15 Pull requests merged by 3 people

• update health check

  #690 merged Dec 27, 2024

• update configs.

  #689 merged Dec 26, 2024

• remove vllm fuse_moe kernel and add moe_sum_reduce, moe_silu_and_mul kernel.

  #688 merged Dec 26, 2024

• add H800 and A800 mla decode configs.

  #687 merged Dec 26, 2024

• add A800 grouped moe kernel json configs.

  #686 merged Dec 25, 2024

• fix config search type error.

  #685 merged Dec 25, 2024

• fix kernel config json load.

  #684 merged Dec 25, 2024

• fix tunning code.

  #683 merged Dec 25, 2024

• add H800 grouped moe kernel configs.

  #682 merged Dec 25, 2024

• add kernel config tuning way to get better performance.

  #681 merged Dec 25, 2024

• Vit triton (fp + quant / tp + dp), custom image pre_process

  #663 merged Dec 25, 2024

• misc: update python requirements

  #680 merged Dec 25, 2024

• add generation cfg parse

  #675 merged Dec 24, 2024

• fix bug for requrements.txt

  #677 merged Dec 24, 2024

• support for dp + tp end2end

  #670 merged Dec 21, 2024

1 Issue closed by 1 person

• [BUG] encountering conflicting dependencies when builting from source

  #676 closed Dec 24, 2024

1 Issue opened by 1 person

• [BUG] cannot launch server after builting from source

  #679 opened Dec 24, 2024