facebookresearch / Detectron

FAIR's research platform for object detection research, implementing popular algorithms like Mask R-CNN and RetinaNet.
Apache License 2.0
26.27k stars 5.46k forks source link

terminate called after throwing an instance #136

Closed zhcf closed 6 years ago

zhcf commented 6 years ago

Actual results

terminate called after throwing an instance of 'caffe2::EnforceNotMet' what(): [enforce fail at context_gpu.h:170] . Encountered CUDA error: invalid device function Error from operator: input: "gpu_0/res2_0_branch2c_bn" input: "gpu_0/res2_0_branch1_bn" output: "gpu_0/res2_0_branch2c_bn" name: "" type: "Sum" device_option { device_type: 1 cuda_gpu_id: 0 } Aborted at 1518191836 (unix time) try "date -d @1518191836" if you are using GNU date PC: @ 0x7f18119a01f7 GI_raise SIGABRT (@0x2c82) received by PID 11394 (TID 0x7f16caffd700) from PID 11394; stack trace: @ 0x7f18124465e0 (unknown) @ 0x7f18119a01f7 __GI_raise @ 0x7f18119a18e8 GI_abort @ 0x7f180af79ac5 (unknown) @ 0x7f180af77a36 (unknown) @ 0x7f180af77a63 (unknown) @ 0x7f180afce345 (unknown) @ 0x7f181243ee25 start_thread @ 0x7f1811a6334d __clone

Detailed steps to reproduce

cd /opt/project/detectron python2 tools/train_net.py \ --cfg configs/getting_started/tutorial_1gpu_e2e_faster_rcnn_R-50-FPN.yaml \ OUTPUT_DIR work/output/coco_train_with_1gpu

System information

zhcf commented 6 years ago

I see there are 2 or 3 posts are about this kind problem in recent days

zhcf commented 6 years ago

clean caffe2 and rebuild, then it worked.

ir413 commented 6 years ago

Closing since the issue seems resolved. Please reopen if that is not the case.