From ec4b9b468ed0226520bb4da9049475f560926856 Mon Sep 17 00:00:00 2001 From: David Zwicker Date: Sat, 13 Apr 2024 20:35:16 +0200 Subject: [PATCH] Adjusted README.md --- README.md | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/README.md b/README.md index fa19dd4..3a60638 100644 --- a/README.md +++ b/README.md @@ -184,7 +184,7 @@ def multiply(a=1, b=2): return a * b if __name__ == "__main__": - submit_job(__file__, parameters={'a': 2}, output="data.hdf5", method="foreground") + submit_job(__file__, parameters={"a": 2}, output="data.hdf5", method="foreground") ``` Here, the `output` argument specifies a file to which the results are written, while `method` chooses how the script is submitted. @@ -201,7 +201,7 @@ def multiply(a=1, b=2): if __name__ == "__main__": for a in range(5): - submit_job(__file__, parameters={'a': a}, output=f"data_{a}.hdf5", method="foreground") + submit_job(__file__, parameters={"a": a}, output=f"data_{a}.hdf5", method="foreground") ``` Note that the safe-guard `if __name__ == "__main__"` is absolutely crucial to ensure that @@ -219,7 +219,7 @@ def multiply(a=1, b=2): return a * b if __name__ == "__main__": - submit_jobs(__file__, parameters={'a': range(5)}, output_folder="data", method="foreground") + submit_jobs(__file__, parameters={"a": range(5)}, output_folder="data", method="foreground") ``` Finally, the packages also offers a method to submit a model script to the cluster using @@ -259,7 +259,7 @@ For instance, the data from the multiple jobs ran above can be collected using from modelrunner import ResultCollection results = ResultCollection.from_folder(".", pattern="data_*.hdf5") -print(results.dataframe) +print(results.to_dataframe()) ``` This example should print all results using a pandas Dataframe, where each row