latest

Bayesian Optimization in PyTorch

Introduction
Get Started
Tutorials

Key Features

Modular

Plug in new models, acquisition functions, and optimizers.

Built on PyTorch

Easily integrate neural network modules. Native GPU & autograd support.

Scalable

Support for scalable GPs via GPyTorch. Run code on multiple devices.

References

BoTorch: A Framework for Efficient Monte-Carlo Bayesian Optimization
@inproceedings{balandat2020botorch,
  title = {{BoTorch: A Framework for Efficient Monte-Carlo Bayesian Optimization}},
  author = {Balandat, Maximilian and Karrer, Brian and Jiang, Daniel R. and Daulton, Samuel and Letham, Benjamin and Wilson, Andrew Gordon and Bakshy, Eytan},
  booktitle = {Advances in Neural Information Processing Systems 33},
  year = 2020,
  url = {http://arxiv.org/abs/1910.06403}
}
Check out some other papers using BoTorch.

Get Started

  1. Install BoTorch:

    via pip (recommended):
    pip install botorch
    via Anaconda (from the unofficial conda-forge channel):
    conda install botorch -c gpytorch -c conda-forge

  2. Fit a model:

    import torch
from botorch.models import SingleTaskGP
from botorch.models.transforms import Normalize, Standardize
from botorch.fit import fit_gpytorch_mll
from gpytorch.mlls import ExactMarginalLogLikelihood

train_X = torch.rand(10, 2, dtype=torch.double) * 2
Y = 1 - torch.linalg.norm(train_X - 0.5, dim=-1, keepdim=True)
Y = Y + 0.1 * torch.randn_like(Y)  # add some noise

gp = SingleTaskGP(
    train_X=train_X,
    train_Y=Y,
    input_transform=Normalize(d=2),
    outcome_transform=Standardize(m=1),
)
mll = ExactMarginalLogLikelihood(gp.likelihood, gp)
fit_gpytorch_mll(mll)

  3. Construct an acquisition function:

    from botorch.acquisition import LogExpectedImprovement

logEI = LogExpectedImprovement(model=gp, best_f=Y.max())

  4. Optimize the acquisition function:

    from botorch.optim import optimize_acqf

bounds = torch.stack([torch.zeros(2), torch.ones(2)]).to(torch.double)
candidate, acq_value = optimize_acqf(
    logEI, bounds=bounds, q=1, num_restarts=5, raw_samples=20,
)
candidate  # tensor([[0.2981, 0.2401]], dtype=torch.float64)