[BUG] wrong import of tools/train_net.py

[rayleizhu]

Sometimes, python interpreter can find the wrong train_net.py (i.e. detrex/detectron2/tools/train_net.py). It is due to in PYTHONPATH, detrex/detection2/ is in front of detrex/.

Possible workarounds are:

1. pip uninstall detrex && pip uninstall detectron2 && pip install -e . && pip install -e detectron2 (tested)
2. export PYTHONPATH=${PWD}:${PYTHONPATH} (not tested)

See if you have more elegant solutions.

[rentainhe]

We will check this later~

[rentainhe]

Sometimes, python interpreter can find the wrong train_net.py (i.e. detrex/detectron2/tools/train_net.py). It is due to in PYTHONPATH, detrex/detection2/ is in front of detrex/.

Possible workarounds are:

1. pip uninstall detrex && pip uninstall detectron2 && pip install -e . && pip install -e detectron2 (tested)
2. export PYTHONPATH=${PWD}:${PYTHONPATH} (not tested)

See if you have more elegant solutions.

I have the same PYTHONPATH but I can not reproduce your bug here, we will do more tests on this

[rayleizhu]

I tried installation on two machines. On the first one, it works well. On the second machine, when I run tools/train_net.py, an error occurs as below.

File "/mnt/lustre/zhulei1/GitRepo/detrex-custom/detectron2/tools/train_net.py", line 125, in main
    cfg = setup(args)
  File "/mnt/lustre/zhulei1/GitRepo/detrex-custom/detectron2/tools/train_net.py", line 117, in setup
    cfg.merge_from_file(args.config_file)
  File "/mnt/lustre/zhulei1/GitRepo/detrex-custom/detectron2/detectron2/config/config.py", line 46, in merge_from_file
    loaded_cfg = self.load_yaml_with_base(cfg_filename, allow_unsafe=allow_unsafe)
  File "/mnt/lustre/zhulei1/miniconda3/envs/detrex/lib/python3.7/site-packages/fvcore/common/config.py", line 61, in load_yaml_with_base
    cfg = yaml.safe_load(f)
  File "/mnt/lustre/zhulei1/miniconda3/envs/detrex/lib/python3.7/site-packages/yaml/__init__.py", line 125, in safe_load
    return load(stream, SafeLoader)
  File "/mnt/lustre/zhulei1/miniconda3/envs/detrex/lib/python3.7/site-packages/yaml/__init__.py", line 81, in load
    return loader.get_single_data()
  File "/mnt/lustre/zhulei1/miniconda3/envs/detrex/lib/python3.7/site-packages/yaml/constructor.py", line 49, in get_single_data
    node = self.get_single_node()
  File "/mnt/lustre/zhulei1/miniconda3/envs/detrex/lib/python3.7/site-packages/yaml/composer.py", line 39, in get_single_node
    if not self.check_event(StreamEndEvent):
  File "/mnt/lustre/zhulei1/miniconda3/envs/detrex/lib/python3.7/site-packages/yaml/parser.py", line 98, in check_event
    self.current_event = self.state()
  File "/mnt/lustre/zhulei1/miniconda3/envs/detrex/lib/python3.7/site-packages/yaml/parser.py", line 174, in parse_document_start
    self.peek_token().start_mark)
yaml.parser.ParserError: expected '<document start>', but found '<scalar>'

Obviously, python found the wrong file /mnt/lustre/zhulei1/GitRepo/detrex-custom/detectron2/tools/train_net.py. I'm not sure why this happens. Anyway, this is not a fatal problem and has already some workarounds.

[rentainhe]

I tried installation on two machines. On the first one, it works well. On the second machine, when I run tools/train_net.py, an error occurs as below.

File "/mnt/lustre/zhulei1/GitRepo/detrex-custom/detectron2/tools/train_net.py", line 125, in main
    cfg = setup(args)
  File "/mnt/lustre/zhulei1/GitRepo/detrex-custom/detectron2/tools/train_net.py", line 117, in setup
    cfg.merge_from_file(args.config_file)
  File "/mnt/lustre/zhulei1/GitRepo/detrex-custom/detectron2/detectron2/config/config.py", line 46, in merge_from_file
    loaded_cfg = self.load_yaml_with_base(cfg_filename, allow_unsafe=allow_unsafe)
  File "/mnt/lustre/zhulei1/miniconda3/envs/detrex/lib/python3.7/site-packages/fvcore/common/config.py", line 61, in load_yaml_with_base
    cfg = yaml.safe_load(f)
  File "/mnt/lustre/zhulei1/miniconda3/envs/detrex/lib/python3.7/site-packages/yaml/__init__.py", line 125, in safe_load
    return load(stream, SafeLoader)
  File "/mnt/lustre/zhulei1/miniconda3/envs/detrex/lib/python3.7/site-packages/yaml/__init__.py", line 81, in load
    return loader.get_single_data()
  File "/mnt/lustre/zhulei1/miniconda3/envs/detrex/lib/python3.7/site-packages/yaml/constructor.py", line 49, in get_single_data
    node = self.get_single_node()
  File "/mnt/lustre/zhulei1/miniconda3/envs/detrex/lib/python3.7/site-packages/yaml/composer.py", line 39, in get_single_node
    if not self.check_event(StreamEndEvent):
  File "/mnt/lustre/zhulei1/miniconda3/envs/detrex/lib/python3.7/site-packages/yaml/parser.py", line 98, in check_event
    self.current_event = self.state()
  File "/mnt/lustre/zhulei1/miniconda3/envs/detrex/lib/python3.7/site-packages/yaml/parser.py", line 174, in parse_document_start
    self.peek_token().start_mark)
yaml.parser.ParserError: expected '<document start>', but found '<scalar>'

Obviously, python found the wrong file /mnt/lustre/zhulei1/GitRepo/detrex-custom/detectron2/tools/train_net.py. I'm not sure why this happens. Anyway, this is not a fatal problem and has already some workarounds.

I am wondering how did you install detectron2 and detrex~

Maybe you can follow the official installation guide to see if this problem can be solved~

[rayleizhu]

I remember that I followed the official installation guide to install detrex exactly. I think you can ignore my problem at this time. Those users who meet the same problem can find this issue and workarounds.