ziatdinovmax/gpax

Issues

• Custom BNN architectures for a fully Bayesian DKL

  #88 opened 3 months ago by ziatdinovmax
  0

• Set noise prior automatically

  #87 opened 3 months ago by ziatdinovmax
  2

• Move `priors` out of utils

  #86 opened 4 months ago by ziatdinovmax
  0

• Minor issue regarding importing sPM model

  #83 opened 4 months ago by utkarshp1161
  2

• Sparse GP

  #80 opened 4 months ago by ziatdinovmax
  1

• Option to use regular NN in viDKL

  #77 opened 4 months ago by ziatdinovmax
  1

• Cost-aware BO

  #68 opened 4 months ago by ziatdinovmax
  1

• UIGP: Allow for different variance along different input feature dimensions

  #70 opened 4 months ago by ziatdinovmax
  1

• Add explanation/examples on how to use utils.priors

  #72 opened 4 months ago by ziatdinovmax
  1

• Quality-of-life suggestions for release, CI, semantic versioning, etc.

  #59 opened 7 months ago by ziatdinovmax
  3

• MCMC prediction stability issue - providing NAN values of variance (Issue in Google Colab GPU/HighRam setting)

  #34 opened 10 months ago by arpanbiswas52
  4

• Unable to use gpu: might be a jax(or jaxlib) issue

  #63 opened 7 months ago by utkarshp1161
  2

• Notebook hypoAL.ipynb appears to just be an older version of gpax_hypo.ipynb

  #61 opened 7 months ago by matthewcarbone
  2

• Implement faster smoke tests of notebooks

  #53 opened 7 months ago by matthewcarbone
  1

• ExactGP.predict and viGP.predict produce inconsistent shapes

  #49 opened 8 months ago by matthewcarbone
  8

• Batch acquisition (bototrch style)

  #58 opened 7 months ago by ziatdinovmax
  1

• Feature: implement simulated campaigning for "hyper parameter tuning"

  #33 opened 10 months ago by matthewcarbone
  7

• Quality-of-life suggestions for release, CI, semantic versioning, etc.

  #41 opened 8 months ago by matthewcarbone
  0

• Suggestion: refactor acquisition.py to a class structure

  #42 opened 8 months ago by matthewcarbone
  2

• Open discussions?

  #52 opened 8 months ago by matthewcarbone
  1

• Feature: allow the user to scale X and y before fitting, predicting, etc.

  #40 opened 8 months ago by matthewcarbone
  3

• Kernels with a different length scale on each axis

  #51 opened 8 months ago by matthewcarbone
  2

• Possible issue in indexes in batch acquisition

  #43 opened 8 months ago by matthewcarbone
  2

• Is it possible to filter nans for ExactGP.predict inside UCB Acquisition functions calls?

  #36 opened 9 months ago by avivajpeyi
  2

• Any suggestions on how to improve acquisition.UCB for active GP example?

  #35 opened 9 months ago by avivajpeyi
  2

• Notebook for Theory-informed data reconstruction and Bayesian optimization appears to be incorrect

  #30 opened 10 months ago by matthewcarbone
  1

• We consider noisy observations of a discontinuous function...

  #6 opened 2 years ago by jecampagne
  1

• Extension of kernel='RBF',...

  #7 opened 2 years ago by jecampagne
  13