search

hassonlab / 247-encoding

Contains python scripts for performing encoding on 247 data.
0 stars 9 forks source link

Dev branch #42

Closed VeritasJoker closed 2 years ago

VeritasJoker commented 2 years ago

Include changes from the other two PRs and more:

  1. pkl_hierarchy in correspondence with 247-pickling
  2. fixed encoding bug in encoding (fold=10, correct train/test fold and mean-centering)
  3. clean up some plotting code
  4. Fixed 'glove50' and 'glove' inconsistency (need to doublecheck)
  5. Added pca as argument
VeritasJoker commented 2 years ago

Still under development. Will test everything as I rerun all encoding on della.

hvgazula commented 2 years ago

My 2 cents: anything to do with embeddings should happen in pickling.. encoding should purely be encoding. Because..pickling is the feeder. Also, encoding is going out of control with too many code changes to account for new experiments.

zkokaja commented 2 years ago

Replace common = True with np array of all Trues. Remove commented out code and then we can merge this.