1 paper
SooHwan Eom, Jay Shim, Gwanhyeong Koo +4
The Transformer's quadratic complexity with input length imposes an unsustainable computational load on large language models (LLMs). In contrast, the Selective Scan Structured Sta…