search

tianzhi0549 / FCOS

FCOS: Fully Convolutional One-Stage Object Detection (ICCV'19)
https://arxiv.org/abs/1904.01355
Other
3.28k stars 630 forks source link

Unexpected Break Down #199

Open wanghao9610 opened 5 years ago

wanghao9610 commented 5 years ago

I missed an unexpected issue that when the training process "start training", I have been used maskrcnn-benchmark for a long time, but I haven't met the issue. Following is the error message, hope someone can give me a solution, thanks in advance! Error Message: UserWarning: torch.distributed.reduce_op is deprecated, please use torch.distributed.ReduceOp instead warnings.warn("torch.distributed.reduce_op is deprecated, please use " (I think it is not an error that can break the training down, but then the training process was down and back to the command line)

bitwangdan commented 4 years ago

@hellohwang hi, I met the same problem as you, have you solved it?

tianzhi0549 commented 4 years ago

@hellohwang The warnings are normal. The breakdown should not be related to it.

wanghao9610 commented 4 years ago

@tianzhi0549 After all, I know that is a normal warning, but there is an unexpected breaking down issue that exists when the training process starts, I don't know the reason.

tianzhi0549 commented 4 years ago

@hellohwang I suggest you re-clone the project and re-compile it.

wanghao9610 commented 4 years ago

@bitwangdan hi, guys. I have a long time don't use the code, finally, I just tried another code that ever can run successfully, it worked. I suggest you can change your code from the ever successfully run code.

wanghao9610 commented 4 years ago

@tianzhi0549 Thanks a lot. I will try.