MCPcopy Create free account
hub / github.com/ImprintLab/MedSegDiff / normal_kl

Function normal_kl

guided_diffusion/losses.py:12–39  ·  view source on GitHub ↗

Compute the KL divergence between two gaussians. Shapes are automatically broadcasted, so batches can be compared to scalars, among other use cases.

(mean1, logvar1, mean2, logvar2)

Source from the content-addressed store, hash-verified

source not stored for this graph (policy: none)

Callers 2

_vb_terms_bpdMethod · 0.85
_prior_bpdMethod · 0.85

Calls

no outgoing calls

Tested by

no test coverage detected