on-policy

on-policy copied to clipboard

When I run train_football_3v1.sh,the same error occurs as Issue #72. The error got an unexpected keyword argument "rnn_states ". Finally, the problem was found in the shared_buffer.py file under utils....

sixie10

Does the state of the grid world need to be normalized?

1

I have made a multi-agent pathfinding in a grid world. The state adopts the Manhattan distance between the agent and the goal. Does it need to be normalized when inputting...

saiyuhang123

无法解析导入“wandb”

1

无法解析导入“wandb”,[{ "resource": "/home/hmp/jwj/project1/off-policy/offpolicy/runner/mlp/mpe_runner.py", "owner": "_generated_diagnostic_collection_name_#1", "code": { "value": "reportMissingImports", "target": { "$mid": 1, "path": "/microsoft/pyright/blob/main/docs/configuration.md", "scheme": "https", "authority": "github.com", "fragment": "reportMissingImports" } }, "severity": 4, "message": "无法解析导入“wandb”", "source": "Pylance", "startLineNumber":...

jun-running

Why share_obs in shared/mpe_runner.py just repeats obs when self.use_centralized_V set to true

1

Why is share_obs repeating the obs value again for n times when using a centralized V? Code taken from [mpe_runner.py](https://github.com/marlbenchmark/on-policy/blob/d53c4902cf2c291c93ced2c42c621371982ca2eb/onpolicy/runner/shared/mpe_runner.py#L133C1-L137C28) ``` if self.use_centralized_V: share_obs = obs.reshape(self.n_rollout_threads, -1) share_obs = np.expand_dims(share_obs,...

Jaroan

I was trying to reproduce your results, however whenever I try to run the script about `simple_speaker_listener`, it crashes for a shape mismatch (Currently trying your last commit, but also...

AlbertoSinigaglia