Config file seeds
- when we comment the seed property line, it does add randomness and the results change however the final config file still hase seed: 2
If possible it would be better to have the seed that was drawn by PNRG (so that we can relaunch the exact them run by using this seed the next time, or at least to have no seed at all in the final config
(note that if i change seed to 5, it has 5 in the final config, so it's only when the property was not set that it records the default one (2) even without using it.
-
maybe the config file generated by default should have the line commented out? Im checking with @thsanche and @pjobic what they think is the most natural in the DL community. @embray you have a strong opinion on this ? Anyway for now Im adding warnings in the training doc so that its not a big issue one way or the other (plus people would realize quite fast that they have completely identical results...)
-
A point that is not important. In the dataset and simulation config file
data_source:
# string template for per-replicate simulation files in Python
# string template format; the following template variables may be
# used: 'name', the same as the name property used in this config
# file; 'scenario', the scenario number, and 'replicate', the
# replicate number of the scenario (if there are multiple
# replicates); path separators may also be used in the template to
# form a directory structure
filename_format: scenario_{scenario}/{dataset_name}_{scenario}_{replicate}.npz
# used: 'name', the same as the name property used in this config
should read
# used: 'dataset_name', the same as the dataset_name property used in this config