兩項目評估報告
- EvalRun: #152(suite #87
auto-f2e2f75c-r1-053958978)
- Target: #2 production-canary
- Cases: 30(N=1, attempts=30/30)
- Generated: 2026-05-16T13:33:38+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 |
❌ 43.6% |
✅ 100.0% |
待修方向(worst-3)
- 知識與產品查詢 × retrieval_relevance = 43.6% — 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 |
11 |
55.0% |
| Step 2: + tool_calling |
11 |
55.0% |
| Step 3: + answer |
4 |
20.0% |
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 (10) |
✅ 10/10 (100.0%) |
✅ 10/10 (100.0%) |
❌ 3/10 (30.0%) |
❌ 3/10 (30.0%) |
| 轉接真人客服 (1) |
✅ 1/1 (100.0%) |
✅ 1/1 (100.0%) |
✅ 1/1 (100.0%) |
✅ 1/1 (100.0%) |
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 (-70.0pp)
- 轉接真人客服 drop at step1 (-0.0pp)
Audit
Reproduce combined report: bin/rails runner "puts Eval::EvaluationReport.call(run: EvalRun.find(152))"
Or fetch each item separately:
bin/rails runner "puts Eval::KbAccuracyReport.call(run: EvalRun.find(152))"
bin/rails runner "puts Eval::ScenarioFunnelReport.call(run: EvalRun.find(152))"
Per-Scenario × Per-Dim — Run #152
Suite: JV3C客服小幫手 (bulk R1) · scenarios: 3 · dims: 5 · populated cells: 12/15
| Scenario |
Scenario |
Tool |
Retrieval |
Faith |
AnsQ |
| early_return |
0.0% [0.0–0.0] (n=9) |
— |
100.0% [100.0–100.0] (n=9) |
100.0% [100.0–100.0] (n=9) |
92.6% [86.7–97.4] (n=9) |
| 知識與產品查詢 |
100.0% [100.0–100.0] (n=20) |
100.0% [100.0–100.0] (n=20) |
85.0% [75.0–95.0] (n=20) |
78.9% [63.2–93.0] (n=19) |
72.2% [57.2–85.2] (n=20) |
| 轉接真人客服 |
100.0% (n=1) |
— |
— |
100.0% (n=1) |
100.0% (n=1) |
Worst-3 cells (lowest primary score)
- early_return × Scenario · 0.0% (n=9) · lowest sub_metric:
scenario_precision
- 知識與產品查詢 × AnsQ · 72.2% (n=20) · lowest sub_metric:
critical_coverage
- 知識與產品查詢 × Faith · 78.9% (n=19) · lowest sub_metric:
rule_compliance