fix: duplicate pad_between_seqs skips and wrong architecture direction - #3390

fix: duplicate pad_between_seqs skips and wrong architecture direction#3390

andrewwhitecdw wants to merge 3 commits into

NVIDIA:mainNVIDIA/TransformerEngine:mainfrom

andrewwhitecdw:bugfix/test-attention-with-cp-duplicate-pad-between-seqs-skips-andandrewwhitecdw/TransformerEngine:bugfix/test-attention-with-cp-duplicate-pad-between-seqs-skips-andCopy head branch name to clipboard