Notes from reading the Overleaf

[gully]

• State more numbers-- the number of grid points you looked at. The number of unique Teff grid points, unique logg grid points, etc.
• We repeated refer to the subset of the PHOENIX model grid the "PHOENIX subset". This almost seems like overly honest methods, since we could simply call it the "model grid" which is more familiar to most folks. I don't have a strong feeling here, just noticed reading that phrase a lot.
• We should cite the Yuan-Sen Ting work on spectral inference if we don't already
• Why does the stored interpolator take up so much more space (13 GB) than the pretrained models themselves(0.5 GB)? What is the intuition here for the factor of >20?
• What ever happend to the phrase "bi-directional mapping"?