Searched refs:requantize (Results 1 – 6 of 6) sorted by relevance
568 requantize = Requantize(cc_emitter, asm_emitter)575 requantize.SpecializeTransform1DKernel('int32_t', 'uint8_t', shape[0],
22 idea was that to automatically requantize 8bit operands to less-than-8bit during99 The general idea is that at the packing stage, we requantize input (Lhs/Rhs)197 stage to requantize input matrix data. See this function in internal/pack.h.
186 the packing stage we have to requantize inputs accordingly. See
319 csinfo_.requantize = "Requantize"; in MklLayoutRewritePass()505 rinfo_.push_back({csinfo_.requantize, in MklLayoutRewritePass()506 mkl_op_registry::GetMklOpName(csinfo_.requantize), in MklLayoutRewritePass()743 string requantize; member
206 tensorflow/core/kernels/requantize.cc
5870 "requantize.cc",6047 "requantize.cc",