1 paper
Sijun Tan, Xiuyu Li, Shishir Patil +5
Processing long contexts remains a challenge for large language models (LLMs) due to the quadratic computational and memory overhead of the self-attention mechanism and the substan…