Search Unity

Search Results

Search Again
Page 1 of 8
  1. ervteng_unity
    Post

    Question How to improve the mean reward of my project (Observations, .yaml file, etc)

    Before changing the observation space, I'd try for running it MUCH longer than 2M steps. Maybe set max_steps to 20M or more. If it becomes...
    Post by: ervteng_unity, Jul 27, 2021 in forum: ML-Agents
  2. ervteng_unity
    Post

    GAIL vs Behavioral Cloning, what's the difference?

    Hey jokerHHH, can you post a picture of your reward curve? It is possible the agent needs to unlearn a bit before going back up, especially if BC...
    Post by: ervteng_unity, Jul 27, 2021 in forum: ML-Agents
  3. ervteng_unity
    Post

    Question Reward/Penalty for a general situation

    Luckily, this is exactly what RL is designed to do. Giving an agent a reward at the end of a sequence of actions is rewarding that trajectory, not...
    Post by: ervteng_unity, Jul 27, 2021 in forum: ML-Agents
  4. ervteng_unity
    Post

    Question SAC training locks-up Unity PPO is fine

    Does it ever unfreeze when using SAC? Or does it just stay stuck forever? Also, try setting "threaded" to false and seeing if the same still happens.
    Post by: ervteng_unity, Jul 27, 2021 in forum: ML-Agents
  5. ervteng_unity
    Post

    Can I open in PyTorch a model trained on mlagents?

    Nope, this should work fine!
    Post by: ervteng_unity, Jul 23, 2021 in forum: ML-Agents
  6. ervteng_unity
    Post

    How can i use Ml-Agents training on my CPU?

    You should be able to train on CPU. I think the Unity environment isn't starting. Does a little window pop up with some agents in it? Actually...
    Post by: ervteng_unity, Jul 23, 2021 in forum: ML-Agents
  7. ervteng_unity
    Post

    Hummingbird Tutorial - MLAgents Python to Unity Connection

    Are you getting any errors on the Unity side?
    Post by: ervteng_unity, Jul 22, 2021 in forum: ML-Agents
  8. ervteng_unity
    Post

    Google Brax Physics Engine

    Totally agree, faster physics would be great! New physics engines though are quite out of the scope of the ML-Agents product. Unity is working...
    Post by: ervteng_unity, Jul 22, 2021 in forum: ML-Agents
  9. ervteng_unity
    Post

    ONNX Model inputs

    I believe they're two sets of observations, one from the joints and body,. and another from those added into the code. You should see an...
    Post by: ervteng_unity, Jul 22, 2021 in forum: ML-Agents
  10. ervteng_unity
    Post

    Can I open in PyTorch a model trained on mlagents?

    This isn't a supported use-case currently, as the architecture is determined by the observations/actions you specify in Unity. Netron...
    Post by: ervteng_unity, Jul 22, 2021 in forum: ML-Agents
  11. ervteng_unity
    Post

    Google Brax Physics Engine

    This isn't in our pipeline. Ml-Agents will support any physics engines that the Unity engine itself supports.
    Post by: ervteng_unity, Jul 21, 2021 in forum: ML-Agents
  12. ervteng_unity
    Post

    Using SAC and Behavioral Cloning together

    Yep, SAC and BC work fine together - just enable it in the same way you would with PPO.
    Post by: ervteng_unity, Jul 21, 2021 in forum: ML-Agents
  13. ervteng_unity
    Post

    Trainning is more and more slow

    You may have a memory leak - can you check the memory usage of both Python and Unity over time?
    Post by: ervteng_unity, Jul 21, 2021 in forum: ML-Agents
  14. ervteng_unity
    Post

    Validity Testing for Reinforcement Learning Projects

    You're right, with RL, the simulator/game IS the dataset. However, you can test the reliability of your results by running the same experiments...
    Post by: ervteng_unity, Jun 25, 2021 in forum: ML-Agents
  15. ervteng_unity
    Post

    How does ML decide value for continuous action space?

    BTW, this Twitter post may be of interest: https://twitter.com/kressdev/status/1399073501810675721 Larger networks also encourage exploration, so...
    Post by: ervteng_unity, Jun 25, 2021 in forum: ML-Agents
  16. ervteng_unity
    Post

    NullReferenceException

    I don't think this is an ML-Agents issue - maybe try re-importing all the assets?
    Post by: ervteng_unity, Jun 25, 2021 in forum: Scripting
  17. ervteng_unity
    Post

    Question Academy Fixed Update Stepper Generating Garbage

    I don't know about XL2 chips - that would depend on Unity compute shaders taking advantage of the hardware. As for the errors - are you still...
    Post by: ervteng_unity, Jun 25, 2021 in forum: ML-Agents
  18. ervteng_unity
    Post

    Delaying episode

    Where are you running your initialization code? If it is in OnEpisodeBegin, it should be called before the first CollectObservations of that...
    Post by: ervteng_unity, Jun 23, 2021 in forum: ML-Agents
  19. ervteng_unity
    Post

    Question Academy Fixed Update Stepper Generating Garbage

    By any chance are you seeing anything in your console? It's possible that Barracuda is writing a warning or info to console and that's eating up a...
    Post by: ervteng_unity, Jun 23, 2021 in forum: ML-Agents
  20. ervteng_unity
    Post

    Resolved Are Agents Set To Heuristic Ignored?

    Heuristic agents don't affect training in any way, from the perspective of the trainer they're the same as a scripted NPC not built with ML-Agents.
    Post by: ervteng_unity, Jun 23, 2021 in forum: ML-Agents
Showing results 1 to 20 of 150 Next >
Search Again
Page 1 of 8