Attention mechanism in depth in NLP

Introduction

Attention helps a model focus on important parts of the input when making decisions. It improves understanding by weighing useful information more.

Translating a sentence from one language to another, where some words depend on others far away.

Summarizing a long article by focusing on key sentences.

Answering questions by looking at relevant parts of a text.

Generating captions for images by focusing on important image regions.

Speech recognition where certain sounds matter more for words.

Syntax

NLP

Attention(Q, K, V) = softmax((Q * K^T) / sqrt(d_k)) * V

Q = Query, K = Key, V = Value are matrices derived from input data.

softmax normalizes scores to probabilities, highlighting important parts.

Examples

Simple example showing how query matches keys and weights values accordingly.

NLP

Q = [[1, 0]]
K = [[1, 0], [0, 1]]
V = [[1, 2], [3, 4]]

scores = Q @ K.T / (2 ** 0.5)
weights = softmax(scores)
output = weights @ V

PyTorch code to compute attention output with softmax weights.

NLP

import torch

Q = torch.tensor([[1., 0.]])
K = torch.tensor([[1., 0.], [0., 1.]])
V = torch.tensor([[1., 2.], [3., 4.]])

scores = torch.matmul(Q, K.T) / (2 ** 0.5)
weights = torch.nn.functional.softmax(scores, dim=-1)
output = torch.matmul(weights, V)
print(output)

Sample Model

This program shows how attention scores are computed, normalized, and used to get a weighted sum of values. It uses simple tensors to demonstrate the core idea.

NLP

import torch
import torch.nn.functional as F

# Define Query, Key, Value tensors
Q = torch.tensor([[1., 0., 1.]])  # Query vector
K = torch.tensor([[1., 0., 1.], [0., 1., 0.], [1., 1., 0.]])  # Key vectors
V = torch.tensor([[1., 2.], [3., 4.], [5., 6.]])  # Value vectors

d_k = Q.size(-1)  # dimension of key

# Calculate scaled dot-product attention
scores = torch.matmul(Q, K.T) / (d_k ** 0.5)  # shape: (1, 3)
weights = F.softmax(scores, dim=-1)  # shape: (1, 3)
output = torch.matmul(weights, V)  # shape: (1, 2)

print(f"Scores: {scores}")
print(f"Weights (attention probabilities): {weights}")
print(f"Output (weighted sum of values): {output}")

Important Notes

Attention scores measure how well each key matches the query.

Scaling by sqrt(d_k) prevents large dot products that hurt learning.

Softmax turns scores into probabilities that sum to 1.

Summary

Attention helps models focus on important parts of input data.

It uses queries, keys, and values to compute weighted sums.

Softmax normalizes scores to highlight relevant information.

Practice

1. What is the main purpose of the attention mechanism in NLP models?

easy

A. To increase the size of the input data

B. To reduce the number of layers in the model

C. To help the model focus on important parts of the input data

D. To randomly shuffle the input tokens

Solution

1. Step 1: Understand attention's role

   Attention helps models decide which parts of the input are most important for the task.

2. Step 2: Compare options

   Only To help the model focus on important parts of the input data correctly describes this focus mechanism; others describe unrelated actions.

3. Final Answer:

   To help the model focus on important parts of the input data -> Option C

4. Quick Check:

   Attention = Focus on important input [OK]

Hint: Remember: attention means focusing on key input parts [OK]

Common Mistakes:

• Thinking attention changes input size
• Confusing attention with model depth
• Assuming attention shuffles data

2. Which of the following correctly represents the formula for attention weights using queries (Q), keys (K), and softmax?

easy

A. softmax(Q x K^T)

B. Q + K

C. softmax(Q - K)

D. Q x K

Solution

1. Step 1: Recall attention weight calculation

   Attention weights are computed by multiplying queries with keys transposed, then applying softmax.

2. Step 2: Evaluate options

   Only softmax(Q x K^T) matches the correct formula softmax(Q x K^T). Others are incorrect operations.

3. Final Answer:

   softmax(Q x K^T) -> Option A

4. Quick Check:

   Attention weights = softmax(Q x K^T) [OK]

Hint: Attention weights = softmax of query-key dot product [OK]

Common Mistakes:

• Using addition instead of multiplication
• Forgetting to transpose keys
• Skipping softmax normalization

3. Given queries Q = [[1, 0]], keys K = [[1, 0], [-10, 1]], and values V = [[10, 20], [30, 40]], what is the output of the attention mechanism (using dot product and softmax)?

medium

A. [[10, 20]]

B. [[20, 30]]

C. [[20, 40]]

D. [[30, 40]]

Solution

1. Step 1: Calculate dot products Q x K^T

   Q = [1,0], K = [[1,0],[-10,1]]; dot products: [1*1+0*0=1, 1*(-10)+0*1=-10]

2. Step 2: Apply softmax to scores

   softmax([1,-10]) ≈ [1, 0] (e^{-10} negligible)

3. Step 3: Compute weighted sum of values

   Output ≈ 1*[10,20] + 0*[30,40] = [[10, 20]]

4. Step 4: Match option

   [[10, 20]] matches exactly.

5. Final Answer:

   [[10, 20]] -> Option A

6. Quick Check:

   Weighted sum of values = [[10, 20]] [OK]

Hint: Calculate dot, softmax, then weighted sum of values [OK]

Common Mistakes:

• Skipping softmax normalization
• Using keys instead of values for output
• Incorrect dot product calculation

4. Identify the error in this attention weight calculation code snippet:

import numpy as np
Q = np.array([[1, 0]])
K = np.array([[1, 0], [-10, 1]])
scores = np.dot(Q, K)
weights = np.exp(scores) / np.sum(np.exp(scores))

medium

A. Values are missing in the calculation

B. Softmax is applied incorrectly

C. Queries and keys have incompatible shapes

D. Keys should be transposed before dot product

Solution

1. Step 1: Check dot product operation

   Dot product should be between Q and K transposed to align dimensions correctly.

2. Step 2: Analyze code

   Code uses np.dot(Q, K) without transposing K, causing wrong shape and incorrect scores.

3. Final Answer:

   Keys should be transposed before dot product -> Option D

4. Quick Check:

   Transpose keys before dot product [OK]

Hint: Always transpose keys before dot product with queries [OK]

Common Mistakes:

• Forgetting to transpose keys
• Misapplying softmax formula
• Ignoring shape compatibility

5. In a transformer model, why is scaling the dot product by the square root of the key dimension important before applying softmax?

hard

A. To increase the dot product values for better attention

B. To prevent large dot product values causing very small gradients

C. To normalize the values between 0 and 1

D. To reduce the number of keys used in attention

Solution

1. Step 1: Understand dot product scaling

   Large dot products can cause softmax to produce very small gradients, slowing learning.

2. Step 2: Role of scaling by sqrt of key dimension

   Scaling reduces dot product magnitude, stabilizing gradients and improving training.

3. Final Answer:

   To prevent large dot product values causing very small gradients -> Option B

4. Quick Check:

   Scaling avoids tiny gradients in softmax [OK]

Hint: Scale dot product to keep gradients healthy [OK]

Common Mistakes:

• Thinking scaling increases dot product
• Confusing scaling with normalization to [0,1]
• Assuming scaling reduces keys count