Skip to content

Latest commit

 

History

History
9 lines (8 loc) · 990 Bytes

optionals.md

File metadata and controls

9 lines (8 loc) · 990 Bytes
Raw

In our study, we compared performances of different representations.

  • Specifically, we focused on: Magpie, Roost and CrabNet.
  • We generate learned embeddings for a material composition by extracting features from trained Roost and CrabNet models.
  • The models we used for this purpose are provided in this repository within saved_models/transfer_learning directory.
  • You will need to move the checkpoint corresponding to Roost inside the Roost directory as: roost/roost/models/oqmd_100_epochs_model/checkpoint-r0.pth.tar
  • Move the checkpoint for CrabNet in the CrabNet directory as: CrabNet/models/trained_models/aflow__agl_thermal_conductivity_300K.pth
  • This will enable you to generate features using Roost and CrabNet.
  • For generating features with CrabNet, you will need to inherit the class and remove the final layers. Roost provides a direct utility for performing this.