Multi-head attention in PyTorch - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to create a multi-head attention layer with 8 heads.

PyTorch

import torch.nn as nn

multihead_attn = nn.MultiheadAttention(embed_dim=64, num_heads=[1])

Drag options to blanks, or click blank then click option'

A8

B4

C16

D32

Attempts:

3 left

💡 Hint

Common Mistakes

Choosing a number of heads that does not divide the embedding dimension evenly.

Using too few or too many heads without considering model size.

The number of heads in multi-head attention is set by the num_heads parameter. Here, 8 heads is the correct choice for this example.

2fill in blank

medium

Complete the code to apply multi-head attention on query, key, and value tensors.

PyTorch

output, attn_weights = multihead_attn([1], key, value)

Drag options to blanks, or click blank then click option'

Aquery

Bkey

Cvalue

Doutput

Attempts:

3 left

💡 Hint

Common Mistakes

Passing key or value as the first argument instead of query.

Confusing the order of inputs.

The first argument to multihead_attn is the query tensor, which is used to compute attention scores.

3fill in blank

hard

Fix the error in the code to correctly reshape the output of multi-head attention.

PyTorch

batch_size, seq_len, embed_dim = x.size()
output = output.transpose(0, 1).reshape([1], seq_len, embed_dim)

Drag options to blanks, or click blank then click option'

Aseq_len

Boutput.size(0)

Cembed_dim

Dbatch_size

Attempts:

3 left

💡 Hint

Common Mistakes

Using seq_len instead of batch_size in reshape.

Not transposing before reshaping.

The output tensor from multihead_attn has shape (seq_len, batch_size, embed_dim). Transposing dims 0 and 1 swaps seq_len and batch_size, so reshaping should use batch_size as the first dimension.

4fill in blank

hard

Fill both blanks to create a mask that prevents attention to future tokens.

PyTorch

import torch
seq_len = 5
mask = torch.triu(torch.ones(seq_len, seq_len), diagonal=[1]) == [2]

Drag options to blanks, or click blank then click option'

A1

B0

CTrue

DFalse

Attempts:

3 left

💡 Hint

Common Mistakes

Using diagonal=0 which masks the main diagonal too.

Comparing to True instead of 0.

To mask future tokens, we use torch.triu with diagonal=1 to get upper triangle above the main diagonal. Comparing to 0 creates a boolean mask where True means attendable positions.

5fill in blank

hard

Fill all three blanks to compute scaled dot-product attention manually.

PyTorch

import torch
import torch.nn.functional as F
import math

d_k = query.size(-1)
scores = torch.matmul(query, key.transpose(-2, -1)) / math.sqrt([1])
attn = F.softmax(scores, dim=[2])
output = torch.matmul(attn, [3])

Drag options to blanks, or click blank then click option'

Ad_k

B-1

Cvalue

Dquery

Attempts:

3 left

💡 Hint

Common Mistakes

Using query instead of value in the last multiplication.

Applying softmax over wrong dimension.

Forgetting to scale scores.

Scaled dot-product attention divides scores by sqrt(d_k), applies softmax over last dimension (-1), then multiplies by value to get output.

Practice

1. What is the main purpose of multi-head attention in PyTorch's MultiheadAttention layer?

easy

A. To reduce the size of the input data

B. To perform simple addition of input vectors

C. To increase the number of layers in the model

D. To allow the model to focus on different parts of the input simultaneously

Solution

1. Step 1: Understand the concept of multi-head attention

   Multi-head attention splits the input into multiple parts (heads) to focus on different information simultaneously.

2. Step 2: Relate to PyTorch's MultiheadAttention layer

   This layer implements this idea to improve model understanding by attending to various input features at once.

3. Final Answer:

   To allow the model to focus on different parts of the input simultaneously -> Option D

4. Quick Check:

   Multi-head attention = focus on multiple input parts [OK]

Hint: Multi-head means multiple focus points at once [OK]

Common Mistakes:

• Thinking it reduces input size
• Confusing it with adding layers
• Assuming it just adds vectors

2. Which of the following is the correct way to create a multi-head attention layer with 8 heads and embedding dimension 64 in PyTorch?

easy

A. torch.nn.MultiheadAttention(8, 64)

B. torch.nn.MultiheadAttention(embed_dim=64, num_heads=8)

C. torch.nn.MultiheadAttention(num_heads=64, embed_dim=8)

D. torch.nn.MultiheadAttention(64)

Solution

1. Step 1: Check PyTorch MultiheadAttention parameter order

   The constructor requires embed_dim first, then num_heads as keyword arguments.

