polish(pu): polish atari ddp configs

puyuan1996 · Dec 19, 2024 · 9c880ad · 9c880ad
1 parent be54aac
commit 9c880ad
Show file tree

Hide file tree

Showing 4 changed files with 10 additions and 10 deletions.
diff --git a/dizoo/atari/config/serial/pong/pong_dqn_ddp_config.py b/dizoo/atari/config/serial/pong/pong_dqn_ddp_config.py
@@ -59,7 +59,7 @@
     Overview:
         This script should be executed with <nproc_per_node> GPUs.
         Run the following command to launch the script:
-        python -m torch.distributed.launch --nproc_per_node=2 --master_port=29501 ./dizoo/atari/config/serial/pong/pong_dqn_ddp_config.py
+        python -m torch.distributed.launch --nproc_per_node=2 ./dizoo/atari/config/serial/pong/pong_dqn_ddp_config.py
     """
     from ding.utils import DDPContext
     from ding.entry import serial_pipeline

diff --git a/dizoo/atari/config/serial/pong/pong_ppo_ddp_config.py b/dizoo/atari/config/serial/pong/pong_ppo_ddp_config.py
@@ -1,7 +1,7 @@
 from easydict import EasyDict
 
-pong_onppo_config = dict(
-    exp_name='data_pong/pong_onppo_ddp_seed0',
+pong_ppo_config = dict(
+    exp_name='data_pong/pong_ppo_ddp_seed0',
     env=dict(
         collector_env_num=8,
         evaluator_env_num=8,
@@ -34,7 +34,7 @@
             clip_ratio=0.2,
             adv_norm=True,
             value_norm=True,
-            # for onppo, when we recompute adv, we need the key done in data to split traj, so we must
+            # for ppo, when we recompute adv, we need the key done in data to split traj, so we must
             # use ignore_done=False here,
             # but when we add key traj_flag in data as the backup for key done, we could choose to use ignore_done=True
             # for halfcheetah, the length=1000
@@ -51,24 +51,24 @@
         eval=dict(evaluator=dict(eval_freq=1000, )),
     ),
 )
-main_config = EasyDict(pong_onppo_config)
+main_config = EasyDict(pong_ppo_config)
 
-pong_onppo_create_config = dict(
+pong_ppo_create_config = dict(
     env=dict(
         type='atari',
         import_names=['dizoo.atari.envs.atari_env'],
     ),
     env_manager=dict(type='subprocess'),
     policy=dict(type='ppo'),
 )
-create_config = EasyDict(pong_onppo_create_config)
+create_config = EasyDict(pong_ppo_create_config)
 
 if __name__ == "__main__":
     """
     Overview:
         This script should be executed with <nproc_per_node> GPUs.
         Run the following command to launch the script:
-        python -m torch.distributed.launch --nproc_per_node=2 --master_port=29501 ./dizoo/atari/config/serial/pong/pong_ppo_ddp_config.py
+        python -m torch.distributed.launch --nproc_per_node=2 ./dizoo/atari/config/serial/pong/pong_ppo_ddp_config.py
     """
     from ding.utils import DDPContext
     from ding.entry import serial_pipeline_onpolicy

diff --git a/dizoo/atari/example/atari_dqn_ddp.py b/dizoo/atari/example/atari_dqn_ddp.py
@@ -60,6 +60,6 @@ def main():
     Overview:
         This script should be executed with <nproc_per_node> GPUs.
         Run the following command to launch the script:
-        python -m torch.distributed.launch --nproc_per_node=2 --master_port=29501 ./dizoo/atari/example/atari_dqn_ddp.py
+        python -m torch.distributed.launch --nproc_per_node=2 ./dizoo/atari/example/atari_dqn_ddp.py
     """
     main()
diff --git a/dizoo/atari/example/atari_ppo_ddp.py b/dizoo/atari/example/atari_ppo_ddp.py
@@ -58,6 +58,6 @@ def main():
     Overview:
         This script should be executed with <nproc_per_node> GPUs.
         Run the following command to launch the script:
-        python -m torch.distributed.launch --nproc_per_node=2 --master_port=29501 ./dizoo/atari/example/atari_ppo_ddp.py
+        python -m torch.distributed.launch --nproc_per_node=2 ./dizoo/atari/example/atari_ppo_ddp.py
     """
     main()