Evaluation & Quality Benchmarks
Automated quality regression testing for retrieval recall, precision, MRR, and generation accuracy.
Pass Rate
96.4%
+1.2% vs baseline
Recall@5
94.2%
Top 5 candidates
Precision@5
89.1%
Relevant density
MRR Score
0.915
Mean Reciprocal Rank
p95 Latency
134 ms
Per evaluation query
Retrieval Strategy Performance Comparison
Benchmarking Recall@5, Precision@5, and MRR (%) across pipeline configurations
Historical Benchmark Runs
| Run Name | Dataset | Samples | Pass Rate | Recall@5 | MRR | Run Time |
|---|---|---|---|---|---|---|
| RAG Retrieval Quality Suite v2.4 | tech_docs_ground_truth_500.json | 500 | 96.4% | 94.2% | 0.915 | 2026-07-26 15:30 |
| Finance QA Benchmark (Hybrid vs Dense) | finance_sec_filings_1000.json | 1000 | 92.1% | 89.5% | 0.864 | 2026-07-25 11:10 |
| Ollama Llama 3 8B Local Embeddings Test | tech_docs_ground_truth_500.json | 500 | 88% | 85.1% | 0.820 | 2026-07-23 20:45 |