warp-tpe-sampler 0.1.2

WarpTpeSampler: Cached TPE + embedded budget policy for Optuna

warp-tpe-sampler

WarpTpeSampler for Optuna: CachedTPESampler + embedded budget policy (refresh/freeze/random).

Install

pip install warp-tpe-sampler

Quickstart

import optuna
from warp_tpe_sampler import WarpTpeSampler, WarpTpeConfig

sampler = WarpTpeSampler(WarpTpeConfig(seed=0, n_startup_trials=20))
study = optuna.create_study(direction="minimize", sampler=sampler)

def objective(trial):
    x = trial.suggest_float("x", 0.0, 1.0)
    return (x - 0.37) ** 2

study.optimize(objective, n_trials=200, n_jobs=1)

Notes

Intended usage: n_jobs == 1.

Depends on Optuna internals; see supported Optuna versions in pyproject.toml.