General & model settings一般與模型設定
Site information, Gemini models and generation parameters. Changes take effect immediately — just refresh the front end.站台資訊、Gemini 模型與生成參數。修改後立即生效,前台重新整理即可看到。
RAG retrieval settingsRAG 檢索參數
Control how many records are retrieved, how they are scored, and how much content is injected. Use the test tool below to verify changes immediately.控制「檢索幾筆、如何計分、注入多少內容」。調整後可用下方的測試工具立即驗證。
Retrieval test檢索測試
Enter a query to see which records the current settings retrieve (no AI generation is called).輸入一段查詢,看看目前參數會檢索出哪些紀錄(不會呼叫 AI 生成)。
Prompt editor提示詞編輯
You can use {{variable}} placeholders. Available variables: pe, subject, grade, tiers, objectives, objectives_block, tier_evidence, source_material, self_eval, media_requests, media_block, rag_context, citation_format. After editing, use "Preview rendered result" to check it first.可使用 {{變數}} 佔位。可用變數:pe, subject, grade, tiers, objectives, objectives_block, tier_evidence, source_material, self_eval, media_requests, media_block, rag_context, citation_format。修改後建議先按「預覽渲染結果」確認。
{{media_block}} is generated by the system from the extension materials the teacher selects — do not remove it from the Develop-stage template.{{media_block}} 由系統依老師勾選的延伸素材自動產生格式要求,請勿從開發階段模板中移除。
Learning target list學習目標清單
This is the data source for the "core learning target" drop-down in the front end. The system already includes NGSS performance expectations (PE) and Taiwan 108 Curriculum natural-science learning performances; schools can import their own targets or supplementary codes here, 前台「課程核心學習目標」下拉選單的資料來源。系統已內建 NGSS 表現期望(PE)與臺灣 108 課綱自然科學學習表現;學校可在此匯入校訂目標或補充代碼,without modifying the built-in default list不會更動程式內建的預設清單 — you can always use "Restore built-in list".,隨時可以「回復內建」。
Current status目前狀態
Import & overwrite匯入覆寫
Each row needs at least the 每列至少需要 code field; recognized field names: 欄位;可辨識的欄位名稱:code, zh/text (Chinese statement), en (original English text), grade/stage, domain, topic, group.code(代碼)、zh/text(中文敘述)、en(英文原文)、grade/stage(年段)、domain(領域)、topic(主題)、group(類別)。
List preview清單預覽
Select a list to preview its first 20 entries.選擇清單後可預覽前 20 筆。
Data import資料匯入
CSV is supported (UTF-8 or Big5). The current flow is CSV import; a Google Sheets API connection can be added later.支援 CSV(UTF-8 或 Big5 皆可)。目前為 CSV 匯入流程,之後可改接 Google Sheets API。
1. Choose a source1. 選擇來源
In Google Sheets, choose File → Download → Comma-separated values (.csv).建議直接從 Google Sheets 選「檔案 → 下載 → 逗號分隔值 (.csv)」。 Download sample CSV下載範例 CSV
2. Import options2. 匯入選項
3. Field mapping3. 欄位對應
Column headers are matched automatically after upload; assign any unmatched fields manually here.上傳後會自動比對欄位標題;若有未對應的欄位,可在此手動指定。
No data loaded yet.尚未載入資料。
Import history匯入歷程
Record management紀錄管理
Browse, edit and delete imported teaching records.瀏覽、編輯、刪除已匯入的教學紀錄。
Edit record編輯紀錄
Vector index maintenance向量索引維護
Semantic retrieval needs a vector index. Building the index calls the Gemini embedding model, so an API key is required (used for this session only).語意檢索需要向量索引。建立索引時會呼叫 Gemini 嵌入模型,需提供 API Key(僅本次使用)。
Build / update index建立/更新索引
Used only for this index build; nothing is written to any server file.只用於本次建立索引;不會寫入伺服器任何檔案。
Backup management備份管理
Every import, deletion and bulk edit is backed up automatically; the latest 10 backups are kept.每次匯入、刪除與大量修改都會自動備份,最多保留 10 份。