Comparison-scoped evidence
Record change explorer
Reference → Candidate · 5 matched records
Filter record changes
5 shown| Record ID | Transition | Reference category | Candidate category | Changed metrics | Evidence |
|---|---|---|---|---|---|
| THQA-001 | Stable pass | PASS | PASS | No metric delta | Inspect evidenceReference retrieved IDs: doc-th-001, doc-th-999 Candidate retrieved IDs: doc-th-001, doc-th-999 Relevant IDs: doc-th-001 Reference metrics: {"hit_rate_at_5":1,"mrr":1,"ndcg_at_5":1,"precision_at_5":0.2,"recall_at_5":1} Candidate metrics: {"hit_rate_at_5":1,"mrr":1,"ndcg_at_5":1,"precision_at_5":0.2,"recall_at_5":1} |
| THQA-002 | Persistent category | SHOULD_ABSTAIN | SHOULD_ABSTAIN | No metric delta | Inspect evidenceReference retrieved IDs: Not included Candidate retrieved IDs: doc-th-999 Relevant IDs: Not included Reference metrics: {"hit_rate_at_5":0,"mrr":0,"ndcg_at_5":0,"precision_at_5":0,"recall_at_5":0} Candidate metrics: {"hit_rate_at_5":0,"mrr":0,"ndcg_at_5":0,"precision_at_5":0,"recall_at_5":0} |
| THQA-003 | Stable pass | PASS | PASS | No metric delta | Inspect evidenceReference retrieved IDs: doc-th-003, doc-th-002 Candidate retrieved IDs: doc-th-003, doc-th-002 Relevant IDs: doc-th-002, doc-th-003 Reference metrics: {"hit_rate_at_5":1,"mrr":1,"ndcg_at_5":1,"precision_at_5":0.4,"recall_at_5":1} Candidate metrics: {"hit_rate_at_5":1,"mrr":1,"ndcg_at_5":1,"precision_at_5":0.4,"recall_at_5":1} |
| THQA-004 | Introduced failure | PASS | RETRIEVAL_MISS | Hit Rate@5 -1 MRR -1 nDCG@5 -1 Precision@5 -0.2 Recall@5 -1 | Inspect evidenceReference retrieved IDs: doc-th-004 Candidate retrieved IDs: doc-th-999 Relevant IDs: doc-th-004 Reference metrics: {"hit_rate_at_5":1,"mrr":1,"ndcg_at_5":1,"precision_at_5":0.2,"recall_at_5":1} Candidate metrics: {"hit_rate_at_5":0,"mrr":0,"ndcg_at_5":0,"precision_at_5":0,"recall_at_5":0} |
| THQA-005 | Stable pass | PASS | PASS | nDCG@5 -0.387 Precision@5 -0.2 Recall@5 -0.5 | Inspect evidenceReference retrieved IDs: doc-th-005, doc-th-006 Candidate retrieved IDs: doc-th-005, doc-th-999 Relevant IDs: doc-th-005, doc-th-006 Reference metrics: {"hit_rate_at_5":1,"mrr":1,"ndcg_at_5":1,"precision_at_5":0.4,"recall_at_5":1} Candidate metrics: {"hit_rate_at_5":1,"mrr":1,"ndcg_at_5":0.6131471927654584,"precision_at_5":0.2,"recall_at_5":0.5} |