Skip to content

feat(vf): implement transform_to_unconstrained z-scoring for vector field estimators#1889

Closed
BHARATH0153 wants to merge 3 commits into
sbi-dev:mainfrom
BHARATH0153:vf-estimators
Closed

feat(vf): implement transform_to_unconstrained z-scoring for vector field estimators#1889
BHARATH0153 wants to merge 3 commits into
sbi-dev:mainfrom
BHARATH0153:vf-estimators

Conversation

@BHARATH0153

@BHARATH0153 BHARATH0153 commented Jun 24, 2026

Copy link
Copy Markdown
Contributor
  • ConditionalVectorFieldEstimator: add prior_transform property
  • FlowMatchingEstimator: apply transform in ode_fn() and loss()
  • ScoreEstimator (ConditionalScore, VP, SubVP, VE): same pattern
  • VectorFieldBasedPotential: transform theta, add log|det J| correction
  • VectorFieldPosterior: sample paths inverse-transform
  • build_vector_field_estimator: wire mcmc_transform, add x_dist param

… field estimators

Closes sbi-dev#1885 (vector field portion)

- Remove assert_transform_to_unconstrained_supported guard from
  build_vector_field_estimator
- Add x_dist parameter; build mcmc_transform when z_score_x is
  transform_to_unconstrained
- Add _input_transform attribute and input_transform property to
  ConditionalVectorFieldEstimator base class
- Update FlowMatchingEstimator.ode_fn() and .loss() to apply the
  transform before forward/loss computation
- Update score estimators (ConditionalScore, VP, SubVP, VE) same
  pattern in ode_fn() and loss()
- Update VectorFieldBasedPotential.__call__() to transform theta,
  add log|det J| correction
- Update VectorFieldPosterior.sample_via_ode() and
  _sample_via_diffusion() to inverse-transform samples
@BHARATH0153
BHARATH0153 deleted the vf-estimators branch June 24, 2026 15:30
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant