test error torch1.7

run demo.py in torch1.9 is ok?but in torch1.7,i get error as follows?
Traceback (most recent call last): File "/data/zhangyong/mypytoch/xyb/inference_x2.py", line 332, in <module> run() File "/data/zhangyong/mypytoch/xyb/inference_x2.py", line 236, in run outputs, img_info = predictor.inference(pic_path) File "/data/zhangyong/mypytoch/xyb/inference_x2.py", line 88, in inference outputs = self.model(img) File "/root/anaconda3/envs/game3/lib/python3.6/site-packages/torch/nn/modules/module.py", line 722, in _call_impl result = self.forward(*input, **kwargs) File "/data/zhangyong/mypytoch/xyb/model/yolox/models/yolox.py", line 46, in forward outputs = self.head(fpn_outs) File "/root/anaconda3/envs/game3/lib/python3.6/site-packages/torch/nn/modules/module.py", line 722, in _call_impl result = self.forward(*input, **kwargs) File "/data/zhangyong/mypytoch/xyb/model/yolox/models/yolo_head.py", line 194, in forward return self.decode_outputs(outputs, dtype=torch.cuda.FloatTensor) File "/data/zhangyong/mypytoch/xyb/model/yolox/models/yolo_head.py", line 228, in decode_outputs strides.append(torch.full((*shape, 1), stride)) RuntimeError: Providing a bool or integral fill value without setting the optional dtypeoroutarguments is currently unsupported. In PyTorch 1.7, whendtypeandoutare not set a bool fill value will return a tensor of torch.bool dtype, and an integral fill value will return a tensor of torch.long dtype.

Reproduce the failure by running demo.py with PyTorch 1.7, then inspect model/yolox/models/yolo_head.py around line 228 in decode_outputs and compare the behavior with PyTorch 1.9. Done when the affected inference path runs under PyTorch 1.7 without the reported fill-value error and the newer version remains working.