A 1D algorithmic environment where the target is the same
as the observation.
The 1D TapeAlgorithmicEnv has an action_space tuple
with shape [2,2,10]. The first item is the chosen movement,
either left or right. The second item is whether to write
to the current position or not. Finally, the last value is
the actual value to write to the tape in the current position.
The environment terminates immediately when an incorrect item
is written or when the entire output is successfully outputted.
The MIN_LENGTH of the tape is 5 plus a random number between
0 and 2.
example
import {Copy} from"gym-js";
let base = 10;
const env = new Copy(base);
console.log(env.action_space.toString());
> DiscreteTuple: 2,2,10console.log(env.observation_space.toString());
> Discrete: 11let action = env.action_space.sample();
let [obs, rew, done, info] = env.step(action);
A 1D algorithmic environment where the target is the same as the observation.
The 1D TapeAlgorithmicEnv has an
action_spacetuple with shape [2,2,10]. The first item is the chosen movement, either left or right. The second item is whether to write to the current position or not. Finally, the last value is the actual value to write to the tape in the current position. The environment terminates immediately when an incorrect item is written or when the entire output is successfully outputted. TheMIN_LENGTHof the tape is5plus a random number between 0 and 2.import {Copy} from "gym-js"; let base = 10; const env = new Copy(base); console.log(env.action_space.toString()); > DiscreteTuple: 2,2,10 console.log(env.observation_space.toString()); > Discrete: 11 let action = env.action_space.sample(); let [obs, rew, done, info] = env.step(action);