Transformer decoder in PyTorch

Introduction

The Transformer decoder helps a model understand and generate sequences, like sentences, by looking at what it has generated so far and what it should pay attention to.

When building a language translator that turns one language into another.

When creating a chatbot that replies to messages step-by-step.

When generating text, like writing stories or summaries.

When predicting the next word in a sentence based on previous words.

Syntax

PyTorch

torch.nn.TransformerDecoder(decoder_layer, num_layers, norm=None)

decoder_layer is a single layer that defines how the decoder works.

num_layers is how many decoder layers you stack to make the full decoder.

Examples

This creates a decoder with 6 layers, each layer has 512 features and 8 attention heads.

PyTorch

decoder_layer = torch.nn.TransformerDecoderLayer(d_model=512, nhead=8)
decoder = torch.nn.TransformerDecoder(decoder_layer, num_layers=6)

Here, tgt is the target sequence input, and memory is the encoded source sequence from the encoder.

PyTorch

output = decoder(tgt, memory)

Sample Model

This code builds a Transformer decoder with 2 layers. It creates random input sequences and passes them through the decoder. The output shape shows the sequence length, batch size, and feature size. The first output vector is printed rounded to 4 decimals.

PyTorch

import torch
from torch import nn

# Set seed for reproducibility
torch.manual_seed(0)

# Define parameters
d_model = 32
nhead = 4
num_layers = 2
seq_len = 5
batch_size = 1

# Create a single decoder layer
decoder_layer = nn.TransformerDecoderLayer(d_model=d_model, nhead=nhead)
# Stack decoder layers
decoder = nn.TransformerDecoder(decoder_layer, num_layers=num_layers)

# Create dummy target sequence (tgt) and memory from encoder
# Shape: (sequence length, batch size, feature size)
tgt = torch.rand(seq_len, batch_size, d_model)
memory = torch.rand(seq_len, batch_size, d_model)

# Pass through decoder
output = decoder(tgt, memory)

# Print output shape and first vector
print(f"Output shape: {output.shape}")
print(f"First output vector:\n{output[0,0,:].round(4)}")

Important Notes

The decoder expects inputs shaped as (sequence length, batch size, feature size).

The memory input comes from the encoder output and helps the decoder focus on the source sequence.

Masking is often used in real tasks to prevent the decoder from seeing future tokens, but is omitted here for simplicity.

Summary

The Transformer decoder generates output sequences by looking at previous outputs and encoder information.

It is built by stacking decoder layers, each with attention mechanisms.

Use it when you want to create models that generate or predict sequences step-by-step.

Practice

1. What is the main role of the Transformer decoder in a sequence-to-sequence model?

easy

A. To reduce the dimensionality of input data

B. To encode input sequences into fixed-size vectors

C. To perform image classification tasks

D. To generate output tokens step-by-step using previous outputs and encoder info

Solution

1. Step 1: Understand the Transformer decoder purpose

   The Transformer decoder generates output sequences by attending to previous outputs and encoder outputs.

2. Step 2: Compare options with decoder role

   Only To generate output tokens step-by-step using previous outputs and encoder info describes generating output tokens step-by-step using previous outputs and encoder info, which matches the decoder's role.

3. Final Answer:

   To generate output tokens step-by-step using previous outputs and encoder info -> Option D

4. Quick Check:

   Decoder generates outputs stepwise = D [OK]

Hint: Decoder creates outputs using past outputs and encoder data [OK]

Common Mistakes:

• Confusing encoder and decoder roles
• Thinking decoder encodes inputs
• Assuming decoder reduces data size

2. Which of the following is the correct way to create a Transformer decoder layer in PyTorch?

easy

A. torch.nn.TransformerDecoderLayer(d_model=512, nhead=8)

B. torch.nn.TransformerEncoderLayer(d_model=512, nhead=8)

C. torch.nn.TransformerDecoder(d_model=512, nhead=8)

D. torch.nn.TransformerLayer(d_model=512, nhead=8)

Solution

1. Step 1: Identify the correct PyTorch class for a decoder layer

   PyTorch provides torch.nn.TransformerDecoderLayer to create a single decoder layer.

