[PyTorch] Inference mode disables initializing quantized weights with column-wise usage#1847

ksivaman merged 4 commits intoNVIDIA:mainNVIDIA/TransformerEngine:mainfrom

timmoon10:inference-mode-weight-inittimmoon10/TransformerEngine:inference-mode-weight-initCopy head branch name to clipboard