how to train?

[linjuechu]

Hello, I just started doing this.I can't find where I can change my training iteration,when I run there will be only Training iteration: 0 Training iteration: 1.

[Kamalnl92]

Hello,

Your question is not clear. What do you mean by "I cannot find where I can change my training iteration?" I see that process is killed and that is probably causing you have run out of memory. Try to use GPU if not possible use the --cpu flag

Cheers

[linjuechu]

Thank you! I'm running on VMware, It seems that CUDA and CUDNN cannot be installed on VMware, no solution has been found. And only train twice every time (like the picture),and I'm not sure if this is the case, snapshot_file issue with no file for the snapshot-backup.reinorcement.pth.(snapshot-backup.reinforcement.pth is being created or found). Looking forward to your answers, thanks again!

On 5/13/2022 @.***> wrote：

Hello,

Your question is not clear. What do you mean by "I cannot find where I can change my training iteration?" I see that process is killed and that is probably causing you have run out of memory. Try to use GPU if not possible use the --cpu flag

Cheers

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>

[Kamalnl92]

Hello,

I would really not recommend using VMware, I think you better have a separate partition with Ubuntu18.04 image (Ubuntu20.04 also works for this project), Else the training would take forever.

Being killed is not related to the snapshot-backup. The snapshot-backup is just saving the model in the training thread, see below https://github.com/xukechun/Efficient_goal-oriented_push-grasping_synergy/blob/feb5dd8196e60f2e454ecf2a26c438043eb65ab1/main.py#L714

Cheers

[linjuechu]

Thank you very much!I will consider your suggestion！