Neither Single nor Multi object model training works

[paucarre]

When running Single object model training: according to documentation in NGDF Training

ros@rusalka:~/NGDF$ bash scripts/train/perobj_Bottle.sh

It produces the following error

/home/ros/.local/lib/python3.8/site-packages/pkg_resources/__init__.py:123: PkgResourcesDeprecationWarning: 1.13.1-unknown is an invalid version and will not be supported in a future release
  warnings.warn(
ngdf/train.py:32: UserWarning:
The version_base parameter is not specified.
Please specify a compatability version level, or None.
Will assume defaults for version 1.1
usage: train.py [--help] [--hydra-help] [--version] [--cfg {job,hydra,all}] [--resolve] [--package PACKAGE] [--run] [--multirun] [--shell-completion] [--config-path CONFIG_PATH]
                [--config-name CONFIG_NAME] [--config-dir CONFIG_DIR] [--experimental-rerun EXPERIMENTAL_RERUN] [--info [{all,config,defaults,defaults-tree,plugins,searchpath}]]
                [overrides [overrides ...]]
train.py: error: unrecognized arguments: --experiment_yml=perobj_Bottle --data_root=/home/thomasweng/projects/ngdf/data --project_root=/home/thomasweng/projects/ngdf
ros@rusalka:~/NGDF$ 

When running multiobject

ros@rusalka:~/NGDF$ bash scripts/train/multobj_Bottle.sh

I get a similar error

/home/ros/.local/lib/python3.8/site-packages/pkg_resources/__init__.py:123: PkgResourcesDeprecationWarning: 1.13.1-unknown is an invalid version and will not be supported in a future release
  warnings.warn(
ngdf/train.py:32: UserWarning:

The version_base parameter is not specified.
Please specify a compatability version level, or None.
Will assume defaults for version 1.1

usage: train.py [--help] [--hydra-help] [--version] [--cfg {job,hydra,all}] [--resolve] [--package PACKAGE] [--run] [--multirun] [--shell-completion] [--config-path CONFIG_PATH]
                [--config-name CONFIG_NAME] [--config-dir CONFIG_DIR] [--experimental-rerun EXPERIMENTAL_RERUN] [--info [{all,config,defaults,defaults-tree,plugins,searchpath}]]
                [overrides [overrides ...]]
train.py: error: unrecognized arguments: --experiment_yml=multobj_Bottle --data_root=/home/thomasweng/projects/ngdf/data --project_root=/home/thomasweng/projects/ngdf
ros@rusalka:~/NGDF$ 

[thomasweng15]

Hi Pau, thanks for raising the issue. You may need to pull the latest code, as this commit replaced the hard-coded paths that you encountered for --data_root for example, sorry about that. With the latest code, do you still get the unrecognized arguments error? It may be a bug I haven't caught yet.

[paucarre]

Single object model training: works perfectly rebasing to last commit, appologies for that!

The multiobject doesn't because of this error:

OSError: Unable to open file (unable to open file: name = '/home/ros/NGDF/scripts/train/../../data/acronym_multobj/grasp-dataset/dataset.hdf5', errno = 2, error message = 'No such file or directory', flags = 0, o_flags = 0)

But I couldn't find the acronym_multobj dataset.