mkabra
commented
1 year ago Hi Stefan,
Thanks for reporting this. We hadn't tested docker on windows 11 yet but we are in the process of testing it. Will update you when it is ready.
As for deeplabcut, yes APT ignores the number of iterations set for it because deeplabcut has a specific learning rate schedule that we didn't know how best to adapt for variable number of iterations.
Mayank
From: Stefan Lemke @.> Sent: Tuesday, April 4, 2023 10:33 PM To: kristinbranson/APT @.> Cc: Subscribed @.***> Subject: [kristinbranson/APT] docker setup on Windows 11 (Issue #410)
External Email: Use Caution
Hello from the Hantman lab - I have been using the docker backend on Windows 10 (multianimal branch, 2 view project) without any issues (except that it seems to ignore any input for N. iterations of training and always trains for 1003000 iterations) but I recently tested it up on a PC running Windows 11 and I am running into the following issues:
The backend configuration test doesn't pass: [image]https://urldefense.com/v3/__https://user-images.githubusercontent.com/8893242/229862178-d83fbda5-bb36-4aba-8ee5-73ab993b47fd.png__;!!Eh6p8Q!Ezlx2mCDmbQTr45VgyymHwm2Oo4EB_Sy5TfoTZHmmwaCK4qZTFsQd8fBJxOwSCXz4c9tRuzMj_IJDnhzjtlBGogp8Q8$
Testing the NVIDIA GPU support using the docker command: docker run --rm -it --gpus=all nvcr.io/nvidia/k8s/cuda-sample:nbody nbody -gpu -benchmark doesn't give an error but the shell doesn't advance or display anything: [image]https://urldefense.com/v3/__https://user-images.githubusercontent.com/8893242/229862699-bbcf42da-61c2-49e3-961f-19c796da254a.png__;!!Eh6p8Q!Ezlx2mCDmbQTr45VgyymHwm2Oo4EB_Sy5TfoTZHmmwaCK4qZTFsQd8fBJxOwSCXz4c9tRuzMj_IJDnhzjtlB4BQv2gg$
I don't get an error on trying to train a network but Matlab doesn't advance after printing "Training started at ... " and the training GUI is not generated. [image]https://urldefense.com/v3/__https://user-images.githubusercontent.com/8893242/229863819-e81f532d-eb01-4ebd-a2cc-fb4d0e1f34b0.png__;!!Eh6p8Q!Ezlx2mCDmbQTr45VgyymHwm2Oo4EB_Sy5TfoTZHmmwaCK4qZTFsQd8fBJxOwSCXz4c9tRuzMj_IJDnhzjtlBdGrG7mw$
Thanks for any help to get this running! Stefan
— Reply to this email directly, view it on GitHubhttps://urldefense.com/v3/__https://github.com/kristinbranson/APT/issues/410__;!!Eh6p8Q!Ezlx2mCDmbQTr45VgyymHwm2Oo4EB_Sy5TfoTZHmmwaCK4qZTFsQd8fBJxOwSCXz4c9tRuzMj_IJDnhzjtlBsBHih28$, or unsubscribehttps://urldefense.com/v3/__https://github.com/notifications/unsubscribe-auth/AAJNKY5INJ5K3XDF7VBVHKLW7RH4TANCNFSM6AAAAAAWS7P37Y__;!!Eh6p8Q!Ezlx2mCDmbQTr45VgyymHwm2Oo4EB_Sy5TfoTZHmmwaCK4qZTFsQd8fBJxOwSCXz4c9tRuzMj_IJDnhzjtlBn_F7b2E$. You are receiving this because you are subscribed to this thread.Message ID: @.***>
Hello from the Hantman lab - I have been using the docker backend on Windows 10 (multianimal branch, 2 view project) without any issues (except that it seems to ignore any input for N. iterations of training and always trains for 1003000 iterations) but I recently tested it up on a PC running Windows 11 and I am running into the following issues:
The backend configuration test doesn't pass:
Testing the NVIDIA GPU support using the docker command: docker run --rm -it --gpus=all nvcr.io/nvidia/k8s/cuda-sample:nbody nbody -gpu -benchmark doesn't give an error but the shell doesn't advance or display anything:
I don't get an error on trying to train a network but Matlab doesn't advance after printing "Training started at ... " and the training GUI is not generated.
Thanks for any help to get this running! Stefan