parsnip

Module Contents

Functions

convert_to_lcdata_h5(dataset_csv, probs_csv, data_dir, ...)

Convert data folder to h5 file compatible with lcdata.

train_parsnip_model(dataset_hdf5, model_prefix)

Train ParSNIP model on custom device.

retrieve_decodings(sn_name, dataset_fn, model_fn)

Retrieve 100 decoded light curves for single

encode_with_parsnip(light_curves, labels, redshifts, ...)

Retrieve ParSNIP features and format

Attributes

DEVICE

DEFAULT_FEATURES

DEVICE = 'mps'[source]
DEFAULT_FEATURES = ['color', 's1', 's2', 's3', 'luminosity', 'reference_time'][source]
convert_to_lcdata_h5(dataset_csv, probs_csv, data_dir, save_dir)[source]

Convert data folder to h5 file compatible with lcdata.

train_parsnip_model(dataset_hdf5, model_prefix)[source]

Train ParSNIP model on custom device.

retrieve_decodings(sn_name, dataset_fn, model_fn)[source]

Retrieve 100 decoded light curves for single light curve name.

encode_with_parsnip(light_curves, labels, redshifts, model_path, save_dir, csv_path, feature_list=DEFAULT_FEATURES)[source]

Retrieve ParSNIP features and format to be read by get_posterior_samples.

Parameters:

  • light_curves (list of Lightcurve) – The light curves to encode into a PosteriorSamples object

  • model_path (str) – filepath of parsnip model

  • save_dir (str) – where to save posterior files.

  • csv_path (str) – where to save training CSV

  • feature_list (list of str) – which features to save