ray.tune.search.Searcher.restore

Searcher.restore(checkpoint_path: str)

恢复此搜索算法的状态

参数:

checkpoint_path – 保存搜索算法状态的文件。此路径应与提供给“save”的路径相同。

示例：

search_alg.save("./my_favorite_path.pkl")

search_alg2 = Searcher(...)
search_alg2 = ConcurrencyLimiter(search_alg2, 1)
search_alg2.restore(checkpoint_path)
tuner = tune.Tuner(
    cost,
    tune_config=tune.TuneConfig(
        search_alg=search_alg2,
        num_samples=5
    ),
)
tuner.fit()