DeepSpeed

ffe0af23 - Fix the bug of deepspeed sequence parallel working with batch size larger than 1 (#5823)

Fix the bug of deepspeed sequence parallel working with batch size larger than 1 (#5823) Modified the `alltoall` function Verified the results with only `TP`:  --------- Co-authored-by: Jinghan Yao <yjhmitweb@ascend-rw02.ten.osc.edu> Co-authored-by: Sam Ade Jacobs <samjacobs@microsoft.com> Co-authored-by: Jinghan Yao <yjhmitweb@ascend-rw01.ten.osc.edu> Co-authored-by: Logan Adams <loadams@microsoft.com>