[Model/Framework] in the model_zoo.py the torch.hub api use wrong

[wangliangwu]

Related to Model/Framework(s) (e.g. GNMT/PyTorch or FasterTransformer/All) PyTorch/Segmentation/MaskRCNN/pytorch/maskrcnn_benchmark/utils/model_zoo.py Describe the bug model_zoo.py at the line 10 from torch.hub import _download_url_to_file must change to from torch.hub import load_state_dict_from_url as _download_url_to_file

because I read the torch api，The i find the torch.hub did not _download_url_to_file method only have load_state_dict_from_url so i rename the _download_url_to_file as load_state_dict_from_url

ref link: https://pytorch.org/docs/1.10/hub.html https://pytorch.org/docs/1.11/hub.html https://pytorch.org/docs/1.13/model_zoo.html

To Reproduce Steps to reproduce the behavior:

1. Install '...'
2. Set "..."
3. Launch '...'

Expected behavior A clear and concise description of what you expected to happen. when I use the pytorch 1.14 (I use the NGC nvcr.io/nvidia/pytorch:23.02-py3 container ),

I use this command : cd /workspace/object_detection && python -m torch.distributed.launch --nproc_per_node=1 tools/test_net.py --config-file configs/e2e_mask_rcnn_R_50_FPN_1x.yaml --skip-eval DATASETS.TEST "(\"coco_2014_minival\",)" DTYPE float16 OUTPUT_DIR /results TEST.IMS_PER_BATCH 1

the i will get this error： ImportError: cannot import name '_download_url_to_file' from 'torch.utils.model_zoo' (/usr/local/lib/python3.8/dist-packages/torch/utils/model_zoo.py)

During handling of the above exception, another exception occurred:

Traceback (most recent call last): File "tools/test_net.py", line 15, in from maskrcnn_benchmark.utils.checkpoint import DetectronCheckpointer File "/workspace/object_detection/maskrcnn_benchmark/utils/checkpoint.py", line 10, in from maskrcnn_benchmark.utils.model_zoo import cache_url File "/workspace/object_detection/maskrcnn_benchmark/utils/model_zoo.py", line 10, in from torch.hub import _download_url_to_file ImportError: cannot import name '_download_url_to_file' from 'torch.hub' (/usr/local/lib/python3.8/dist-packages/torch/hub.py) /usr/local/lib/python3.8/dist-packages/torch/distributed/launch.py:180: FutureWarning: The module torch.distributed.launch is deprecated and will be removed in future. Use torchrun. Note that --use_env is set by default in torchrun. If your script expects --local_rank argument to be set, please change it to read from os.environ['LOCAL_RANK'] instead. See https://pytorch.org/docs/stable/distributed.html#launch-utility for further instructions

Environment Please provide at least:

• Container version (e.g. pytorch:19.05-py3):nvcr.io/nvidia/pytorch:23.02-py3
• GPUs in the system: (e.g. 8x Tesla V100-SXM2-16GB): NVIDIA A10
• CUDA driver version (e.g. 418.67):CUDA Version: 12.0