PyMC v3.1 Release Notes
Release Date: 2017-06-23 // almost 7 years ago-
๐ New features
๐ New user forum at http://discourse.pymc.io
๐ Much improved variational inference support:
- Add Operator Variational Inference (experimental).
- Add Stein-Variational Gradient Descent as well as Amortized SVGD (experimental).
- Add pm.Minibatch() to easily specify mini-batches.
- Added various optimizers including ADAM.
- Stopping criterion implemented via callbacks.
0๏ธโฃ sample() defaults changed: tuning is enabled for the first 500 samples which are then discarded from the trace as burn-in.
๐ MvNormal supports Cholesky Decomposition now for increased speed and numerical stability.
Many optimizations and speed-ups.
NUTS implementation now matches current Stan implementation.
โ Add higher-order integrators for HMC.
ADVI stopping criterion implemented.
๐ Improved support for theano's floatX setting to enable GPU computations (work in progress).
๐ MvNormal supports Cholesky Decomposition now for increased speed and numerical stability.
โ Added support for multidimensional minibatches
โ Added
Approximation
class and the ability to convert a sampled trace into an approximation via itsEmpirical
subclass.๐
Model
can now be inherited from and act as a base class for user specified models (see pymc3.models.linear).โ Add MvGaussianRandomWalk and MvStudentTRandomWalk distributions.
GLM models do not need a left-hand variable anymore.
๐จ Refactored HMC and NUTS for better readability.
โ Add support for Python 3.6.
๐ Fixes
Bound now works for discrete distributions as well.
Random sampling now returns the correct shape even for higher dimensional RVs.
๐ Use theano Psi and GammaLn functions to enable GPU support for them.