space

Author: wenscarl

csrc/nv_internal/tensorrt_llm/thop/fp4Quantize.h

@@ -33,6 +33,7 @@ void fp4_quantize(Tensor self, Optional<Tensor> const& globalScale, Tensor value
 
 void fp4_batched_quantize(Tensor self, Optional<Tensor> const& mask, Tensor globalScale,
                           Tensor valueE2M1, Tensor scaleFP8SF, int64_t sfVecSize, bool sfUseUE8M0);
+
 void silu_and_mul_fp4_batched_quantize(Tensor const& self, Tensor const& mask,
                                        Tensor const& globalScale, Tensor valueE2M1,
                                        Tensor scaleFP8SF, int64_t sfVecSize);