diff --git a/vllm/nvfp4_cutedsl.py b/vllm/nvfp4_cutedsl.py index e6b4567b..7a67a111 100644 --- a/vllm/nvfp4_cutedsl.py +++ b/vllm/nvfp4_cutedsl.py @@ -18,6 +18,7 @@ import torch from cutedsl.bridge import ( quantize_activation_nvfp4, quantize_weight_to_nvfp4, + quantize_to_nvfp4, make_b_k_major, assemble_scales_3d_side, run_nvfp4_grouped_gemm,