torch.nn.Module.named_parameters()

• Iterator-based: Returns an iterator rather than a list, making it memory-efficient for large models

• Hierarchical naming: Parameter names reflect the module hierarchy using dot notation (e.g., layer1.weight)

• Recursive by default: Includes parameters from all submodules unless specified otherwise

import torch
import torch.nn as nn

class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.fc1 = nn.Linear(10, 5)
        self.fc2 = nn.Linear(5, 2)
    
    def forward(self, x):
        x = torch.relu(self.fc1(x))
        return self.fc2(x)

model = SimpleModel()

for name, param in model.named_parameters():
    print(f"{name}: {param.shape}")

# fc1.weight: torch.Size([5, 10])
# fc1.bias: torch.Size([5])
# fc2.weight: torch.Size([2, 5])
# fc2.bias: torch.Size([2])

Selective

Parameter Freezing

for name, param in model.named_parameters():
    if 'fc2' not in name:
        param.requires_grad = False
    print(f"{name}: requires_grad={param.requires_grad}")

Module — PyTorch 2.2 documentation

Modules can also contain other Modules, allowing to nest them in a tree structure. You can assign the submodules as regular attributes:

https://pytorch.org/docs/stable/generated/torch.nn.Module.html#torch.nn.Module.named_parameters