You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
By saving and loading spectra in HDF5 it allows you to only access a few spectra at any position in a list of spectra. Storing spectra in this format is especially helpful for training models, since there is no need to have all spectra in memory.
It stores the spectra like same size vectors. This means the spectra need to have a predefined number of fragments and if they have less, they need to be padded with zeroes. But due to efficient compression this is not really a memory issue.
By saving and loading spectra in HDF5 it allows you to only access a few spectra at any position in a list of spectra. Storing spectra in this format is especially helpful for training models, since there is no need to have all spectra in memory.
It stores the spectra like same size vectors. This means the spectra need to have a predefined number of fragments and if they have less, they need to be padded with zeroes. But due to efficient compression this is not really a memory issue.
@roman-bushuiev implemented something similar already for https://github.com/roman-bushuiev/DreaMS and will try to implement it for matchms, so make the DreaMS package build on matchms.
The text was updated successfully, but these errors were encountered: