali
added a commit
that referenced
this pull request
May 3, 2026
Fix #2 — _planner_exec.py: GSWPlan now validates per-kind constraint shape at plan-emit time. argmax/argmin require non-empty candidate_entity_ids and sort_by_blank_ids; equals/gt/lt require both left_ref and right_ref; derived/in_list require non-empty args_blanks. Catches q227-style "Winner B" placeholder leaks where the orchestrator submitted whatever string was in scope because the executor couldn't compute the target. Fix #5 — _planner_emit.py: replaced the single repair retry with a loop of up to MAX_REPAIR_ATTEMPTS=3 calls with progressively sharper prompts. After all repairs fail, raise PlanEmitError(kind="exhausted") and the orchestrator stamps planner_failed=True with empty pred instead of falling through to the flat ReAct adapter, which used to hallucinate sentinels like 'Zero.' / 'Data not available.' on planner-side failures. Fix #3 (cap raise 2→4) was attempted on top but reverted: 2x100Q probe showed wrong_synthesis grew 12.5 → 18 mean as the extra cap budget converted "give up empty" into "submit stale wrong value". Cap stays at 2. Tests: +8 constraint-shape validators (test_planner_exec.py) + 8 multi-retry repair tests (test_planner_emit.py, new). Three existing cap-related orchestrator tests pin MAX_PLAN_UPDATES_PER_RUN=2 via monkeypatch so they remain robust to future cap changes. 137 tests green (was 121). FRAMES dev-100 mean (gpt-4.1 judge, hybrid retriever, 2 runs each): Baseline: 70.0 (73, 67) Phase-3.6.2: 68.5 (69, 68) Within ±3-pt LLM stochasticity band; deterministic q227/q83/q118 flips ✗→✓ confirmed in both runs. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
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
1 participant
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.
No description provided.