ai4co
some questions
Hello,
Thank you for sharing the codebase! I’ve made a few modifications to the worker code:
1.Added TensorBoard to track the training process. 2.Updated the code so that training can resume if interrupted unexpectedly or stopped for any other reason. 3.Implemented parallel computation for each validation case, improving the overall validation efficiency.
However, I encountered a couple of issues:
My code isn’t utilizing the GPU for training and only uses the CPU. This isn't a critical issue since it doesn't affect the algorithm’s functionality, but I wanted to point it out.I don't know why it's happening.torch.cuda.is_available() showing True
The valid_set folder contains only three files (40length_16agents_0.3density.pth, 40length_32agents_0.3density.pth, and 40length_64agents_0.3density.pth) used for validation. I wanted to know if there is a way to generate other test environments defined in test_env_settings in the config file, which are not present in the folder.
Any guidance on these points would be greatly appreciated!
Thanks again for your help!
Hi, thanks for your interest!
My code isn’t utilizing the GPU for training and only uses the CPU. This isn't a critical issue since it doesn't affect the algorithm’s functionality, but I wanted to point it out.I don't know why it's happening.torch.cuda.is_available() showing True
Is this related to our codebase? If you have a GPU but you are not using it, I expect torch.cuda.is_available() will still show True
The valid_set folder contains only three files (40length_16agents_0.3density.pth, 40length_32agents_0.3density.pth, and 40length_64agents_0.3density.pth) used for validation. I wanted to know if there is a way to generate other test environments defined in test_env_settings in the config file, which are not present in the folder.
(If you refer to how you can generate more data) Yes, you can generate more! Here are the instructions: https://github.com/ai4co/eph-mapf/tree/main/src/data
I just wanted to take a moment to express my heartfelt gratitude for sharing your code and providing such clear and helpful explanations. Your insights were invaluable in resolving my questions, and I truly appreciate the time and effort you put into assisting me.