Self-attention mechanism in PyTorch - Model Pipeline Trace

Model Pipeline - Self-attention mechanism

The self-attention mechanism helps a model look at all parts of a sentence to understand the importance of each word when making predictions. It compares each word to every other word to decide what to focus on.

Data Flow - 5 Stages

1Input Embeddings

1 sentence x 5 words x 8 features→Convert words into vectors representing their meaning→1 sentence x 5 words x 8 features

[[0.1, 0.3, ..., 0.2], [0.0, 0.5, ..., 0.1], ...]

↓

2Linear Projections (Q, K, V)

1 x 5 x 8→Create Query, Key, and Value vectors by multiplying embeddings with weight matrices→1 x 5 x 8 for each Q, K, V

Q: [[0.2, 0.1, ...], ...], K: [[0.3, 0.0, ...], ...], V: [[0.5, 0.2, ...], ...]

↓

3Attention Scores

Q: 1 x 5 x 8, K: 1 x 5 x 8→Calculate scores by dot product of Q and K transpose, then scale→1 x 5 x 5

[[1.2, 0.5, 0.3, 0.7, 0.9], ...]

↓

4Softmax on Scores

1 x 5 x 5→Convert scores to probabilities that sum to 1 for each word→1 x 5 x 5

[[0.4, 0.1, 0.1, 0.2, 0.2], ...]

↓

5Weighted Sum of Values

Attention weights: 1 x 5 x 5, V: 1 x 5 x 8→Multiply attention weights by V vectors and sum to get output→1 x 5 x 8

[[0.3, 0.2, ..., 0.4], ...]

Training Trace - Epoch by Epoch

Loss
1.2 |****
0.9 |***
0.7 |**
0.5 |*
0.4 |

Epoch	Loss ↓	Accuracy ↑	Observation
1	1.2	0.45	Model starts learning, loss is high, accuracy low
2	0.9	0.60	Loss decreases, accuracy improves
3	0.7	0.72	Model learns important word relations
4	0.5	0.80	Better focus on relevant words
5	0.4	0.85	Training converges with good accuracy

Prediction Trace - 5 Layers

Layer 1: Input Embeddings

Layer 2: Linear Projections to Q, K, V

Layer 3: Attention Scores Calculation

Layer 4: Softmax on Scores

Layer 5: Weighted Sum of Values

Model Quiz - 3 Questions

What does the softmax step do in self-attention?

A Turns scores into probabilities that sum to 1

B Multiplies queries and keys

C Creates word embeddings

D Calculates loss during training

Key Insight

Self-attention lets the model weigh how important each word is compared to others in a sentence, helping it understand context better. Training shows loss going down and accuracy going up, meaning the model learns to focus on the right words.

Practice

1. What is the main purpose of the self-attention mechanism in models like Transformers?

easy

A. To convert input data into images

B. To increase the size of the input data

C. To reduce the number of layers in the model

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

Solution

1. Step 1: Understand self-attention's role

   Self-attention helps the model decide which parts of the input are important by comparing each part to others.

2. Step 2: Identify the main goal

   The main goal is to focus on important parts, not to change input size or model layers.

3. Final Answer:

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

4. Quick Check:

   Self-attention = focus on important parts [OK]

Hint: Self-attention means focusing on important input parts [OK]

Common Mistakes:

• Thinking self-attention changes input size
• Confusing self-attention with model depth
• Assuming it converts data formats

2. Which of the following is the correct PyTorch code snippet to compute scaled dot-product attention scores given query Q and key K tensors?

easy

A. scores = torch.matmul(Q, K.transpose(-2, -1)) / math.sqrt(Q.size(-1))

B. scores = torch.matmul(K, Q.transpose(-2, -1)) * math.sqrt(Q.size(-1))

C. scores = torch.matmul(Q, K) / Q.size(0)

D. scores = torch.matmul(Q.transpose(-2, -1), K) / math.sqrt(K.size(-1))

Solution

1. Step 1: Recall scaled dot-product attention formula

   Attention scores = (Q x K^T) / sqrt(d_k), where d_k is the dimension of keys.

