sbrl-github/research-rainfallradar
1
0
Fork 0
mirror of https://github.com/sbrl/research-rainfallradar synced 2024-12-22 06:05:01 +00:00
Code Issues Projects Releases Packages Wiki Activity

README: note about RANDSEED

Browse source
This commit is contained in:
Starbeamrainbowlabs 2024-08-30 19:02:18 +01:00
parent e908e4b693
commit 05fd955067
Signed by: sbrl
GPG key ID: 1BE5172E637709C2
1 changed files with 3 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

4
README.md
View file

@ -170,7 +170,9 @@ ARGS | Optional. Any additional arguments to pass to the python program.
> It is strongly advised that all filepaths do **NOT** contain spaces. > It is strongly advised that all filepaths do **NOT** contain spaces.
**Making predictions:** Set `PATH_CHECKPOINT` to point to a checkpoint file to make predictions with an existing model that you trained earlier instead of training a new one. Data is pulled from the given dataset, same as during training. The first `PREDICT_COUNT` items in the dataset are picked to make a prediction. **Making predictions:** Set `PATH_CHECKPOINT` to point to a checkpoint file to make predictions with an existing model that you trained earlier instead of training a new one. Data is pulled from the given dataset, same as during training. The first `PREDICT_COUNT` items in the dataset are picked to make a prediction.
**`RANDSEED`:** The experiment runs in the series in the conference paper (see above!) uses `RANDSEED=QPnwyRZbLJBaTS7pjo4JZcl8vy9yv1D0SCpbS6olD5cmy`
> [!NOTE] > [!NOTE]
> The dataset pipeline is naturally non-deterministic with respect to the order in which samples are read. Ensuring the ordering of samples is not mangled is only possible when making predictions, and requires a number of environment variables to be set: > The dataset pipeline is naturally non-deterministic with respect to the order in which samples are read. Ensuring the ordering of samples is not mangled is only possible when making predictions, and requires a number of environment variables to be set: