evanthebouncy

# a self play version of 20 quetsions with gpt playing against itself
from langchain.llms import OpenAI
# from langchain.chat_models import ChatOpenAI
llm = OpenAI(model_name="gpt-3.5-turbo-0301")

# an oracle
def get_oracle(concept):
    def give_answer(question):
        prompt = f"For a {concept}, \
        {question}. Answer with a single word [YES, NO]"

evanthebouncy

# get all possible next states
def get_next_state(tuple_state):
    ret = []
    # for all index of tuple_state, subtract any number of stones 1 through maximum possible
    for i in range(len(tuple_state)):
        for j in range(1, tuple_state[i]+1):
            # create new state
            new_state = list(tuple_state)
            new_state[i] -= j

evanthebouncy

from rectangle import is_inside, is_correct, inside, outside, W
import random
import string 

# for the purpose of showing this is a "language model", all programs here are 
# written as STRINGS, you need to call eval(prog) on them to get the actual program

def writer1():
    return ''.join(random.choice(string.printable) for i in range(9))

evanthebouncy

from rectangle import is_inside, is_correct, inside, outside, W
import random

def random_writer(spec):
    # ignores the spec
    T, D, L, R = random.randint(0,W), random.randint(0,W), random.randint(0,W), random.randint(0,W)
    return [T, D, L, R]

def better_writer(spec):
    # get the coordinates of spec that are inside

evanthebouncy

import random
# some globals
W = 6

inside = True
outside = False

def interpret(program, inputt):
    T, D, L, R = program
    i, j = inputt

evanthebouncy

import random
# some globals
W = 6

def exe(prog, x):
    T, D, L, R = prog
    i, j = x
    return i >= L and i <= R and j >= T and j <= D

# check if a spec is satisfied

evanthebouncy

# the tiling environment, a grid world where you place tetris pieces

import random

# the dimension of the grid
L = 6

# hardcoding a few pieces, it's fine
PIECES = [
    [[1,1,1,1]], # horizontal line

evanthebouncy

evanthebouncy

import numpy as np

np.random.seed(0)

def generate_pair():
    # make two random numbers x and y
    # make x first randomly from 1 to 10
    x = np.random.randint(1, 11)
    y = np.random.randint(1, 11)
    x_y = x + y

evanthebouncy

track = track4()
car = newCar(track)
startDisplay(track, car)

def control(sd, pastsd):
    # average these values to be more smooth
    leftt = sd[0] + pastsd[0];
    mid = sd[1] + pastsd[1];
    # this bit of assymetry is actually very clever, because 
    # "intuitively" assymetrical things has more information on it