TECHNOLOGY PROFILE · TECH-0024
BoTorch
Bayesian optimization toolkit built on PyTorch.
Reference link only — not reviewed this edition. This profile preserves the prepared catalog’s original descriptions. Advantages depend on the workload; comparisons are editorial, not benchmark results.
Where it has an edge
Differentiable acquisition functions and custom Gaussian-process models suit research beyond standard tuning APIs.
Best-fit problems
Expensive black-box experiments and novel Bayesian optimization methods.
Tradeoffs & cautions
Surrogate modeling and acquisition optimization require statistical expertise.
What to test before choosing
Reproduce a run from its recorded code/data/config; test artifact access, search budget and validation leakage.
Runtime & hardware
Local, server or managed
Access & licensing
Open-source cores; hosted/commercial features vary
Execution
Local / self-hosted
Lifecycle
Not independently audited
Comparison basis
Editorial problem-fit judgement; no head-to-head benchmark performed
Connected concepts
Alternatives to investigate
Optuna · Ax · SciPy