benchmark: RS-LoCoMo-Full-v5-strong — the strong-agent protocol variant - #176
Merged
Conversation
fazpu
enabled auto-merge (squash)
July 29, 2026 16:22
Coverage report
This PR does not seem to contain any modification to coverable code. |
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
The owner-approved strong-agent variant (step 3 of the measurement plan), motivated by three controlled smoke passes on a healthy GLM-5.2 store (session recall 0.5, gold facts at rank 1) scoring 1–2/8 with the pinned
gpt-4o-minianswer agent looping past the tool-call limit or returning invalid responses — the harness agent is the floor, not the memory.A typed frozen protocol registry:
full-v5(default, bit-identical to today) andfull-v5-strong(answer agentopenai/gpt-5.6-luna; judge, prompts, schemas, tool catalog, budgets, temperature identical).--protocolselects at prepare; downstream stages read the pin from the run. The two fingerprints differ (answer model is in the derivation) — scores are never comparable across protocols, per the dated design note.Review: Grok APPROVE WITH NOTES (none ship-blocking; old run dirs need re-preparing before reuse — moot, new runs prepare fresh). Final review by Claude. Fast gate local (53 passed); full matrix in CI.