/
solver.py
165 lines (139 loc) · 5.49 KB
/
solver.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
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
from functools import partial
import glob
import torch
from torch.utils.data import DataLoader
import torch.nn as nn
import torch.optim as optim
import wandb
from data import CNFDataset, train_eval_test_split, collate_fn
PREDICT_SOLUTION = False # predict assignments instead of satisfiability
class TransformerBinaryClassifier(nn.Module):
def __init__(self, input_dim, num_layers, num_heads):
super(TransformerBinaryClassifier, self).__init__()
self.encoder_layer = nn.TransformerEncoderLayer(
d_model=input_dim, nhead=num_heads
)
self.transformer_encoder = nn.TransformerEncoder(
self.encoder_layer, num_layers=num_layers
)
self.fc = nn.Linear(input_dim, 1)
self.predict_solution = PREDICT_SOLUTION
def forward(self, src):
output = self.transformer_encoder(src)
output = torch.sum(output, dim=0)
if not self.predict_solution:
output = self.fc(output)
return torch.sigmoid(output)
def train_model(train_dataset, eval_dataset):
with wandb.init():
hparams = wandb.config
model = TransformerBinaryClassifier(
input_dim=hparams["input_dim"],
num_layers=hparams["num_layers"],
num_heads=hparams["num_heads"],
)
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model.to(device)
train_loader = DataLoader(
train_dataset,
batch_size=hparams["batch_size"],
shuffle=True,
collate_fn=collate_fn,
)
eval_loader = DataLoader(
eval_dataset,
batch_size=hparams["batch_size"],
shuffle=False,
collate_fn=collate_fn,
)
if PREDICT_SOLUTION:
criterion = nn.MSELoss() # or nn.BCEWithLogitsLoss() if treating as element-wise classification
else:
criterion = nn.BCELoss()
optimizer = optim.Adam(model.parameters(), lr=hparams["learning_rate"])
scheduler = torch.optim.lr_scheduler.StepLR(
optimizer, step_size=hparams["step_size"], gamma=hparams["gamma"]
)
for epoch in range(hparams["num_epochs"]):
model.train()
total_loss = 0
for seq, labels in train_loader:
seq, labels = seq.to(device), labels.to(device)
outputs = model(seq).squeeze(dim=1)
loss = criterion(outputs, labels)
optimizer.zero_grad()
loss.backward()
optimizer.step()
total_loss += loss.item()
scheduler.step()
avg_loss = total_loss / len(train_loader)
wandb.log(
{
"epoch": epoch,
"train_loss": avg_loss,
"learning_rate": scheduler.get_last_lr()[0],
}
)
model.eval()
with torch.no_grad():
eval_loss = 0
correct_predictions = 0
total_predictions = 0
for seq, labels in eval_loader:
seq, labels = seq.to(device), labels.to(device)
outputs = model(seq).squeeze(dim=1)
loss = criterion(outputs, labels)
eval_loss += loss.item()
# Calculate accuracy
predicted_labels = outputs > 0.5
correct_predictions += (
(predicted_labels == labels.bool()).sum().item()
)
total_predictions += labels.size(0)
avg_eval_loss = eval_loss / len(eval_loader)
accuracy = correct_predictions / total_predictions
wandb.log(
{"epoch": epoch, "eval_loss": avg_eval_loss, "accuracy": accuracy}
)
print(
f"Epoch [{epoch+1}/{hparams['num_epochs']}], Train Loss: {avg_loss:.4f}, Eval Loss: {avg_eval_loss:.4f}, Accuracy: {accuracy:.4f}"
)
if __name__ == "__main__":
cnf_files = glob.glob("data/*.cnf")
cnf_dataset = CNFDataset(cnf_files)
train_dataset, eval_dataset, test_dataset = train_eval_test_split(cnf_dataset)
# hparams = {
# "batch_size": 32,
# "num_epochs": 20,
# "learning_rate": 0.001,
# "num_layers": 3,
# "num_heads": 1,
# "input_dim": cnf_dataset.input_dim,
# "max_seq_len": cnf_dataset.max_seq_len,
# }
# wandb.init(
# project="SATScale",
# entity="rspandey",
# config=hparams,
# )
# train_model(train_dataset, eval_dataset)
sweep_config = {
"method": "bayes",
"metric": {"name": "eval_loss", "goal": "minimize"},
"parameters": {
"learning_rate": {"min": 0.0001, "max": 0.01},
"num_layers": {"values": [1, 2, 4, 8, 16]},
"num_heads": {"values": [1, 2, 4, 8]},
"batch_size": {"values": [16, 32, 64]},
"num_epochs": {"values": [5, 20, 50]},
"step_size": {"values": [5, 10, 20]},
"gamma": {"values": [0.1, 0.5, 0.9]},
"input_dim": {"values": [cnf_dataset.input_dim]},
"max_seq_len": {"values": [cnf_dataset.max_seq_len]},
},
}
sweep_id = wandb.sweep(sweep_config, project="SATScale", entity="rspandey")
train_model_partial = partial(
train_model, train_dataset=train_dataset, eval_dataset=eval_dataset
)
wandb.agent(sweep_id, train_model_partial)