虼蚻_cmchao
@cmchao
Mon, Jun 17, 2024 4:55 AM
1
上星期有一篇
虼蚻_cmchao
@cmchao
Mon, Jun 17, 2024 4:56 AM
另一篇報導
Never Again MatMul: A New Era for Lightweight Langua...
虼蚻_cmchao
@cmchao
Mon, Jun 17, 2024 4:56 AM
路人一:"They're still multiplying matrices, it's just that one of the matrices has entries -1, 0, 1."
虼蚻_cmchao
@cmchao
Mon, Jun 17, 2024 4:57 AM
路人二:"This kind of weight is called tenary weight. There is a paper called “Ternary Weight Networks” which is published in 2016.
However, LLM has less error tolerance in quantization. I think this method still has long path to be widely used in LLM"
虼蚻_cmchao
@cmchao
Mon, Jun 17, 2024 4:59 AM
對噗頭的那個人的評價....
載入新的回覆
However, LLM has less error tolerance in quantization. I think this method still has long path to be widely used in LLM"