Random Search Sampler

Generate n reproducible random-search configurations with np.random.default_rng(seed). For each configuration, in order, draw the learning rate as 10 ** rng.uniform(-5, -1) and then the tree depth as rng.integers(2, 10).

Implement solve(n, seed=0).

Returns. Return a list of n dicts {'lr': float, 'depth': int} with lr between ${10}^{-5}10^\{-5\}$ and ${10}^{-1}10^\{-1\}$ and depth between 2 and 9.

Examples

Example 1

Input:

solve(2, seed=0)

Output:

[{'lr': 0.003531, 'depth': 6}, {'lr': 1.5e-05, 'depth': 4}]

Example 2

Input:

solve(1, seed=7)

Output:

[{'lr': 0.003165, 'depth': 7}]