run · run_wrjeqlxp5nq1yy9e
RRF 融合(BM25+图像+硬度)
Zhuaiz/elastic-mineral-hackson@bdd7b05 ↗
on Mineral Identification from Field Observations
total score
0.920
engine
rrf(bm25+image) → qwen-plus
cases
46 passed · 4 failed
latency
—
cost
—
submitted
2026-07-11 05:54:53
Run summary
- score
- 0.920
- total cases
- 50
- passed cases
- 46
- skipped cases
- 0
Per-case results
46 pass4 fail
reasonanswer 'rhodonite' ≠ 'andalusite'
answer
rhodonite
expected
andalusite
- id
- andalusite
- category
- orthorhombic
- difficulty
- hard
reasonanswer 'grossular' ≠ 'andradite'
answer
grossular
expected
andradite
- id
- andradite
- category
- isometric
- difficulty
- hard
reasonanswer 'albite' ≠ 'microcline'
answer
albite
expected
microcline
- id
- microcline
- category
- triclinic
- difficulty
- hard
reasonanswer 'diopside' ≠ 'spodumene'
answer
diopside
expected
spodumene
- id
- spodumene
- category
- monoclinic
- difficulty
- hard