I use ray tune HyperBandForBOHB function as schedule, the optimization method is BOHB, and the setting is

bohb = HyperBandForBOHB(

time_attr=‘training_iteration’,

max_t=5,

reduction_factor=3,

)

I set num_sample=50, I want to know how it calculate each round train how many trials?

here is the log:

Round #0:

Bracket(Max Size (n)=1, Milestone (r)=3, completed=100.0%): {TERMINATED: 3}

Bracket(Max Size (n)=2, Milestone (r)=5, completed=100.0%): {TERMINATED: 2}

Round #1:

Bracket(Max Size (n)=1, Milestone (r)=3, completed=100.0%): {TERMINATED: 3}

Bracket(Max Size (n)=2, Milestone (r)=5, completed=100.0%): {TERMINATED: 2}

Round #2:

Bracket(Max Size (n)=1, Milestone (r)=3, completed=100.0%): {TERMINATED: 3}

Bracket(Max Size (n)=2, Milestone (r)=5, completed=100.0%): {TERMINATED: 2}

Round #3:

Bracket(Max Size (n)=1, Milestone (r)=3, completed=100.0%): {TERMINATED: 3}

Bracket(Max Size (n)=2, Milestone (r)=5, completed=100.0%): {TERMINATED: 2}

Round #4:

Bracket(Max Size (n)=1, Milestone (r)=3, completed=100.0%): {TERMINATED: 3}

Bracket(Max Size (n)=2, Milestone (r)=5, completed=100.0%): {TERMINATED: 2}

Round #5:

Bracket(Max Size (n)=1, Milestone (r)=3, completed=100.0%): {TERMINATED: 3}

Bracket(Max Size (n)=2, Milestone (r)=5, completed=100.0%): {TERMINATED: 2}

Round #6:

Bracket(Max Size (n)=1, Milestone (r)=3, completed=100.0%): {TERMINATED: 3}

Bracket(Max Size (n)=2, Milestone (r)=5, completed=100.0%): {TERMINATED: 2}

Round #7:

Bracket(Max Size (n)=1, Milestone (r)=3, completed=100.0%): {TERMINATED: 3}

Bracket(Max Size (n)=2, Milestone (r)=5, completed=100.0%): {TERMINATED: 2}

Round #8:

Bracket(Max Size (n)=1, Milestone (r)=3, completed=100.0%): {TERMINATED: 3}

Bracket(Max Size (n)=2, Milestone (r)=5, completed=100.0%): {TERMINATED: 2}

Round #9:

Bracket(Max Size (n)=1, Milestone (r)=3, completed=100.0%): {TERMINATED: 3}

Bracket(Max Size (n)=2, Milestone (r)=5, completed=100.0%): {TERMINATED: 2}

it seems each round has 1 trail training 3 iterations, 2 trial training 5 iterations, I want to know how adjust these setting, I think is related to max_t and reduction_factor, but how the relation between num_sample=50, max_t=5 and reduction_factor=3?