memit-iclr-2023-gptj-gptneox

IN premise — summaries/2026/08/24/meng-2022-memit-s0-abstract.md

Created 2026-08-25T02:58:13+00:00

MEMIT was published at ICLR 2023 and demonstrated on GPT-J (6B) and GPT-NeoX (20B)

Summary

MEMIT is a peer-reviewed method for editing the knowledge stored inside a language model, validated at a top-tier machine learning conference in 2023. It was tested on two mid-sized open-source models (6B and 20B parameters), showing the technique works beyond tiny toy models but within a practical range of model sizes.