1 paper · 1 filter
Mo Li, L. H. Xu, Qitai Tan +3
Large Language Models (LLMs) suffer from significant performance degradation when processing long contexts due to proactive interference, where irrelevant information in earlier pa…