From 8a1f9e4e1a81ffa02c79ccd9fd3e0cd1f4e96def Mon Sep 17 00:00:00 2001
From: Inhyuk Andy Cho <andy.inhyuk.jo@intel.com>
Date: Thu, 5 Jan 2023 17:29:39 +0900
Subject: [PATCH] Revert "[OTX] Bugfix: multi GPU raise error when num_workers
 isn't set as 0. (#1475)"

This reverts commit c076902a791e069cb89abab89550cc13449d6b53.
---
 otx/cli/utils/multi_gpu.py | 1 -
 1 file changed, 1 deletion(-)

diff --git a/otx/cli/utils/multi_gpu.py b/otx/cli/utils/multi_gpu.py
index 651a7be09cd..50ccb9bc8cd 100644
--- a/otx/cli/utils/multi_gpu.py
+++ b/otx/cli/utils/multi_gpu.py
@@ -169,7 +169,6 @@ def run_child_process(train_func: Callable, rank: int, gpu_ids: List[int], outpu
             output_path (str): output path where task output are saved.
             multi_gpu_port (str): port for communication between multi GPU processes.
         """
-        mp.set_start_method(method="fork", force=True)  # set multi porcess method as fork
         gpus_arg_idx = sys.argv.index("--gpus")
         for _ in range(2):
             sys.argv.pop(gpus_arg_idx)