fix: follow-up 追問對準實質與情緒核心,不再咬語氣詞#26
Merged
Merged
Conversation
pilot 真人給情緒很重的回答(工作不開心/付出可有可無/不想做了), bot 卻咬住語氣詞「有點」追問「這個『有點』背後是什麼感覺」。 根因:generate_follow_up 的 prompt 只傳 rule code「R1」、從不傳規則 語意 → LLM 無從依規則發問;且「Keep their wording」無「對準實質」 指引 → 咬住任何可引用的字,包括語氣緩衝詞。 Gemini + Codex 雙審收斂: - 新增 FOLLOW_UP_RULE_PROMPTS:把 R1-R5 規則語意傳進 prompt - prompt 加 salience priority(情緒 > 意義/自我價值 > 決定/意圖 > 具體事件 > 用字) - 明列 hedge words(有點/好像/可能/大概/其實/就是),明令不得追問 語氣詞的「背後」 Constraint: pilot hotfix;不改 select_rule、不擴張 ontology Directive: 這是更大根因「bot 不知道自己的目的與累積脈絡」的局部修正; 根因修法(共用 purpose + cross-sitting context)另案處理 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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
摘要
pilot hotfix(接續 #23 #24 #25)。真人 pilot 給情緒很重的回答「我覺得我工作得很不開心,我的付出好像可有可無,有點不太想做了」,bot 回「你說『有點不太想做了』——這個『有點』背後,實際上是什麼感覺?」—— 咬住語氣緩衝詞「有點」,略過情緒核心。使用者:「它連人話都聽不懂。」
根因
generate_follow_up的 prompt 只傳 rule code(字串"R1"),從不解釋 R1-R5 的語意 →select_rule精心選的規則在生成端被丟掉,LLM 只能瞎掰。加上「Keep their wording」無「對準實質」指引 → LLM 抓到任何可引用的字,包括語氣詞。修法(Gemini + Codex 雙審收斂)
FOLLOW_UP_RULE_PROMPTS:把 R1-R5 各自的規則語意(事實→模式、模式→原則、原則→反例、抽象→具體、重複→換角度)傳進 prompt。驗證
新測試 3 個:
FOLLOW_UP_RULE_PROMPTS涵蓋全規則;prompt 組裝含 rule instruction + anti-hedge 指令(mock create_message);R5 早 return 不呼叫 LLM。review chain
Gemini(多方案)+ Codex(工程穩健度)→ 強烈收斂 → Codex 實作 → Claude 獨立驗證(278 全綠)。
注意
🤖 Generated with Claude Code