NM512 / dreamerv3-torch

Implementation of Dreamer v3 in pytorch.
MIT License
422 stars 96 forks source link

load a trained model and evaluate it visually #61

Open luokuang2001 opened 3 months ago

luokuang2001 commented 3 months ago

Hello! I have trained the Dreamer on my self-defined env and saved the log, How can I load a trained model “latest.pt" in my logdir, and evaluate the model policy visually on my self-defined env ?

NM512 commented 1 month ago

Hello!

If you specify the same log directory, you can restart both the training and evaluation processes seamlessly. Additionally, to visually evaluate the model policy, you can check the images saved in TensorBoard by turning on video_pred_log in the config file.