metric

`n_returned`

better: neutral: a description
kind: descriptive: says what the output was, never whether it was good
decides: no goal
weighed by: no goal
recorded on: scifact_200

01what it measures, and how

How many distinct document ids the ranking returned.

A project explains its own metric at length with METRIC_DESCRIPTIONS = {"n_returned": {"summary": "…", "details": "…"}}.

02on scifact_200

scored against silverscored by a judgewrote the reference it is scored againstscored against a label set, not a reference setunmarked rows are scored against gold

arm	mean n_returned	items	scored against
sf_bge_small_n200_v1	100.0	200	gold/scifact_200 gold
sf_bm25_n200_v1	100.0	200	gold/scifact_200 gold
sf_deepseek_m_n200_v1	100.0	200	gold/scifact_200 gold
sf_deepseek_s_n200_v1	100.0	200	gold/scifact_200 gold
sf_e5_base_n200_v1	100.0	200	gold/scifact_200 gold
sf_first_k_n200_v1	100.0	200	gold/scifact_200 gold
sf_gemma_l_n200_v1	100.0	200	gold/scifact_200 gold
sf_gemma_m_n200_v1	100.0	200	gold/scifact_200 gold
sf_gemma_s_n200_v1	100.0	200	gold/scifact_200 gold
sf_glm_m_n200_v1	100.0	200	gold/scifact_200 gold
sf_glm_s_n200_v1	100.0	200	gold/scifact_200 gold
sf_llama_l_n200_v1	100.0	200	gold/scifact_200 gold
sf_llama_m_n200_v1	100.0	200	gold/scifact_200 gold
sf_llama_s_n200_v1	100.0	200	gold/scifact_200 gold
sf_minilm_n200_v1	100.0	200	gold/scifact_200 gold
sf_mistral_s_n200_v1	100.0	200	gold/scifact_200 gold
sf_mpnet_n200_v1	100.0	200	gold/scifact_200 gold
sf_qwen_s_n200_v1	100.0	200	gold/scifact_200 gold
sf_random_n200_v1	100.0	200	gold/scifact_200 gold
sf_glm_s_bm25first_n200_v1 · pair	100.0	200	gold/scifact_200 gold
sf_glm_s_e5first_n200_v1 · pair	100.0	200	gold/scifact_200 gold
sf_qwen_s_bm25first_n200_v1 · pair	100.0	200	gold/scifact_200 gold
sf_qwen_s_e5first_n200_v1 · pair	100.0	200	gold/scifact_200 gold
sf_llama_l_n200_v1 · repeats	100.0	200	gold/scifact_200 gold
sf_llama_s_n200_v1 · repeats	100.0	200	gold/scifact_200 gold
sf_qwen_s_n200_v1 · repeats	100.0	200	gold/scifact_200 gold