Today we’re announcing Ternary Bonsai: Top intelligence at 1.58 bits
Using ternary weights {-1, 0, +1}, we built a family of models that are 9x smaller than their 16-bit counterparts while outperforming most models in their respective parameter classes on standard benchmarks.… https://twitter.com/PrismML/status/2044833023682896134/photo/1
中文: 今天我们将公布《Ternary Bonsai:顶级智能》,售价1.58位
使用三元权重{-1、0、+1},我们构建了一个模型系列,其体积比16位对应的小9倍,但在标准基准测试中优于大多数模型的参数类别。