Go to file
2020-04-21 13:38:02 -04:00
map2map Add explicit barrier after init dist groups 2020-04-21 13:38:02 -04:00
scripts Add runtime address and port determination and share them via file 2020-02-13 19:56:54 -06:00
.gitignore Initial commit 2019-11-30 15:27:25 -05:00
LICENSE Initial commit 2019-11-30 15:27:25 -05:00
README.md Add TOC 2020-04-21 12:34:14 -04:00
setup.py Change tensorboard to optional dependency 2020-01-23 09:58:54 -05:00

map2map

Neural network emulators to transform field/map data

Installation

Install in editable mode

pip install -e .

Usage

Take a look at the examples in scripts/*.slurm, and the command line options in map2map/args.py or by m2m.py -h.

Data

Structure your data to start with the channel axis and then the spatial dimensions. Put each sample in one file. Specify the data path with glob patterns.

Data normalization

Input and target (output) data can be normalized by functions defined in map2map2/data/norms/.

Model

Find the models in map2map/models/. Customize the existing models, or add new models there and edit the __init__.py.

Training

Files generated

  • *.out: job stdout and stderr
  • state_*.pth: training state including the model parameters
  • checkpoint.pth: symlink to the latest state
  • runs/: directories of tensorboard logs

Tracking

Install tensorboard and launch it by

tensorboard --logdir PATH --samples_per_plugin images=IMAGES --port PORT
  • Use . as PATH in the training directory, or use the path to some parent directory for tensorboard to search recursively for multiple jobs.
  • Show IMAGES images, or all of them by setting it to 0.
  • Pick a free PORT. For remote jobs, do ssh port forwarding.