Multi-head attention

As explained in Deep Learning with PyTorch, Second Edition

An attention mechanism that runs several attention heads in parallel, each learning different projections and relationships.

Multi-head attention runs several attention mechanisms in parallel instead of relying on one. Each head learns its own query, key, and value projections, so different heads can focus on different relationships in the same sequence. section 9.6.1

For example, one head might learn syntactic relationships while another captures semantic associations. The head outputs are concatenated and passed through a linear layer to form the final representation. section 9.6.1

A transformer block uses multi-head attention alongside:

• Residual connections, which let gradients flow directly through deeper networks. section 9.6.1
• Positional embeddings, which add token-order information to otherwise parallel processing. section 9.6.1
• Layer normalization and a feed-forward MLP, which complete the surrounding transformer block. section 9.6.1

A PyTorch transformer block

A block can set num_heads=4, require n_embd to be divisible by that number, and create separate linear query, key, and value projections. A larger Transformer then adds character and positional embeddings, stacks these blocks, and projects the result to vocabulary logits.

Figure 9.13 Diagram of the Transformer architecture from the original paper “Attention Is All You Need.” The encoder is on the left, and the decoder is on the right. The encoder output is fed into the decoder as the key and value for the multi-head attention mechanism.

Use the architecture diagram to trace how encoder outputs become the keys and values for decoder cross-attention, supporting sequence generation such as translation. section 9.7.2

How Deep Learning with PyTorch, Second Edition covers it

How deep it goes

Explains learned projections, head specialization, concatenation, and surrounding components, while stopping short of detailed attention-score mathematics.

Examples

Demonstrates a PyTorch TransformerBlock and a stacked character-level Transformer that generates sample names.

What it assumes

Assumes basic attention, embeddings, neural-network modules, and transformer-block components such as residual connections and normalization.

What you can do after

Builds it from scratch in PyTorch as part of a transformer block and connects it to sequence generation.

How it connects

Connects attention heads to query, key, and value projections, residual connections, positional embeddings, layer normalization, and encoder-decoder attention.

liveBookmanning.com

Read Deep Learning with PyTorch, Second Edition on liveBook Get the book at manning.com

Also explained in

• 

  liveBookmanning.com
  Natural Language Processing in Action, Second EditionMulti-head attention computes multiple attention representations in parallel and combines them, potentially using distinct query and source sequences.
• 

  liveBookmanning.com
  Build a Text-to-Image Generator (from Scratch)An attention mechanism that computes attention through multiple separate query, key, and value subspaces to capture different relationships simultaneously.
• 

  liveBookmanning.com
  Transformers in ActionMultihead attention is an attention mechanism that computes multiple attention patterns in parallel so a model can represent different relationships among sequence positions.
• 

  liveBookmanning.com
  Learn Generative AI with PyTorchA self-attention variant that splits representations into multiple attention heads to capture different relationships in parallel.
• 

  liveBookmanning.com
  Inside Deep Learning: Math, Algorithms, ModelsAn attention mechanism that uses multiple attention heads to attend to different contexts or aspects of the input in parallel.
• 

  liveBookmanning.com
  Deep Learning with Python, Third EditionAn attention variant that runs several attention computations in parallel with different parameters to learn different relationships.
• 

  liveBookmanning.com
  Graph Neural Networks in ActionAn attention technique that computes multiple attention operations in parallel and combines their outputs.
• 

  liveBookmanning.com
  Build a Large Language Model (From Scratch)An attention technique that computes several independent attention heads in parallel and combines their outputs to capture different relationships in the same sequence.
• 

  liveBookmanning.com
  Deep Learning with R, Third EditionMultiHeadAttention is a Keras layer that computes attention through multiple parallel attention heads and combines their outputs.
• 

  liveBookmanning.com
  Time Series Forecasting Using Foundation ModelsAn attention mechanism with multiple parallel attention heads that combines information from different representation subspaces.
• 

  liveBookmanning.com
  Machine Learning Algorithms in DepthMultihead attention runs multiple attention mechanisms in parallel and combines their output representations.
• 

  liveBookmanning.com
  Optimization Algorithms: AI techniques for design, planning, and control problemsMulti-head self-attention computes several attention operations in parallel so a sequence representation can capture different relationships among its elements.
• 

  liveBookmanning.com
  The AI Pocket BookA transformer attention mechanism that splits representations into multiple heads to attend to different relationships in parallel.

Compare how 14 books explain multi-head attention →

Related concepts

• graph attention networkused by · A graph attention network is a graph neural network that uses attention weights to aggregate information from neighboring nodes, allowing their contributions to differ.
• grouped-query attentionequivalent to · Grouped-query attention partitions query heads into groups whose members share key and value projections, reducing attention computation and memory use.
• attentionkind of · Attention is a neural-network mechanism that computes input-dependent weights to combine information according to its relevance to a computation.
• attention headhas part · An attention head is an independently parameterized attention computation that evaluates relationships among input elements and combines information from them into a contextualized representation.
• transformer architectureused by · A transformer architecture is a neural-network architecture that processes sequences or other structured inputs using attention mechanisms and related transformations.
• encoderpart of · An encoder is a component that transforms input data into an encoded representation for subsequent processing, reconstruction, comparison, or transmission.

This explainer was generated by Manning's knowledge graph from the text of Deep Learning with PyTorch, Second Edition. The section links point to the parts of the book it draws on.