Repository navigation
Routing: SessionStart hook lifts Haiku on bare questions 66% to 92%; held-out sets 3 and 4 - #7
Merged
Merged
Conversation
… routing Two description rewordings (assume-AgentBase sentence, general-question clause) were measured on a fresh 36-case set and did not move Haiku 5.5 (75% -> 76%), so descriptions are unchanged; the finding and the new set are documented. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
… held-out set 4 Haiku 5.5 on bare questions 66% -> 92% (fresh set), 64% -> 95% (held-out2), tuning 95% -> 100%. The last wording tweak (skip agents on another named stack) is not re-measured: weekly limit reached. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
khanhphamj
changed the title
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Two description rewordings did not help Haiku 5.5 on bare questions (75% to 76%) and were reverted. What worked: a SessionStart hook (Claude Code only) that says "my agent" means a GreenNode AgentBase agent unless another framework or platform is named.
Haiku 5.5, 2 runs per case: fresh held-out4 66% to 92%, held-out2 64% to 95%, tuning 95% to 100%; all 8 other-stack cases stayed none. Sonnet 5.5 on held-out4: 98%, one false positive (CrewAI) that the final wording targets.
Not verified: the final wording (after the CrewAI fix) has no score, the Claude Code weekly limit was reached. Rerun held-out4 on Sonnet and Haiku and the tuning set on Sonnet after it resets (Nov 2).
🤖 Generated with Claude Code