2. Step 2: Match formula with code

   scores = torch.matmul(Q, K.transpose(-2, -1)) / math.sqrt(Q.size(-1)) correctly transposes K, multiplies Q and K^T, and divides by sqrt of last dimension size.

3. Final Answer:

   scores = torch.matmul(Q, K.transpose(-2, -1)) / math.sqrt(Q.size(-1)) -> Option A

4. Quick Check:

   Scaled dot-product = QxK^T / sqrt(d_k) [OK]

Hint: Transpose keys and divide by sqrt of key dimension [OK]

Common Mistakes:

• Multiplying K by Q instead of Q by K^T
• Forgetting to transpose keys
• Using multiplication instead of division by sqrt

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

import torch
Q = torch.randn(2, 4, 8)  # batch=2, seq_len=4, dim=8
K = torch.randn(2, 4, 8)
V = torch.randn(2, 4, 8)

scores = torch.matmul(Q, K.transpose(-2, -1)) / (8 ** 0.5)
weights = torch.nn.functional.softmax(scores, dim=-1)
attn_output = torch.matmul(weights, V)

medium

A. (4, 2, 8)

B. (2, 4, 8)

C. (2, 8, 4)

D. (4, 8, 2)

Solution

1. Step 1: Analyze input tensor shapes

   Q, K, V have shape (batch=2, seq_len=4, dim=8).

2. Step 2: Understand matrix multiplications

   Scores shape: (2,4,4), weights shape: (2,4,4), multiplying weights (2,4,4) by V (2,4,8) results in (2,4,8).

3. Final Answer:

   (2, 4, 8) -> Option B

4. Quick Check:

   Output shape = (batch, seq_len, dim) = (2,4,8) [OK]

Hint: Output shape matches Q and V batch and seq_len dims [OK]

Common Mistakes:

• Confusing batch and sequence dimensions
• Mixing up transpose effects on shape
• Assuming output shape changes dimension size

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

import torch
Q = torch.randn(1, 3, 5)
K = torch.randn(1, 3, 5)
V = torch.randn(1, 3, 5)

scores = torch.matmul(Q, K) / (5 ** 0.5)
weights = torch.nn.functional.softmax(scores, dim=-1)
output = torch.matmul(weights, V)

medium

A. Q and V should have the same shape

B. Softmax should be applied on dim=0 instead of dim=-1

C. K should be transposed before multiplication with Q

D. Division by sqrt(5) is incorrect

Solution

1. Step 1: Check matrix multiplication for scores

   Q shape is (1,3,5), K shape is (1,3,5). To multiply Q by K^T, K must be transposed on last two dims.

2. Step 2: Identify missing transpose

   Code multiplies Q by K directly, causing shape mismatch or wrong attention scores.

3. Final Answer:

   K should be transposed before multiplication with Q -> Option C

4. Quick Check:

   QxK^T needed, missing transpose causes error [OK]

Hint: Always transpose keys before multiplying with queries [OK]

Common Mistakes:

• Skipping transpose of K
• Applying softmax on wrong dimension
• Confusing shapes of Q and V

5. You want to implement multi-head self-attention in PyTorch with 4 heads. If your input tensor has shape (batch=2, seq_len=6, dim=32), what should be the dimension size of each head's query, key, and value vectors?

hard

A. 8

B. 4

C. 16

D. 32

Solution

1. Step 1: Understand multi-head attention dimension split

   Total dimension 32 is split evenly across 4 heads, so each head gets 32 / 4 = 8 dimensions.

2. Step 2: Confirm dimension per head

   Each head's query, key, and value vectors have dimension 8 to keep total dimension consistent.

3. Final Answer:

   8 -> Option A

4. Quick Check:

   32 total dim / 4 heads = 8 per head [OK]

Hint: Divide total dim by number of heads for each head size [OK]

Common Mistakes:

• Using total dimension for each head
• Dividing by wrong number of heads
• Confusing batch size with dimension