jyjang13/

demo-my-eval-075520

Fixed evaluation-set composition built with BenchPress. This repo stores only the recipe (manifest.json) — source data is streamed at load time. Abilities: external_knowledge_retrieval, contextual_retrieval, temporal_reasoning, commonsense_causal_reasoning

	Sources




	

benchmark source repo config split n_samples license gated

test 100 mit yes

train 100 odc-by no