eval-harness

metric

recall

better
higher
kind
quality: measured against a reference or by a judge
decides
no goal
weighed by
no goal
recorded on
few_nerd_280

01what it measures, and how

Typed recall: the share of the gold entities found with the right type.

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

02on few_nerd_280

gold/few_nerd_280fn_span_marker_n200_v1: 0.774fn_openai_m_n200_v1: 0.719fn_anthropic_l_n200_v1: 0.707fn_qwen_m_n200_v1: 0.705fn_openai_l_n200_v1: 0.704fn_qwen_s_n200_v1: 0.702fn_gemma_m_n200_v1: 0.7fn_deepseek_m_n200_v1: 0.681fn_anthropic_m_n200_v1: 0.678fn_gemma_l_n200_v1: 0.676fn_qwen_l_n200_v1: 0.663fn_mistral_m_n200_v1: 0.655fn_glm_m_n200_v1: 0.655fn_openai_s_n200_v1: 0.651fn_llama_l_n200_v1: 0.647fn_deepseek_l_n200_v1: 0.639fn_deepseek_s_n200_v1: 0.639fn_anthropic_s_n200_v1: 0.63fn_mistral_l_n200_v1: 0.63fn_gemma_s_n200_v1: 0.629fn_llama_m_n200_v1: 0.621fn_llama_s_n200_v1: 0.617fn_mistral_s_n200_v1: 0.613fn_glm_l_n200_v1: 0.612fn_glm_s_n200_v1: 0.593fn_gliner_n200_v1: 0.518fn_capitalized_n200_v1: 0.222fn_nothing_n200_v1: 0.1250.125better · 0.774

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