Maryam Honari comments

Results 8 comments of


                                            Maryam Honari

performances of rnn2.py

Hi elasryidriss, The reported accuracy seemed to be from doc-cnn4.py model. I'm still curious why the the rnn2.py doesn't improve and keeps around 49%.

Defining an empty observation for Buffer Sensor

Hi, Thanks for bringing up this request. Currently we consider a nearly zero entity as an empty observation and it's a fair point that it might be a valid observation....

Using ML agent on other engines

Hi @Ali-khn, Thank you for being an ML-Agents user. Adding support for external engines in not on the ML-Agents Roadmap right now. We are focused on adding support for custom...

Evaluation during training and it’s statistics

Hi @chenzhutian, This is a useful feature. I logged it internally under [MLA-2465](https://jira.unity3d.com/browse/MLA-2465) and we will prioritize accordingly.

Support for Torch 1.9

Supporting M1 mac is definitely on our roadmap! I've logged the upgrade to PyTorch in our internal tracker as MLA-2273 and will be prioritized accordingly.

SaveJsonTimers shouldn't save to dataPath

Thanks for bringing this to our attention. That's a fair enhancement and I logged it in our internal tracker.

multiprocessing for inference and training, to make full use of CPU and GPU.

Hi, We have this under our radar to improve the efficiency of our trainers. We also had prototypes for parallelizing the SubprocessEnvManager. Thanks for bringing this to our attention, will...

fatal error: mjmodel.h: No such file or directory

@adityagupte95 for me moving everything inside ~/.mujoco/mujoco200/mujoco200_linux to ~/.mujoco/mujoco200/ and deleting the mujoco200_linux fixed the problem