0.3.2

gptorch 0.3

Change log

0.3.0

Changes breaking backward compatibility:

• GPR, VFE, SVGP: training inputs order is changed from (y, x) to (x, y) on
  model __init__()s.
• .predict() functions return the same type as the inputs provided
  (numpy.ndarray->numpy.ndarray, torch.Tensor->torch.Tensor)
• Remove util.as_variable()
• Remove util.tensor_type()
• Remove util.KL_Gaussian()
• Remove util.gammaln()
• GPModel method .loss() generally replaces .compute_loss().
• .compute_loss() methods in models generally renamed to .log_likelihood()
  and signs flipped to reflect the fact that the loss is generally the negative
  LL.

Changes not breaking backward compatibility:

• GPR, VFE: Allow specifying training set on .compute_loss() with x, y kwargs
• GPR, VFE: Allow specifying training inputs on ._predict() with x kwarg
• GPU supported with .cuda()
• Remove GPModel.evaluate()
• Don't print inducing inputs on sparse GP initialization
• Suport for priors in gptorch.model.Models

0.3.1

• Fix some places where .compute_loss() wasn't replaced, causing GPModel.optimize() not to work.

0.3.2

• Issue 20 related to installing gptorch on top of pip-installed versions of PyTorch with non-standard device configurations.
• Issue 22 where importing gptorch changes the default dtype in PyTorch from single- to double-precision.
• Added gptorch.__version__

Authors

• Steven Atkinson

0.3.1

gptorch 0.3

Change log

0.3.0

Changes breaking backward compatibility:

• GPR, VFE, SVGP: training inputs order is changed from (y, x) to (x, y) on
  model __init__()s.
• .predict() functions return the same type as the inputs provided
  (numpy.ndarray->numpy.ndarray, torch.Tensor->torch.Tensor)
• Remove util.as_variable()
• Remove util.tensor_type()
• Remove util.KL_Gaussian()
• Remove util.gammaln()
• GPModel method .loss() generally replaces .compute_loss().
• .compute_loss() methods in models generally renamed to .log_likelihood()
  and signs flipped to reflect the fact that the loss is generally the negative
  LL.

Changes not breaking backward compatibility:

• GPR, VFE: Allow specifying training set on .compute_loss() with x, y kwargs
• GPR, VFE: Allow specifying training inputs on ._predict() with x kwarg
• GPU supported with .cuda()
• Remove GPModel.evaluate()
• Don't print inducing inputs on sparse GP initialization
• Suport for priors in gptorch.model.Models

0.3.1

• Fix some places where .compute_loss() wasn't replaced, causing GPModel.optimize() not to work.

Authors

• Steven Atkinson

0.3.0

gptorch 0.3

Change log

0.3.0

Changes breaking backward compatibility:

• GPR, VFE, SVGP: training inputs order is changed from (y, x) to (x, y) on
  model __init__()s.
• .predict() functions return the same type as the inputs provided
  (numpy.ndarray->numpy.ndarray, torch.Tensor->torch.Tensor)
• Remove util.as_variable()
• Remove util.tensor_type()
• Remove util.KL_Gaussian()
• Remove util.gammaln()
• GPModel method .loss() generally replaces .compute_loss().
• .compute_loss() methods in models generally renamed to .log_likelihood()
  and signs flipped to reflect the fact that the loss is generally the negative
  LL.

Changes not breaking backward compatibility:

• GPR, VFE: Allow specifying training set on .compute_loss() with x, y kwargs
• GPR, VFE: Allow specifying training inputs on ._predict() with x kwarg
• GPU supported with .cuda()
• Remove GPModel.evaluate()
• Don't print inducing inputs on sparse GP initialization
• Suport for priors in gptorch.model.Models

Authors

• Steven Atkinson

0.2.3

gptorch 0.2

Change log

0.2.1

• Add missing .predict_f_samples() method to GPModel
• Add missing diag kwarg to GPModel.predict_f().

0.2.2

• Remove instances of torch.set_default_dtype() from codebase
• Add CircleCI and CodeCov.io

0.2.3

• Fix gradient-shunting behavior caused by torch.clamp() used in util.squared_distance()
• Replace deprecated uses of as_variable() in densities.py with as_tensor()

Authors

• Steven Atkinson

0.2.2

gptorch 0.2

Change log

0.2.1

• Add missing .predict_f_samples() method to GPModel
• Add missing diag kwarg to GPModel.predict_f().

0.2.2

• Remove instances of torch.set_default_dtype() from codebase
• Add CircleCI and CodeCov.io

Authors

• Steven Atkinson

0.2.1

Version 0.2.1

Change log

• Add missing .predict_f_samples() method to GPModel
• Add missing diag kwarg to GPModel.predict_f().

Authors

• Steven Atkinson

0.2.0

Version 0.2.0

Change log

• Support for PyTorch 1.0 up through the latest current release.
• Can initialize sparse GP inducing inputs using centers of k-means clusters
• Suppress warning messages by default when jitter is required for linear algebra operations

Authors

• Steven Atkinson

v0.1.0

Version 0.1.0

Initial Release.
For use with Python 2 or 3, PyTorch version 0.3.1

Features

• Models supported:
  • Vanilla GP regression
  • Sparse GP regression with variational inference of inducing points.
• Kernels supported:
  • White
  • Constant
  • RBF
  • Matern 5/2
  • Matern 3/2
  • Exponential (Matern 1/2)
  • Linear
  • Sum kernels
  • Product kernels
• Mean functions supported: zero only
• Training: optimization using a variety of PyTorch and Scipy optimizers
• Tests in place for most kernels, likelihoods, densities.

Contributors:

• Steven Atkinson
• Yinhao Zhu
• Nicholas Geneva