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
"PPPL deep learning disruption prediction package" repository description
"Fusion Recurrent Neural Network" (FRNN) title of README.md and internal project reference name
are inconsistent and somewhat misleading names for our efforts. While deep RNNs have been the main thrust of the project since its inception in the summer of 2016, using "FRNN" as a shorthand name might be too narrow going forward. Already, the repository contains implementations of a variety of shallow ML methods: random forests, SVMs, single hidden layer perceptrons, and gradient-boosted trees. We have added, or are about to add, deep learning methods that are not based on RNNs, e.g. TCN and Transformers.
Also, not sure if we want to keep PPPL in our GitHub org name, given that we have researchers and team members elsewhere.
I am coming around to "FusionDL". The name of our AESP Project's Unix Group is fusiondl_aesp, and I am now using "FusionDL" in the project tracking documents at ALCF.
Activity
felker commentedon Jan 23, 2020
Also, I have started referring to the project as DeepFusionAESP (Aurora Early Science Program) to distinguish it from the first-principles simulation-based project in the A21 ESP with PI C.S. Chang:
https://www.alcf.anl.gov/science/projects/high-fidelity-simulation-fusion-reactor-boundary-plasmas
felker commentedon Jun 9, 2020
I am coming around to "FusionDL". The name of our AESP Project's Unix Group is
fusiondl_aesp, and I am now using "FusionDL" in the project tracking documents at ALCF.