nikhilbarhate99
Training is ok, but failed to eval.
Hello👋,
Thank you for open-sourcing the code for the min decision transformer. Your code has been tremendously helpful in helping me understand DT.
However, I am currently facing an issue. During the training process, the action loss is indeed steadily decreasing, but the test results have consistently been subpar, to the point of having no discernible impact. I've been grappling with this problem for a while now and can't seem to figure out why this is happening.
By the way, I haven't tested it on the three environments, namely halfcheetah, hopper, and walker2d, mainly because I've been struggling with configuring d4rl. I'm using the upgraded version of d4rl provided by Farama, specifically on the pointmaze offline dataset.
If you could spare some time to assist me with this, I would be immensely grateful!
The Decision Transformer paper does not provide results for pointmaze environment, it is a difficult env and I would not expect DT to work well on it out of the box.
If you are struggling to install d4rl, you can refer to the first couple of blocks on the colab notebook in the repo that installs the required libraries on google colab.
Anyway, Thank for your time! 发件人: ***@***.*** ***@***.***> 代表 Nikhil Barhate ***@***.***>日期: 星期二, 2023年10月17日 05:02收件人: nikhilbarhate99/min-decision-transformer ***@***.***>抄送: Godw ***@***.***>, Author ***@***.***>主题: Re: [nikhilbarhate99/min-decision-transformer] Training is ok, but failed to eval. (Issue #5)The Decision Transformer paper does not provide results for pointmaze environment, it is a difficult env and I would not expect DT to work well on it out of the box.—Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you authored the thread.Message ID: ***@***.***>
May DT failed to do path planning. I have tried it on the Minari dataset. It seems like it can only work in the Umaze environment.
Hello👋,
Thank you for open-sourcing the code for the min decision transformer. Your code has been tremendously helpful in helping me understand DT.
However, I am currently facing an issue. During the training process, the action loss is indeed steadily decreasing, but the test results have consistently been subpar, to the point of having no discernible impact. I've been grappling with this problem for a while now and can't seem to figure out why this is happening.
By the way, I haven't tested it on the three environments, namely halfcheetah, hopper, and walker2d, mainly because I've been struggling with configuring d4rl. I'm using the upgraded version of d4rl provided by Farama, specifically on the pointmaze offline dataset.
If you could spare some time to assist me with this, I would be immensely grateful!
@353055619 Hi, I found the value of state_mean and state_std differed in train.py and test.py recently. Maybe It's this bug that caused this issue.
I will completely consider this later! And offer you a full insight into that.