Skip to contents

Validate a DQN replay buffer of (s, a, r, s2[, done]) rows (hmdqn)

Usage

morie_check_buffer(buffer, n_states, n_actions, name)

Arguments

buffer

List of length-4/5 transitions.

n_states, n_actions

Bounds.

name

Caller name for error messages.

Value

List with s,a,r,s2,done (0-based indices as given).

Examples

buf <- list(c(0L, 1L, 1.0, 2L, FALSE), c(2L, 0L, 0.5, 1L, TRUE))
morie_check_buffer(buf, n_states = 3L, n_actions = 2L, name = "demo")
#> $s
#> [1] 0 2
#> 
#> $a
#> [1] 1 0
#> 
#> $r
#> [1] 1.0 0.5
#> 
#> $s2
#> [1] 2 1
#> 
#> $done
#> [1] FALSE  TRUE
#>