#llm memory evaluation

RENDER: Controlling Reader-Facing Evidence in LLM Memory EvaluationNEW⭐8

Evaluating LLM memory with RENDER: how reader-facing formats—summaries, typed records, raw text—impact accuracy, showing up to 72.6-point variance across models...