vllm/cacheflow/model_executor/parallel_utils/__init__.py

13 lines
320 B
Python
Raw Normal View History

2023-05-10 06:30:12 +08:00
import cacheflow.model_executor.parallel_utils.parallel_state
import cacheflow.model_executor.parallel_utils.tensor_parallel
import cacheflow.model_executor.parallel_utils.utils
# Alias parallel_state as mpu, its legacy name
mpu = parallel_state
__all__ = [
"parallel_state",
"tensor_parallel",
"utils",
]