AlphaGateau/match.py at master · Akulen/AlphaGateau · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
import argparse
import os
import json

import numpy as np
import jax
import flax.linen as nn
import pgx
import pgx.chess as pgc
import rich.progress as rp
from sklearn.linear_model import LinearRegression
from rich.pretty import pprint

from models import load_model
import mcts
from utils import to_pgn
from ranking import compute_elo


devices = jax.local_devices()
num_devices = len(devices)

games_dir = f"./tournaments/ranking matches"
os.makedirs(games_dir, exist_ok=True)

def main():
    parser = argparse.ArgumentParser(
        prog='MatchMaker',
        description='Run a match between two models',
    )
    parser.add_argument('model1')
    parser.add_argument('iteration1', type=int)
    parser.add_argument('model2')
    parser.add_argument('iteration2', type=int)
    parser.add_argument('-n', '--n-games', type=int, default=64)
    parser.add_argument('-m', '--max-plies', type=int, default=256)
    parser.add_argument('-s', '--n-sims', type=int, default=128)
    parser.add_argument('-r', '--seed', type=int, default=42)

    args = parser.parse_args()

    assert(os.path.isfile(f"./models/{args.model1}/{args.iteration1:06}.ckpt"))
    assert(os.path.isfile(f"./models/{args.model2}/{args.iteration2:06}.ckpt"))

    env_id = "chess"
    env = pgx.make(env_id)
    model1, model_params1 = load_model(
        env,
        f"./models/{args.model1}/{args.iteration1:06}.ckpt",
        f"{args.model1}-{args.iteration1:03}"
    )
    model2, model_params2 = load_model(
        env,
        f"./models/{args.model2}/{args.iteration2:06}.ckpt",
        f"{args.model2}-{args.iteration2:03}"
    )

    with open("rankings.json", "r") as file:
        data = json.load(file)
    with open("rankings.backup.json", "w") as file:
        json.dump(data, file)

    R, games = mcts.full_pit(
        env,
        model1,
        jax.device_put_replicated(model_params1, devices),
        model2,
        jax.device_put_replicated(model_params2, devices),
        jax.random.PRNGKey(args.seed),
        n_games=args.n_games,
        max_plies=args.max_plies,
        n_sim=args.n_sims,
        n_devices=num_devices
    )
    wins, draws, losses = map(
        lambda r: ((R == r).sum()).item(),
        [1, 0, -1]
    )
    model1_name = f'{args.model1}/{args.iteration1:03}'
    model2_name = f'{args.model2}/{args.iteration2:03}'
    if model1_name not in data['results']:
        data['results'][model1_name] = {}
    if model2_name not in data['results'][model1_name]:
        data['results'][model1_name][model2_name] = [0, 0, 0]
    if model2_name not in data['results']:
        data['results'][model2_name] = {}
    if model1_name not in data['results'][model2_name]:
        data['results'][model2_name][model1_name] = [0, 0, 0]
    data['results'][model1_name][model2_name][0] += wins
    data['results'][model1_name][model2_name][1] += draws
    data['results'][model1_name][model2_name][2] += losses
    data['results'][model2_name][model1_name][0] += losses
    data['results'][model2_name][model1_name][1] += draws
    data['results'][model2_name][model1_name][2] += wins

    print(f'{model1_name}[{data["elo"][model1_name]:4}]'
          f' vs '
          f'[{data["elo"][model2_name]:4}]{model2_name}: '
          f'{wins:2}/{draws:2}/{losses:2}')

    data['elo'] = compute_elo(data['results'])

    count = [128] * 3
    with open(os.path.join(
        games_dir,
        f"{model1.id} vs {model2.id}.pgn"
    ), "w") as f:
        for r, g in zip(R, games):
            r_i = int(np.round(r))
            if count[r_i+1] > 0:
                count[r_i+1] -= 1
                print(to_pgn(
                    g,
                    round=f"Ranking Match {model1_name} vs {model2_name}",
                    player0=model1.id,
                    player1=model2.id,
                    result=r_i,
                    pgc=pgc
                ), file=f)

    pprint(data['elo'])
    with open("rankings.json", "w") as file:
        json.dump(data, file)

if __name__ == "__main__":
    main()