I feel that there is an issue with the running efficiency of this code.

qzp2018 / MCLN

This is a PyTorch implementation of MCLN proposed by our paper "Multi-branch Collaborative Learning Network for 3D Visual Grounding"(ECCV2024)

Other

11 stars 0 forks source link

I feel that there is an issue with the running efficiency of this code. #4

Closed RobinLiuZX closed 3 months ago

RobinLiuZX commented 4 months ago

I am running the code in a dual NVIDIA RTX 3090 environment, but the time to train one epoch has increased from around 2 hours to now over 10 hours, and it still hasn't finished training one epoch. Could you please tell me why this is happening?

qzp2018 commented 4 months ago

I tried on the official code and ran successfully. It takes 26 minutes to train one epoch with 4 3090 GPUs, and 48 minutes with 2 3090 GPUs. You may check whether your data is correctly organized.

RobinLiuZX commented 4 months ago

Ok, thanks!