search

Xharlie / BtcDet

Behind the Curtain: Learning Occluded Shapes for 3D Object Detection
Apache License 2.0
194 stars 43 forks source link

CUDA error: invalid configuration argument #3

Closed Yzichen closed 2 years ago

Yzichen commented 2 years ago

When I run multifindbestfit.py, It reports an error: CUDA error: invalid configuration argument. How can I fix it ？

Yzichen commented 2 years ago

I increased the number of loops in the get_iou function, The problem is solved .

laiyingxin2 commented 2 years ago

same question,I tried to change the for i in range(11): increase or decrease to 200, 12,1000 ,1......There are still mistakes,请问你是怎么更改的

laiyingxin2 commented 2 years ago

def get_iou(box_tensor): limit = len(box_dims_lst) start = 0 iou3d_lst = [] for i in range(64): end = min(start + limit // 10, limit) iou3d = iou3d_nms_utils.boxes_iou3d_gpu(box_tensor[start:end, :], box_tensor) iou3d_lst.append(iou3d) start = end iou3d = torch.cat(iou3d_lst, dim=0) print("iou3d", iou3d.shape) return iou3d

in this function

DheerajYarlagadda commented 2 years ago

when I'm trying to increase the loops in the 'get_iou' function. it is giving me an error 'CUDA error: invalid configuration argument'. But when I set back the loop length to normal/original, I'm not getting this error. Could you please let me know how you overcome this error?