main
[MoE/ZeRO] Moe refactor with zero refactor (#5821)
fix dist log prob test
[feat] support qwen3 in shardformer
[shardformer] adapted T5 and LLaMa test to use kit (#4049)
[Feature] Zigzag Ring attention (#5905)
[misc] update pre-commit and run all files (#4752)
[misc] refactor launch API and tensor constructor (#5666)