spacekit - JWST calibration pipeline resource prediction modeling

Inference

Generate estimated memory footprints on unlabeled data.

• predict
  • JwstCalPredict
  • load_pretrained_model()
  • predict_handler()

Setup

Install with pip

$ pip install spacekit

$ git clone https://github.com/spacetelescope/spacekit
$ cd spacekit
$ pip install -e .

Run Inference

*from the command line*

$ python -m spacekit.skopes.jwst.cal.predict /path/to/inputs

*from python*

from spacekit.skopes.jwst.cal.predict import JwstCalPredict
input_path = "/path/to/level1/exposures"
jcal = JwstCalPredict(input_path)
jcal.run_inference()
jcal.predictions
{
    'jw01076-o101-t1_nircam_clear-f212n': {'gbSize': 10.02},
    'jw01076-o101-t1_nircam_clear-f210m': {'gbSize': 8.72},
    'jw01076-o101-t1_nircam_clear-f356w': {'gbSize': 7.38},
}

Outputs: dictionary of level 3 products and estimated memory footprint (GB)