vllm.models.minimax_m3.nvidia.specialized_decode.runtime
¶
Modules:
-
decode–The specialized decode executor.
-
model_state–The model state of the specialized decode.
-
warmup–Tuning and compiling the specialized decode's stages.