v0.4.2
[gemini] fixes for benchmarking (#5847)
[npu] change device to accelerator api (#5239)
[shardformer]: support gpt-j, falcon, Mistral and add interleaved pipeline for bert (#5088)
[gemini] quick fix on possible async operation (#5803)
[chore] remove unnecessary assert since compute list might not be recorded
[gemini] async grad chunk reduce (all-reduce&reduce-scatter) (#5713)
[bug] continue fix