Modularize fused experts and integrate PPLX kernels (#15956)
This commit is contained in:
@@ -390,7 +390,8 @@ class CPUWorker(LocalOrDistributedWorkerBase):
|
||||
|
||||
ensure_model_parallel_initialized(
|
||||
parallel_config.tensor_parallel_size,
|
||||
parallel_config.pipeline_parallel_size)
|
||||
parallel_config.pipeline_parallel_size,
|
||||
parallel_config.enable_expert_parallel)
|
||||
|
||||
def get_cache_block_size_bytes(self) -> int:
|
||||
"""Return the size in bytes of a single KV cache block.
|
||||
|
||||
Reference in New Issue
Block a user