Troubleshooting & FAQ

Common Issues

1. Installation Errors

• Solution: Ensure Python 3.8+, pip, and all dependencies are installed. Use a virtual environment.

2. CUDA/Device Errors

• Solution: Check CUDA version and PyTorch installation. Run on CPU with --device cpu if needed.

3. Poor Training Performance

• Solution: Tune hyperparameters, check environment compatibility, increase training steps.

4. Checkpoint Not Found

• Solution: Ensure the path to .pth files is correct and the file exists.

5. Unsupported Environment

• Solution: Add a wrapper or check the environment name.

Frequently Asked Questions

Q: Can I use my own environment?
A: Yes, add a wrapper in environments/ and update the config.

Q: How do I visualize training?
A: Use TensorBoard or the built-in plotting utilities.

Q: How do I contribute?
A: See Contributing.

If your issue is not listed, please open an issue on GitHub.