WavePrune2 min read
WavePrune: One period is often enough for RoPE
Rotary Position Embedding (RoPE) encodes token positions by rotating each two-dimensional channel of the query and key vectors at a channel-specific frequency, making the attention logits invariant to a common shift of positions.
By Guancheng Du, Luotian Huang, Shaowen Wang, Si Li, Kaifeng LyuarXiv - cs.CL (NLP)
Read the full story
Originally published by arXiv - cs.CL (NLP) on Oct 7, 2026.
Comments
Sign in to join the discussion.