tinyzqh / light_mappo Goto Github PK
View Code? Open in Web Editor NEWLightweight version of MAPPO to help you quickly migrate to your local environment.
Lightweight version of MAPPO to help you quickly migrate to your local environment.
师兄您好,我在调整env_continuous.py文件中第30行代码u_action_space = spaces.Box(low=0.0, high=90.0, shape=(self.signal_action_dim,), dtype=np.float32)后,action取值并没有限制在0和90之间,请问师兄这是为什么呢?谢谢师兄了。
from envs.env_core import EnvCore
I am using my customized env and want to render in 'human' and 'rgb_array' mode. Could you please give some examples or implementation?
直接把env_discrete文件里的 self.discrete_action_input = False是不是就是用于连续动作
在env_wrappers.py中,step_wait()的"obs[i] = self.envs[i].reset()"判断episode是否结束,这里将reset之后的观测值传给了obs[i],导致episode结束的那一刻的obs被覆盖。这样赋值是否不妥?因为reset之后的obs可以认为是随机的,不应该将其赋给obs[i],而应该直接调用"self.envs[i].reset()"?
給出的范例只有 sub-agnet_obs ,这里是不是没有特别区分观测信息与全局状态信息? 这里的 sub_agent_obs 就是指智能体的部分观测信息的列表吗?那这样全局信息是怎么处理的呢,就是把部分观测信息的融合作为全局信息?
是不是连续动作空间的环境不能用eval
Traceback (most recent call last):
File "G:\lcz\mappo\train\train.py", line 149, in
main(sys.argv[1:])
File "G:\lcz\mappo\train\train.py", line 137, in main
runner.run()
File "G:\lcz\mappo\runner\shared\env_runner.py", line 88, in run
self.eval(total_num_steps)
File "C:\Users\ljh99\anaconda3\lib\site-packages\torch\autograd\grad_mode.py", line 27, in decorate_context
return func(*args, **kwargs)
File "G:\lcz\mappo\runner\shared\env_runner.py", line 183, in eval
raise NotImplementedError
NotImplementedError
RT,runner在训练的时候,如果环境给出了done,runner没有进行reset,在某些环境中可能会导致不收敛
您好!如果agent的动作维度不一致时,light-mappo如何进行action mask?
每个智能体的观测空间是一个3宽高的Box,那么在env_core.py中self.obs_dim该如何设置呢?以及env_discrete.py里的observation_space又该如何设置?最后是否还有其他要做特别修改的地方吗?
训练出来的模型真的有效果吗?
env_runner warmup()函数位置在episode外
报错定位于runner/separated/env_runner.py中的collcet函数中的
actions = np.array(actions).transpose(1, 0, 2)
初步排查发现在这句代码上面的循环中,当agent_id取1时,生成的动作的shape与agent_id取0时不同
There is the mistake in the env_wrappers.py that is the VecEnvWrappercan is the unresolved reference.
Thanks very much for your codes.
Have you considered to extend it into other variants of MAPPO, such as MAPPO-L?
I want to use continuous actions, but an error is reported after setting the self.discrete_action_space
in the environment to false.
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.