main
[shardformer] adapted T5 and LLaMa test to use kit (#4049)
[shardformer] support ep for deepseek v3 (#6185)
[CI/tests] simplify some test case to reduce testing time (#5755)
[Feature] Split cross-entropy computation in SP (#5959)
[upgrade]Upgrade transformers (#6320)
[release] update version (#6195)
[release] update version (#5752)
[Zerobubble] merge main. (#6142)
[hotfix] fix hybrid checkpointio for sp+dp (#6184)
[misc] refactor launch API and tensor constructor (#5666)
[feat] support qwen3 in shardformer
[misc] update torch version (#6206)