BENCHMARK · MESA 0.9.0
MESA 0.9.0 benchmark report
Tuned BGE-M3 retrieved the correct passage in the first five results on 62.0% of answerable questions. The offline Jina experiment reached 76.0% with a separate article-aware pipeline.
62.0%
Hit@5
Correct passage in the first five results
39.7%
Hit@1
Correct passage in the first result
0.500
MRR
Ranking of the first correct passage
MESA 0.9.0 · BGE-M3, tuned vector weight (S4)
Search results
October 29, 2026 · 141 documents · 121 / 145 answerable test questions
| Configuration | Hit@5 | Hit@1 | MRR | Hit@10 | Doc@1 | Doc@5 | p50 | p95 | Hardware |
|---|---|---|---|---|---|---|---|---|---|
| Keyword search (Mode A)20261029T190820Z | 33.1% | 22.3% | 0.293 | 38.8% | 36.4% | 60.3% | 0.96 s | 1.93 s | Colab T4 |
| BGE-M3, equal weights (S0)20261029T234645Z_exp | 51.2% | 34.7% | 0.431 | 60.3% | 58.7% | 70.3% | 1.91 s | 3.49 s | Colab A100 |
| BGE-M3, tuned vector weight (S4)20261029T234645Z_exp | 62.0% | 39.7% | 0.500 | 70.3% | 60.3% | 75.2% | 1.87 s | 3.29 s | Colab A100 |
| Configuration | Hit@5 | Hit@1 | MRR | Hit@10 | Doc@1 | Doc@5 | p50 | p95 | Hardware |
|---|---|---|---|---|---|---|---|---|---|
| Article chunks + offline Jina20261029T234645Z_exp | 76.0% | 57.0% | 0.654 | 80.2% | 69.4% | 82.6% | — | — | Colab A100 |
- Mode A ran on a T4. The S0 and S4 rows use the A100 reruns from the article-chunking study. The original T4 embedding runs are separate measurements.
- The Jina reranking ran offline over saved results. It is reported separately from built-in configurations and excluded from the comparison.
Results by question type
Hit@5 per question type. Darker cells mean the right passage was found more often.
| Question type | Keyword search (Mode A) | BGE-M3, equal weights (S0) | BGE-M3, tuned vector weight (S4) | Article chunks + offline Jina |
|---|---|---|---|---|
| Exact wording (36) | 66.7% | 86.1% | 86.1% | 94.4% |
| Paraphrase (39) | 10.3% | 28.2% | 56.4% | 79.5% |
| Cross-document (17) | 11.8% | 47.1% | 47.1% | 70.6% |
| Number, date or docket (29) | 34.5% | 41.4% | 48.3% | 51.7% |
Time until searchable
| Configuration | Total time | Chunks | Hardware |
|---|---|---|---|
| BGE-M3, equal weights (S0) | 29.7 min | 3,480 | Colab A100 |
Download results
Report and aggregate metrics (JSON) ↓How we measured
Tier M contains 141 Turkish laws, regulations and court decisions. A language model generated questions from the corpus. A separate model and rule checks reviewed them. Configurations were chosen on 101 development questions before evaluation on 145 held-out questions. Retrieval scores use the 121 answerable questions.
What do the metrics mean?
- Hit@1 · Hit@5 · Hit@10
- Share of questions with the correct passage in the first 1, 5 or 10 results.
- MRR
- Average reciprocal rank of the first correct passage. A score of 1 means it always came first.
- Doc@1 · Doc@5
- Share of questions with the correct document among the first 1 or 5 results.
- p50 · p95
- Half of queries finished within p50 and 95% within p95. Both are measured in seconds.
Scope and limitations
The test measures whether MESA retrieves the correct passage or document. The 24 questions with no answer in the corpus are outside the retrieval scores. The corpus is small and its decisions come from the Court of Cassation. The questions are synthetic.
