forked from foersterrobert/AlphaZeroFromScratch
-
Notifications
You must be signed in to change notification settings - Fork 0
/
AlphaZeroP.py
139 lines (101 loc) · 5.26 KB
/
AlphaZeroP.py
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
127
128
129
130
131
132
133
134
135
136
137
138
139
import random
import numpy as np
import torch
import torch.nn.functional as F
from MCTSP import MCTSP
from tqdm import trange
from file_handler import write_to_file
import scipy.stats
from visualize import plot_hex_grid
class AlphaZeroP:
def __init__(self, model, optimizer, game, args):
self.model = model
self.optimizer = optimizer
self.game = game
self.args = args
self.mcts = MCTSP(game, args, model)
self.counter = 0
self.policy_loss_history = []
self.value_loss_history = []
self.entropy = 0
self.steps = 0
def selfPlay(self):
return_memory = []
player = 1
spGames = [SPG(self.game) for spg in range(self.args['num_parallel_games'])]
while len(spGames) > 0:
states = np.stack([spg.state for spg in spGames])
neutral_states = self.game.change_perspective(states, player)
self.mcts.search(neutral_states, spGames)
for i in range(len(spGames))[::-1]:
spg = spGames[i]
action_probs = np.zeros(self.game.action_size)
for child in spg.root.children:
action_probs[child.action_taken] = child.visit_count
action_probs /= np.sum(action_probs)
spg.memory.append((spg.root.state, action_probs, player))
temperature_action_probs = action_probs ** (1 / self.args['temperature'])
temperature_action_probs /= np.sum(temperature_action_probs)
action = np.random.choice(self.game.action_size, p=temperature_action_probs)
entropy = scipy.stats.entropy(action_probs)
self.entropy += entropy # Accumulate entropy over the game
spg.state = self.game.get_next_state(spg.state, action, player)
self.counter += 1
value, is_terminal = self.game.get_value_and_terminated(spg.state, action)
if is_terminal:
write_to_file(data1=player, data2=None, data3=None, filename="training1/winner.csv")
write_to_file(data1=self.entropy/self.counter, data2=None, data3=None, filename="training1/entropy.csv")
self.counter = 0
self.entropy = 0
for hist_neutral_state, hist_action_probs, hist_player in spg.memory:
hist_outcome = value if hist_player == player else self.game.get_opponent_value(value)
return_memory.append((
self.game.get_encoded_state(hist_neutral_state),
hist_action_probs,
hist_outcome
))
del spGames[i]
player = self.game.get_opponent(player)
return return_memory
def train(self, memory):
random.shuffle(memory)
for batchIdx in range(0, len(memory), self.args['batch_size']):
sample = memory[batchIdx:min(len(memory) - 1, batchIdx + self.args['batch_size'])]
state, policy_targets, value_targets = zip(*sample)
state, policy_targets, value_targets = np.array(state), np.array(policy_targets), np.array(
value_targets).reshape(-1, 1)
state = torch.tensor(state, dtype=torch.float32, device=self.model.device)
policy_targets = torch.tensor(policy_targets, dtype=torch.float32, device=self.model.device)
value_targets = torch.tensor(value_targets, dtype=torch.float32, device=self.model.device)
out_policy, out_value = self.model(state)
policy_loss = F.cross_entropy(out_policy, policy_targets)
value_loss = F.mse_loss(out_value, value_targets)
loss = policy_loss + value_loss
self.optimizer.zero_grad()
loss.backward()
self.optimizer.step()
self.steps += 1
write_to_file(data1=policy_loss.item(), data2=None, data3=None, filename="training1/policy_loss.csv")
write_to_file(data1=value_loss.item(), data2=None, data3=None, filename="training1/value_loss.csv")
write_to_file(data1=loss.item(), data2=None, data3=None, filename="training1/total_loss.csv")
write_to_file(data1=self.steps, data2=None, data3=None, filename="training1/steps.csv")
#self.policy_loss_history.append(policy_loss.item())
#self.value_loss_history.append(value_loss.item())
def learn(self):
for iteration in range(self.args['num_iterations']):
memory = []
self.model.eval()
for selfPlay_iteration in trange(self.args['num_selfPlay_iterations'] // self.args['num_parallel_games']):
memory += self.selfPlay()
self.model.train()
for epoch in trange(self.args['num_epochs']):
self.train(memory)
self.mcts.save_depth_counts_to_csv("depth_counts.csv")
torch.save(self.model.state_dict(), f"model_{iteration}_{self.game}.pt")
torch.save(self.optimizer.state_dict(), f"optimizer_{iteration}_{self.game}.pt")
class SPG:
def __init__(self, game):
self.state = game.get_initial_state()
self.memory = []
self.root = None
self.node = None