conda env create --prefix "prefix of your choice" --file mop_conda_env.yml
python src/mop/scripts/ray_trainer.py
This will start multiple training sessions with hyperparameters option defined in the file ray_trainer.py
and provided in the file src/mop/scripts/hparams.yaml
. The logs and the checkpoints will be available in the directory nmg_logs
.
Edit the checkpoint path (ckpt_path) in the file src/mop/scripts/eval.py
.
python src/nmg/scripts/eval.py
This will save the output ".npz" file in the same directory as the checkpoint. Visualize the results by running the src/mop/scripts/viz_scripts/viz_nmg.py
in the blender. For the reconstruction modes, the first half samples are the original and the rest half are the reconstructions.