v0.3.4
[kernel] support pure fp16 for cpu adam and update gemini optim tests (#4921)
[misc] update pre-commit and run all files (#4752)
[Kernels]Updated Triton kernels into 2.1.0 and adding flash-decoding for llama token attention (#4965)
[hotfix] Fix import error: colossal.kernel without triton installed (#4722)