兩項目評估報告
- EvalRun: #124（suite #69
auto-89ac7fad-r1-053958761）
- Target: #2 production-canary
- Cases: 30（N=1, attempts=30/30）
- Generated: 2026-05-16T13:33:54+08:00
本報告含兩個獨立評估項目。項目一 評估 bot 進入「知識與產品查詢」fallback
後的撈取 + 回答能力；項目二 評估 bot 跨全 enabled scenarios 的 routing →
tool → answer 三段 funnel 健康度。
項目一: 知識庫精準度
評估範圍 (prerequisite: 進對 scenario)
- 進對 scenario 的 attempts: 10 / 30 (33.3%)
- 只對這些 attempts 算 retrieval + answer
- 沒進對 scenario 的 attempts → 排除（routing 失敗不該污染 KB 訊號）
Per-scenario 評估
| Scenario |
qualifying / total attempts |
retrieval_relevance |
answer_correctness |
| 知識與產品查詢 |
10 / 10 |
❌ 0.0% |
✅ 100.0% |
待修方向（worst-3）
- 知識與產品查詢 × retrieval_relevance = 0.0% — bot 進對 scenario 但沒呼叫 search tool — 流程 / prompt 問題（不是 KB content gap）
- 知識與產品查詢 × answer_correctness = 100.0% — 回答內容有問題 — 細看 hallucination_rate vs answer_quality 哪邊低
項目二: 情境調用與完成
整體 funnel（全 scenarios 加總）
| Stage |
Pass count |
% of total |
| Total attempts |
20 |
100.0% |
| Step 1: scenario_routing |
5 |
25.0% |
| Step 2: + tool_calling |
5 |
25.0% |
| Step 3: + answer |
0 |
0.0% |
Per-scenario funnel
| Scenario (n_attempts) |
Step 1 (routing) |
Step 2 (tool) |
Step 3 (answer) |
end-to-end |
| early_return (9) |
❌ 0/9 (0.0%) |
❌ 0/9 (0.0%) |
❌ 0/9 (0.0%) |
❌ 0/9 (0.0%) |
| 知識與產品查詢 (4) |
✅ 4/4 (100.0%) |
✅ 4/4 (100.0%) |
❌ 0/4 (0.0%) |
❌ 0/4 (0.0%) |
| 訂單查詢 (7) |
❌ 1/7 (14.3%) |
❌ 1/7 (14.3%) |
❌ 0/7 (0.0%) |
❌ 0/7 (0.0%) |
Drop-off 最大的 5 個 scenario
- early_return drop at step1 (-100.0pp)
- 知識與產品查詢 drop at step3 (-100.0pp)
- 訂單查詢 drop at step1 (-85.7pp)
Audit
Reproduce combined report: bin/rails runner "puts Eval::EvaluationReport.call(run: EvalRun.find(124))"
Or fetch each item separately:
bin/rails runner "puts Eval::KbAccuracyReport.call(run: EvalRun.find(124))"bin/rails runner "puts Eval::ScenarioFunnelReport.call(run: EvalRun.find(124))"
Per-Scenario × Per-Dim — Run #124
Suite: GH Hotel (bulk R1) · scenarios: 3 · dims: 5 · populated cells: 13/15
| Scenario |
Scenario |
Tool |
Retrieval |
Faith |
AnsQ |
| early_return |
0.0% [0.0–0.0] (n=9) |
— |
11.1% [0.0–33.3] (n=9) |
100.0% [100.0–100.0] (n=3) |
68.1% [47.8–86.7] (n=9) |
| 知識與產品查詢 |
100.0% [100.0–100.0] (n=14) |
100.0% [100.0–100.0] (n=14) |
35.7% [14.3–57.1] (n=14) |
66.7% [36.7–90.0] (n=10) |
60.5% [46.4–75.2] (n=14) |
| 訂單查詢 |
57.1% [14.3–85.7] (n=7) |
14.3% [0.0–42.9] (n=7) |
— |
83.3% [50.0–100.0] (n=6) |
79.0% [65.7–90.0] (n=7) |
Worst-3 cells (lowest primary score)
- early_return × Scenario · 0.0% (n=9) · lowest sub_metric:
scenario_precision
- early_return × Retrieval · 11.1% (n=9) · lowest sub_metric:
knowledges_source_recall
- 訂單查詢 × Tool · 14.3% (n=7) · lowest sub_metric:
tools_precision