Xiong et al. (2020) — ARXIV-2002.04745
Standing in the anthology
Retired as a duplicate rather than for anything about the work. LIT-114 carries the reading.
The two notes reached the same paper from opposite directions, which is why the duplication survived a year of lints. This one was filed correctly and described wrongly: the identifier is right, and the single takeaway under it names RMSNorm, which the paper never discusses. LIT-114 was filed wrongly and corrected: it arrived as a fabricated title against an unrelated arXiv id, and the September repair re-sourced it to Xiong et al. Nothing compared the two, because no check in the record compares identifiers across notes (luria#165).
The cost was not the extra document. It was SOTA-032, which sourced this note and inherited its conflation — recommending pre-norm and RMSNorm as one practice, citing a paper that argues only the first. That practice is now split, and the half this note misattributed is sourced to LIT-023, where it belonged all along.