作者位次 #5/8ACL 2026Findings· Findings of ACL
L a MPE : Length-aware Multi-grained Positional Encoding for Adaptive Long-context Scaling Without Training
Sikui Zhang · Guangze Gao · Ziyun Gan · Chunfeng Yuan · Zefeng Lin · Houwen Peng · Bing Li · Weiming Hu
摘要
Large language models (LLMs) experience significant performance degradation when the input exceeds the pretraining context window, primarily due to the out-of-distribution (OOD) behavior of Rotary Position Embedding (RoPE). Recent studies mitigate this problem by remapping OOD positions into the...