TP/quantization/weight loading refactor part 1 - Simplify parallel linear logic (#1181)

vllm/model_executor/utils.py

@@ -4,9 +4,6 @@ import random
 import numpy as np
 import torch
 
-from vllm.model_executor.parallel_utils.parallel_state import model_parallel_is_initialized
-from vllm.model_executor.parallel_utils.tensor_parallel import model_parallel_cuda_manual_seed
-
 
 def set_random_seed(seed: int) -> None:
     random.seed(seed)
@@ -14,6 +11,3 @@ def set_random_seed(seed: int) -> None:
     torch.manual_seed(seed)
     if torch.cuda.is_available():
         torch.cuda.manual_seed_all(seed)
-
-    if model_parallel_is_initialized():
-        model_parallel_cuda_manual_seed(seed)