Skip to content

Latest commit

 

History

History
36 lines (26 loc) · 1.41 KB

README.md

File metadata and controls

36 lines (26 loc) · 1.41 KB
Raw

drawing

Harnessing the power of deep learning to accurately and efficiently predict crucial properties of the stars in our galaxy.

Training StarNet

A script has been provided (starnet/train_StarNet.py) to make training StarNet simple. An example of how to use it to train the basic StarNet2017 architecture on a dataset named training_dataset.h5 to predict the parameters T_eff, logg, [M/H], [alpha/M]:

python /path/to/starnet/train_StarNet.py \
--data_path /path/to/training_dataset.h5 \
--num_train 50000 \
--targets teff logg M_H a_M \
--spec_key spectra_starnetnorm \
--save_folder results_stored_here/ \
--batch_size 32 \
--epochs 35 \
--model_type StarNet2017

NOTE: the --targets and --spec_key arguments expect the keywords that are used in your training_dataset.h5 file to store the training labels and training features, respectively.

Citing this work

Here's the BibTeX:

@article{bialek2020assessing,
  title={Assessing the performance of LTE and NLTE synthetic stellar spectra in a machine learning framework},
  author={Bialek, Spencer and Fabbro, S{\'e}bastien and Venn, Kim A and Kumar, Nripesh and O’Briain, Teaghan and Yi, Kwang Moo},
  journal={Monthly Notices of the Royal Astronomical Society},
  year={2020}
}