ncnn
ncnn copied to clipboard
lstm int8前向
参考教程https://github.com/Tencent/ncnn/wiki/quantized-int8-inference 将 lstm模型转为int8,但Debug模式下,代码运行和非量化没有区别。
lstm在windows下,没有实现量化前向?
https://github.com/Tencent/ncnn/pull/5435
implemented in https://github.com/Tencent/ncnn/commit/08b7d99a75ad70fdc5c1d9e5debc675fce5e83a9