Run artifact · miracl retrieval

miracl-th-mini

System evalops-bm25-mini evaluated against synthetic-fixture-v1. This page presents only approved public evidence.

Verification: VERIFIEDData: SYNTHETIC_FIXTUREClaim: INTEGRATION_ONLY

Artifact ID

demo-miracl-th-mini-v1

Claim scope: Integration Only.

Aggregate values

Metrics

Hit Rate@5

1

k preserved in metric name · demo-miracl-th-mini-v1

MRR

1

k preserved in metric name · demo-miracl-th-mini-v1

nDCG@5

1

k preserved in metric name · demo-miracl-th-mini-v1

Precision@5

0.24

k preserved in metric name · demo-miracl-th-mini-v1

Recall@5

1

k preserved in metric name · demo-miracl-th-mini-v1

Failure summary

0 labeled failures

No public failure records are included for this artifact. This is not proof of perfect model performance on every possible input.

Provenance

Dataset
miracl-th-mini · synthetic-fixture-v1
System
evalops-bm25-mini
Benchmark context
miracl · th · dev
Evaluation
miracl retrieval
Recorded
15 Jan 2026
Top-k
5
Evaluator
retrieval: deterministic-metrics-v1
Retriever
bm25 · bm25-local-v1

Record-level evidence

5 approved records
Approved record-level retrieval evidence
RecordStatusMetricsRetrieved IDsRelevant IDs
1Passk=1: Hit Rate@1 1 · MRR 1; k=5: Hit Rate@5 1 · MRR 1d1, d3, d2, d5, d4d1
2Passk=1: Hit Rate@1 1 · MRR 1; k=5: Hit Rate@5 1 · MRR 1d3, d6, d5, d4, d1d3
3Passk=1: Hit Rate@1 1 · MRR 1; k=5: Hit Rate@5 1 · MRR 1d3, d1, d2, d6, d4d1, d3
4Passk=1: Hit Rate@1 1 · MRR 1; k=5: Hit Rate@5 1 · MRR 1d4, d1, d5, d2, d3d4
5Passk=1: Hit Rate@1 1 · MRR 1; k=5: Hit Rate@5 1 · MRR 1d2, d5, d1, d6, d4d2

Limitations

  • BM25 is a local deterministic baseline; no model-superiority claim is supported.
  • Synthetic MIRACL-shaped Thai mini fixture; not official MIRACL benchmark data or a leaderboard result.

Contract notes

Metric names and denominators are inherited from the deterministic evaluator output. No combined quality score, missing denominator, raw response, prompt, corpus text, or hidden reasoning is inferred here.

Schema public-evidence-v1 · Verification text remains visible alongside color.