struct BanditAgent {
int kind;
unsigned long k;
&FloatTensor counts;
&FloatTensor values;
float param;
float baseline;
unsigned long t;
struct Rng rng;
}Fields
pulls per arm
sample means (preferences for BANDIT_GRADIENT)
mean reward so far
Methods
= drop()
§
static struct BanditAgent new(int kind, unsigned long k, float param, unsigned long seed)source · definition