search

Megvii-BaseDetection / YOLOX

YOLOX is a high-performance anchor-free YOLO, exceeding yolov3~v5 with MegEngine, ONNX, TensorRT, ncnn, and OpenVINO supported. Documentation: https://yolox.readthedocs.io/
Apache License 2.0
9.48k stars 2.21k forks source link

4G显存跑1batch都out of memry #1039

Closed sq2012 closed 2 years ago

sq2012 commented 2 years ago

4G显存跑1batch都out of memry，显卡是GTX1650，换了3090，发现显存占用12G，想问下，yolox这么吃显存吗？以往yolo也就4G显存随便都能跑，是我操作问题还是yolox真的就吃显存，-o --fp16已经关掉了,用的是yolox_x

FateScript commented 2 years ago

What if you use some some small model like nano, tiny or small ? YOLOX-X is a large model and you need more GPU mem to train such a model.

sq2012 commented 2 years ago

谢谢