deep-learning

name	deep-learning
description	Build and train neural networks with PyTorch - MLPs, CNNs, and training best practices
version	1.4.0
sasmp_version	1.4.0
bonded_agent	04-deep-learning
bond_type	PRIMARY_BOND
parameters	{"required":[{"name":"model","type":"nn.Module","validation":"Valid PyTorch model"},{"name":"train_loader","type":"DataLoader","validation":"Non-empty DataLoader"}],"optional":[{"name":"epochs","type":"integer","default":10,"validation":"1 <= epochs <= 1000"},{"name":"lr","type":"float","default":0.001,"validation":"1e-6 <= lr <= 1"}]}
retry_logic	{"strategy":"exponential_backoff","max_attempts":3,"base_delay_ms":1000}
logging	{"level":"info","metrics":["train_loss","val_loss","val_accuracy","gpu_memory"]}

Deep Learning Skill

Build and train neural networks using PyTorch.

Quick Start

import torch
import torch.nn as nn
from torch.utils.data import DataLoader, TensorDataset

# Define model
class SimpleNN(nn.Module):
    def __init__(self, input_dim, hidden_dim, output_dim):
        super().__init__()
        self.layers = nn.Sequential(
            nn.Linear(input_dim, hidden_dim),
            nn.ReLU(),
            nn.Dropout(0.3),
            nn.Linear(hidden_dim, output_dim)
        )

    def forward(self, x):
        return self.layers(x)

# Train
model = SimpleNN(10, 64, 2)
optimizer = torch.optim.AdamW(model.parameters(), lr=1e-3)
criterion = nn.CrossEntropyLoss()

for epoch in range(10):
    model.train()
    for batch_x, batch_y in train_loader:
        optimizer.zero_grad()
        output = model(batch_x)
        loss = criterion(output, batch_y)
        loss.backward()
        optimizer.step()

Key Topics

1. Neural Network Architectures

Architecture	Use Case	Key Layers
MLP	Tabular data	Linear, ReLU, Dropout
CNN	Images	Conv2d, MaxPool2d, BatchNorm
RNN/LSTM	Sequences