2. Step 2: Verify correct syntax

   torch.nn.MultiheadAttention(embed_dim=64, num_heads=8) correctly uses embed_dim=64 and num_heads=8. torch.nn.MultiheadAttention(num_heads=64, embed_dim=8) swaps parameters incorrectly, A and D use positional arguments wrongly.

3. Final Answer:

   torch.nn.MultiheadAttention(embed_dim=64, num_heads=8) -> Option B

4. Quick Check:

   embed_dim first, num_heads second [OK]

Hint: Remember: embed_dim first, num_heads second in constructor [OK]

Common Mistakes:

• Swapping embed_dim and num_heads
• Using positional args incorrectly
• Missing one of the required parameters

3. Given the following PyTorch code, what is the shape of the output tensor attn_output?

import torch
mha = torch.nn.MultiheadAttention(embed_dim=16, num_heads=4)
query = torch.rand(5, 3, 16)  # (sequence_length, batch_size, embed_dim)
key = torch.rand(5, 3, 16)
value = torch.rand(5, 3, 16)
attn_output, attn_weights = mha(query, key, value)

medium

A. (5, 3, 16)

B. (3, 5, 16)

C. (5, 16, 3)

D. (3, 16, 5)

Solution

1. Step 1: Understand input shapes for MultiheadAttention

   Inputs are (sequence_length, batch_size, embed_dim). Here, query shape is (5, 3, 16).

2. Step 2: Check output shape of attn_output

   Output shape matches query shape: (5, 3, 16).

3. Final Answer:

   (5, 3, 16) -> Option A

4. Quick Check:

   Output shape = query shape = (5, 3, 16) [OK]

Hint: Output shape matches query input shape [OK]

Common Mistakes:

• Mixing batch and sequence dimensions
• Confusing embed_dim position
• Assuming output shape changes with heads

4. Identify the error in this PyTorch multi-head attention code snippet:

import torch
mha = torch.nn.MultiheadAttention(embed_dim=32, num_heads=4)
query = torch.rand(10, 2, 32)
key = torch.rand(10, 2, 16)
value = torch.rand(10, 2, 32)
output, weights = mha(query, key, value)

medium

A. The embed_dim of key does not match embed_dim of query and value

B. The batch size is incorrect

C. The sequence length is too large

D. The number of heads is too high

Solution

1. Step 1: Check embed_dim consistency

   MultiheadAttention requires query, key, and value to have the same embedding dimension. Here, key has embed_dim 16, while query and value have 32.

2. Step 2: Identify the error cause

   This mismatch causes a runtime error because the attention mechanism cannot align different embedding sizes.

3. Final Answer:

   The embed_dim of key does not match embed_dim of query and value -> Option A

4. Quick Check:

   Embed dims must match for query, key, value [OK]

Hint: Check all embed_dims match before calling attention [OK]

Common Mistakes:

• Ignoring embed_dim mismatch
• Assuming batch size causes error
• Thinking sequence length is the problem

5. You want to apply multi-head attention to a batch of sentences with varying lengths. Which PyTorch feature helps you handle padding tokens correctly during attention computation?

hard

A. Increasing the number of heads to cover padding

B. Setting batch_first=True in MultiheadAttention

C. Using the key_padding_mask argument in MultiheadAttention

D. Manually removing padding tokens before attention

Solution

1. Step 1: Understand padding in variable-length sequences

   Sentences have different lengths, so padding tokens are added to make batch tensors uniform.

2. Step 2: Use key_padding_mask to ignore padding

   PyTorch's MultiheadAttention accepts key_padding_mask to mask out padded positions during attention, preventing them from affecting results.

3. Final Answer:

   Using the key_padding_mask argument in MultiheadAttention -> Option C

4. Quick Check:

   Mask padding tokens with key_padding_mask [OK]

Hint: Use key_padding_mask to ignore padded tokens [OK]

Common Mistakes:

• Removing padding manually (inefficient)
• Confusing batch_first with padding handling
• Increasing heads does not fix padding