2. Step 2: Check options for correct class and parameters

   torch.nn.TransformerDecoderLayer(d_model=512, nhead=8) uses torch.nn.TransformerDecoderLayer with correct parameters. Others are either encoder layers or invalid classes.

3. Final Answer:

   torch.nn.TransformerDecoderLayer(d_model=512, nhead=8) -> Option A

4. Quick Check:

   Decoder layer class is TransformerDecoderLayer = A [OK]

Hint: Decoder layer class ends with 'DecoderLayer' in PyTorch [OK]

Common Mistakes:

• Using EncoderLayer instead of DecoderLayer
• Trying to instantiate TransformerDecoder directly
• Using non-existent class names

3. Given the code below, what is the shape of the output tensor?

import torch
from torch.nn import TransformerDecoder, TransformerDecoderLayer

layer = TransformerDecoderLayer(d_model=256, nhead=8)
decoder = TransformerDecoder(layer, num_layers=3)

memory = torch.rand(10, 32, 256)  # (source_seq_len, batch_size, d_model)
tgt = torch.rand(20, 32, 256)     # (target_seq_len, batch_size, d_model)

output = decoder(tgt, memory)

medium

A. (32, 20, 256)

B. (10, 32, 256)

C. (20, 32, 256)

D. (20, 256, 32)

Solution

1. Step 1: Understand input and output shapes for TransformerDecoder

   The decoder input shape is (target_seq_len, batch_size, d_model). The output shape matches the target input shape.

2. Step 2: Check given shapes and output

   Target input 'tgt' has shape (20, 32, 256), so output will have the same shape.

3. Final Answer:

   (20, 32, 256) -> Option C

4. Quick Check:

   Output shape matches target input shape = C [OK]

Hint: Output shape matches target input shape in decoder [OK]

Common Mistakes:

• Confusing source and target sequence lengths
• Mixing batch size and feature dimensions
• Assuming output shape changes after decoding

4. What is wrong with this PyTorch Transformer decoder usage?

import torch
from torch.nn import TransformerDecoder, TransformerDecoderLayer

layer = TransformerDecoderLayer(d_model=128, nhead=4)
decoder = TransformerDecoder(layer, num_layers=2)

memory = torch.rand(15, 16, 128)
tgt = torch.rand(15, 16, 64)  # Incorrect feature size

output = decoder(tgt, memory)

medium

A. The memory tensor has wrong batch size

B. The target tensor 'tgt' has wrong feature dimension

C. The number of decoder layers is too low

D. The nhead parameter should be equal to batch size

Solution

1. Step 1: Check feature dimensions of inputs

   The decoder expects target and memory tensors to have the same feature dimension (d_model=128).

2. Step 2: Identify mismatch in target tensor

   Target tensor 'tgt' has feature size 64, which does not match d_model=128, causing an error.

3. Final Answer:

   The target tensor 'tgt' has wrong feature dimension -> Option B

4. Quick Check:

   Feature size mismatch in target tensor = A [OK]

Hint: Check if target and memory feature sizes match d_model [OK]

Common Mistakes:

• Ignoring feature dimension mismatch
• Assuming batch size must match nhead
• Thinking number of layers causes this error

5. You want to build a Transformer decoder that generates text step-by-step using previous outputs and encoder memory. Which of the following is the best approach to implement the decoder's forward pass in PyTorch?

hard

A. Use a TransformerDecoder with masked target input to prevent attending to future tokens

B. Use a TransformerDecoder without masking so it sees all target tokens

C. Use only a TransformerEncoder for generation

D. Use a TransformerDecoder but feed the entire output sequence at once without masking

Solution

1. Step 1: Understand the need for masking in decoding

   During generation, the decoder must not look at future tokens to avoid cheating, so masking is required.

2. Step 2: Identify correct usage in PyTorch

   Using TransformerDecoder with a mask on target input prevents attending to future tokens, enabling step-by-step generation.

3. Final Answer:

   Use a TransformerDecoder with masked target input to prevent attending to future tokens -> Option A

4. Quick Check:

   Mask future tokens in decoder input = B [OK]

Hint: Mask future tokens in decoder input for stepwise generation [OK]

Common Mistakes:

• Not using masks and allowing future token attention
• Using encoder only for generation
• Feeding full output sequence without masking