Baselines of grid dispatching competition by zenghsh3 · Pull Request #709 · PaddlePaddle/PARL

zenghsh3 · 2021-08-20T08:47:51Z

No description provided.

TomorrowIsAnOtherDay · 2021-08-25T12:26:48Z

examples/Baselines/GridDispatch_competition/README.md

@@ -0,0 +1,7 @@
+## Baselines for grid dispatching competition


We can write in Chinese. The participants only come from China.

Some of the participants are from Portugal.

examples/Baselines/GridDispatch_competition/paddle/README.md

TomorrowIsAnOtherDay · 2021-08-25T12:36:04Z

examples/Baselines/GridDispatch_competition/paddle/env_wrapper.py

+    return env
+
+
+if __name__ == '__main__':


We can remove the test code.

TomorrowIsAnOtherDay · 2021-08-25T12:36:30Z

examples/Baselines/GridDispatch_competition/paddle/env_wrapper.py

+        low_bound = gen_p_action_space.low
+        high_bound = gen_p_action_space.high
+
+        mapped_action = low_bound + (action - (-1.0)) * (


Why don't we use the action_mapping wrapper here?

The current wrapper cannot support it. (We fix it in #673)

TomorrowIsAnOtherDay · 2021-08-25T12:37:41Z

examples/Baselines/GridDispatch_competition/paddle/env_wrapper.py

+
+
+class MaxTimestepWrapper(gym.Wrapper):
+    def __init__(self, env):


can we pass the max_timestep her instead of declaring it globally?

TomorrowIsAnOtherDay · 2021-08-25T12:40:01Z

examples/Baselines/GridDispatch_competition/paddle/train.py

+            critic_lr=CRITIC_LR)
+        self.agent = GridAgent(algorithm)
+
+        self.agent.restore("./paddle_pretrain_model")


What if the model does not exist?

It will raise an exception.

examples/Baselines/GridDispatch_competition/torch/env_wrapper.py

TomorrowIsAnOtherDay · 2021-08-27T03:07:06Z

examples/Baselines/GridDispatch_competition/paddle/README.md

+
+2. Update the data path for distributed training (Using an absoluate path).
+```bash
+export PWD=`pwd`


this line does not support the Windows system.

Added Linux dependency.

TomorrowIsAnOtherDay · 2021-08-27T03:07:41Z

examples/Baselines/GridDispatch_competition/paddle/README.md

+
+xparl start --port 8010 --cpu_num 6
+```
+


Note that you only need to start the cluster once.

TomorrowIsAnOtherDay · 2021-08-27T03:07:53Z

examples/Baselines/GridDispatch_competition/paddle/README.md

+xparl start --port 8010 --cpu_num 6
+```
+
+5. Run training script


start training.

TomorrowIsAnOtherDay · 2021-08-27T03:08:38Z

examples/Baselines/GridDispatch_competition/paddle/README.md

+python train.py --actor_num 6
+```
+
+6. Visualize


visualize the training curve and other information.

TomorrowIsAnOtherDay · 2021-08-27T03:09:52Z

examples/Baselines/GridDispatch_competition/paddle/README.md

@@ -0,0 +1,60 @@
+## SAC baseline for grid dispatching competition
+
+In this example, we provide a distributed SAC baseline based on PARL and paddlepaddle to solve the [grid dispatching competition](https://aistudio.baidu.com/aistudio/competition/detail/111) task.


to solve the task -> for the task.

TomorrowIsAnOtherDay · 2021-08-27T03:10:09Z

examples/Baselines/GridDispatch_competition/paddle/README.md

+In this example, we provide a distributed SAC baseline based on PARL and paddlepaddle to solve the [grid dispatching competition](https://aistudio.baidu.com/aistudio/competition/detail/111) task.
+
+### Dependencies
+* python3


added python3.6+

TomorrowIsAnOtherDay

LGTM.

zenghsh3 added 5 commits August 20, 2021 16:42

add baselines for grid dispatching competition

c947826

update doc

6fd740d

remove model

505bf5c

update link

6056117

fix yapf

b0ee005

zenghsh3 requested a review from TomorrowIsAnOtherDay August 24, 2021 02:14

TomorrowIsAnOtherDay reviewed Aug 25, 2021

View reviewed changes

refine code and doc

f67d5d1

TomorrowIsAnOtherDay reviewed Aug 27, 2021

View reviewed changes

refine doc

6b5a690

TomorrowIsAnOtherDay approved these changes Aug 27, 2021

View reviewed changes

TomorrowIsAnOtherDay merged commit 7a910d3 into develop Aug 27, 2021

zenghsh3 deleted the grid_baselines branch January 20, 2022 07:57

		@@ -0,0 +1,7 @@
		## Baselines for grid dispatching competition



		class MaxTimestepWrapper(gym.Wrapper):
		def __init__(self, env):

		@@ -0,0 +1,60 @@
		## SAC baseline for grid dispatching competition

		In this example, we provide a distributed SAC baseline based on PARL and paddlepaddle to solve the [grid dispatching competition](https://aistudio.baidu.com/aistudio/competition/detail/111) task.

Conversation

zenghsh3 commented Aug 20, 2021

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

TomorrowIsAnOtherDay left a comment

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants