Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[MISC] Update Drone Entity and Training Performance Enhancements #598

Merged
merged 3 commits into from
Jan 18, 2025

Conversation

KafuuChikai
Copy link
Contributor

Changes:

  1. Update drone_entity:

    • Converted data types from NumPy arrays to PyTorch tensors to improve performance.
    • The hover_train.py script no longer requires using tensor.cpu() to transfer data from CUDA to CPU.
    • In tests with parallel environments of 8192, 16384, 32768, and 65536, we observed an average training performance improvement of 20%.
  2. Adjust Hyperparameters:

    • Updated default hyperparameters to ensure more stable and higher reward training without affecting speed.
    • Note: Further improvements in reward may require additional training time.
  3. Update README:

    • Added instructions for training with visualization.

Please review the changes and provide feedback. Thank you!

convert numpy to tensor for better training performance, there is no need to use tensor.cpu()
change params for stable and better training, update readme and add Train with visulization
@KafuuChikai
Copy link
Contributor Author

I have added a link to my repository in the README to provide the latest updates, detailed documentation, and additional resources. If you agree, we can keep it; otherwise, I can remove it. Please let me know your preference.

For the latest updates, detailed documentation, and additional resources, visit this repository: GenesisDroneEnv.

@Genesis-Embodied-AI Genesis-Embodied-AI merged commit 512ad1a into Genesis-Embodied-AI:main Jan 18, 2025
1 check passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants