在PyTorch中进行超参数搜索通常有两种常用的方法:
- 使用Grid Search:通过定义一个超参数的候选值列表,对所有可能的组合进行穷举搜索,选择表现最好的超参数组合。可以使用GridSearchCV类来实现这一过程。
from sklearn.model_selection import GridSearchCV
from sklearn.metrics import accuracy_score
from torch import nn, optim
from torch.utils.data import DataLoader
# Define your model
class MyModel(nn.Module):
def __init__(self, input_dim, hidden_dim, output_dim):
super(MyModel, self).__init__()
self.hidden = nn.Linear(input_dim, hidden_dim)
self.relu = nn.ReLU()
self.output = nn.Linear(hidden_dim, output_dim)
def forward(self, x):
x = self.hidden(x)
x = self.relu(x)
x = self.output(x)
return x
# Define your dataset and dataloader
# dataset = ...
# dataloader = DataLoader(dataset, batch_size=64, shuffle=True)
# Define parameter grid
param_grid = {
'hidden_dim': [64, 128, 256],
'learning_rate': [0.001, 0.01, 0.1]
}
# Create a GridSearchCV object
grid_search = GridSearchCV(MyModel, param_grid, scoring='accuracy', cv=3)
# Fit the model
grid_search.fit(dataloader)
# Print best parameters
print(grid_search.best_params_)
- 使用Random Search:与Grid Search不同,Random Search是随机地在指定的参数空间中采样,从而更有效地搜索超参数空间。可以使用RandomizedSearchCV类来实现这一过程。
from sklearn.model_selection import RandomizedSearchCV
from sklearn.metrics import accuracy_score
from torch import nn, optim
from torch.utils.data import DataLoader
# Define your model
# Define your model
class MyModel(nn.Module):
def __init__(self, input_dim, hidden_dim, output_dim):
super(MyModel, self).__init__()
self.hidden = nn.Linear(input_dim, hidden_dim)
self.relu = nn.ReLU()
self.output = nn.Linear(hidden_dim, output_dim)
def forward(self, x):
x = self.hidden(x)
x = self.relu(x)
x = self.output(x)
return x
# Define your dataset and dataloader
# dataset = ...
# dataloader = DataLoader(dataset, batch_size=64, shuffle=True)
# Define parameter grid
param_dist = {
'hidden_dim': [64, 128, 256],
'learning_rate': [0.001, 0.01, 0.1]
}
# Create a RandomizedSearchCV object
random_search = RandomizedSearchCV(MyModel, param_dist, n_iter=10, scoring='accuracy', cv=3)
# Fit the model
random_search.fit(dataloader)
# Print best parameters
print(random_search.best_params_)
无论选择哪种方法,超参数搜索是一个耗时的过程,需要谨慎选择超参数的范围和步长,以及合适的评估指标来评估模型性能。