search

facebookresearch / ContrastiveSceneContexts

Code for CVPR 2021 oral paper "Exploring Data-Efficient 3D Scene Understanding with Contrastive Scene Contexts"
MIT License
218 stars 29 forks source link

RuntimeError: Could not detect "srun", are you indeed on a slurm cluster? #30

Open Heroism502 opened 2 years ago

Heroism502 commented 2 years ago

hello, i have a environment problem, how can i solve? thanks image

Sekunde commented 2 years ago

Please remove -m in the script if you are not in a slurm system.

Heroism502 commented 2 years ago

thanks for you reply, i got another problem, have you ever meeting ? image

Sekunde commented 2 years ago

hi, this is probably due to the different versions of omegaconf, you can try to remove .pretty() in this line: https://github.com/facebookresearch/ContrastiveSceneContexts/blob/main/downstream/votenet/lib/ddp_trainer.py#L44

Heroism502 commented 2 years ago

Thanks a lot, i have another question, did you have try to use MinkowskiEngine == v0.5 ( or more hight), i want to use your code with 0.5.4 but got lot of problem, do you have any plan to update it with the most new of MinkowskiEngine, Thanks for your answer.

Sekunde commented 2 years ago

Hi, currently we don't have the plan to migrate the code and pre-trained weights to the newest MinkowskiEngine. Notice, we do observe a performance drop if loading our pre-trained weights in ME0.5, due to the different way of kernel arrangements.

Gofinge commented 1 year ago

@Heroism502 , you can refer LanguageGroundedSemseg (ScanNet200) [here]. More specifically, only modify the voxelizer and replace the MinkUNet model with v0.5 version.