兩項目評估報告
- EvalRun: #298（suite #150
auto-3eb1e09f-r1-053959834）
- Target: #2 production-canary
- Cases: 30（N=1, attempts=30/30）
- Generated: 2026-05-16T14:15:35+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 |
❌ 47.5% |
✅ 100.0% |
待修方向（worst-3）
- 知識與產品查詢 × retrieval_relevance = 47.5% — 撈到的 KB chunks 不夠相關 — KB content gap，跟 Neptune team 提 enrichment
- 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 |
6 |
30.0% |
| Step 2: + tool_calling |
6 |
30.0% |
| Step 3: + answer |
3 |
15.0% |
Per-scenario funnel
| Scenario (n_attempts) |
Step 1 (routing) |
Step 2 (tool) |
Step 3 (answer) |
end-to-end |
| early_return (10) |
❌ 0/10 (0.0%) |
❌ 0/10 (0.0%) |
❌ 0/10 (0.0%) |
❌ 0/10 (0.0%) |
| 知識與產品查詢 (5) |
✅ 5/5 (100.0%) |
✅ 5/5 (100.0%) |
❌ 2/5 (40.0%) |
❌ 2/5 (40.0%) |
| 訂單查詢 (5) |
❌ 1/5 (20.0%) |
❌ 1/5 (20.0%) |
❌ 1/5 (20.0%) |
❌ 1/5 (20.0%) |
Drop-off 最大的 5 個 scenario
- early_return drop at step1 (-100.0pp)
- 訂單查詢 drop at step1 (-80.0pp)
- 知識與產品查詢 drop at step3 (-60.0pp)
Audit
Reproduce combined report: bin/rails runner "puts Eval::EvaluationReport.call(run: EvalRun.find(298))"
Or fetch each item separately:
bin/rails runner "puts Eval::KbAccuracyReport.call(run: EvalRun.find(298))"bin/rails runner "puts Eval::ScenarioFunnelReport.call(run: EvalRun.find(298))"
Per-Scenario × Per-Dim — Run #298
Suite: 克莉絲丹 香氛客服 (bulk R1) · scenarios: 3 · dims: 5 · populated cells: 15/15
| Scenario |
Scenario |
Tool |
Retrieval |
Faith |
AnsQ |
| early_return |
0.0% [0.0–0.0] (n=10) |
0.0% (n=1) |
88.9% [66.7–100.0] (n=9) |
100.0% [100.0–100.0] (n=9) |
89.3% [77.0–98.7] (n=10) |
| 知識與產品查詢 |
100.0% [100.0–100.0] (n=15) |
100.0% [100.0–100.0] (n=15) |
70.0% [46.7–90.0] (n=15) |
88.9% [73.3–100.0] (n=15) |
88.0% [76.0–96.9] (n=15) |
| 訂單查詢 |
80.0% [40.0–100.0] (n=5) |
20.0% [0.0–60.0] (n=5) |
40.0% [20.0–50.0] (n=5) |
100.0% [100.0–100.0] (n=4) |
94.0% [88.0–99.3] (n=5) |
Worst-3 cells (lowest primary score)
- early_return × Scenario · 0.0% (n=10) · lowest sub_metric:
scenario_precision
- early_return × Tool · 0.0% (n=1) · lowest sub_metric:
tools_precision
- 訂單查詢 × Tool · 20.0% (n=5) · lowest sub_metric:
tools_precision