feat(compact): drop the forced summarize tool_choice after tool results - #2330
Draft
mikasenghaas wants to merge 1 commit into
Draft
feat(compact): drop the forced summarize tool_choice after tool results#2330mikasenghaas wants to merge 1 commit into
mikasenghaas wants to merge 1 commit into
Conversation
Co-Authored-By: Claude Fable 5 <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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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
tool_choice={"function": {"name": "summarize"}}on the compact harness's post-tool-result model call —summarizestays the only advertised tool there and the system prompt still requires saving notes, but the model now has to choose the call itselfsummarizecall after a tool result still ends the rollout with no answer, so protocol violations become an observable (and trainable) negative signal instead of being mechanically preventedchat()accordingly (thetool_choiceparameter had no other caller)Verification
16-task wikispeedia eval (eval split, Qwen/Qwen3-4B on a local vLLM), this branch vs the forced protocol:
wiki_click_link×11 + one more later,wiki_go_back×3) instead ofsummarize— even with onlysummarizeadvertised, the server-side parser extracts the call and the protocol ends the rollout. Exactly one rollout followed the click→summarize loop unforced (7 turns) and then answered wrong. Trace dashboard: https://app.primeintellect.ai/dashboard/evaluations/wk9qajkt4afdnsx95enhgqooThis reproduces the GLM-5.2 finding that motivated the forcing (advertising
summarizealone is not enough). As-is, the unforced protocol yields no successful rollouts for Qwen3-4B — fine as a hard negative training signal, unusable for evals. If we want this behavior switchable, aforce_summarize: bool = Trueknob onCompactingHarnessConfigis probably the better shape than removing the forcing outright.Breaking
summarizecall (or a plain-text finish) now end without an answer instead of being steered intosummarize. Weaker instruction-followers will score lower under this harness.🤖 Generated with Claude Code