All pull requests

Pull requests

• Fix Jacobian multiplicity for matrix-valued matmul logp

  bug
  #8571·jontb opened Oct 25, 2026·5 tasks done·

  ·
  1

  1

• Fix observation-noise variance in MarginalKron predictions

  bug
  #8570·jontb opened Oct 25, 2026·5 tasks done·

  ·
  1

  2

• Accept any narwhals-supported DataFrame or Series in pm.Data and as_tensor_variable

  #8569·ColtAllen opened Oct 25, 2026·

  ·
  1

• Rewrite more distributions symbolic for backend compatibility

  #8525·ricardoV94 opened Oct 23, 2026Member·

  ·
  2

• Support batched MatrixNormal logp evaluation

  #8521·S-h-u-b-h-1 opened Oct 23, 2026·

  ·
  3

• Fix acceptance ratio of CategoricalGibbsMetropolis proportional proposal

  #8517·mottopanikeiku opened Oct 22, 2026·6 of 11 tasks·

  ·
  1

  2

• Fix CAR forward sampling covariance

  #8514·mottopanikeiku opened Oct 20, 2026·6 of 11 tasks·

  ·
  1

  3

• Record step method sampling state in NDArray backend

  enhancementsfeature requesttrace-backend
  #8511·aniketpandey05 opened Oct 20, 2026·8 of 14 tasks·

  ·
  1

  2

• fix: exclude current gradient from its adagrad_window normaliser (#8444)

  #8508·AdityaBil opened Oct 19, 2026·11 tasks·

  ·
  1

• Support pandas Series and DataFrame inputs in Gaussian Processes

  bug
  #8496·hrideymarwah15 opened Oct 17, 2026Contributor·

  ·
  1

  2

• Add pm.Placeholder for variables that take their values from a trace

  bugfeature request
  #8489·kojikojikojikoji opened Oct 14, 2026Contributor·8 of 11 tasks·

  ·
  2

  2

• Support user callback with nutpie sampler

  feature request
  #8488·saheli56 opened Oct 13, 2026·

  ·
  1

  2

• Add Model.dependent_logp and use it in discrete Metropolis and Slice samplers

  enhancementssamplers
  #8476·bhaskargurram-ai opened Oct 12, 2026·6 of 11 tasks·

  ·
  1

  2

• Support pymc.dims models in vectorize_over_posterior

  #8469·raashish1601 opened Oct 6, 2026Contributor·5 of 11 tasks·

  ·
  1

  6

• Add explicit dtype support to pm.Data

  bug
  #8467·leninworld opened Oct 5, 2026·8 of 11 tasks·

  ·
  1

  9

• Convert integer and boolean DataFrames like Series in convert_data

  bug
  #8463·Eliaaazzz opened Oct 2, 2026·5 of 11 tasks·

  ·
  1

  1

• Port ZarrTrace backend to zarr 3

  enhancementstrace-backend
  #8462·williambdean opened Oct 1, 2026Contributor·

  ·
  1

  15

• Add icdf for AsymmetricLaplace distribution

  #8453·Sharman1661 opened Sep 25, 2026·4 of 11 tasks·

  ·
  9

• Compile the SMC trace function once and pass compile_kwargs to it

  bug
  #8450·RitiGrover opened Sep 23, 2026·4 of 11 tasks·

  ·
  1

  2

• Exclude current gradient from its adagrad_window normaliser

  #8447·shaneraphel opened Sep 20, 2026·6 tasks done·

  ·
  1

• Fix icdf precision in the far tails for Cauchy, Exponential and Weibull

  bug
  #8446·siddhant-shahhh opened Sep 20, 2026·6 of 11 tasks·

  ·
  1

  1

• Raise ShapeError for shape conflicts in observed data

  bug
  #8443·siddhant-shahhh opened Sep 18, 2026·4 of 7 tasks·

  ·
  1

  2

• Fix LogitNormal losing default logodds transform

  bug
  #8442·wolfgang-aura opened Sep 18, 2026·4 of 5 tasks·

  ·
  1

  11

• Fix logcdf lower-tail underflow for Cauchy and StudentT

  bug
  #8436·siddhant-shahhh opened Sep 14, 2026·6 of 11 tasks·

  ·
  1

  1

• ENH: Enable batch size > 1 for CAR distribution (#8414)

  #8434·ChiragKaushik4 opened Sep 14, 2026·4 of 9 tasks·

  ·
  1

  1