Add NVFP4 variant quantized with NVIDIA ModelOpt for vLLM bb42b17 verified switzerchees commited on Feb 14