Sequence-to-sequence basics in TensorFlow - Model Metrics & Evaluation

Metrics & Evaluation - Sequence-to-sequence basics

Which metric matters for Sequence-to-sequence and WHY

In sequence-to-sequence tasks like translation or summarization, accuracy and BLEU score are important. Accuracy shows how often the model predicts the exact next token correctly. BLEU score measures how close the whole predicted sequence is to the correct sequence, even if not exact. This helps us understand if the model is learning to generate meaningful sequences.

Confusion matrix or equivalent visualization

For sequence-to-sequence, confusion matrices are less common because outputs are sequences, not single labels. Instead, we look at token-level accuracy or sequence-level metrics.

Token-level example:
True sequence:    [I, am, happy]
Predicted:        [I, am, sad]

Token accuracy = 2 correct tokens / 3 total tokens = 66.7%

BLEU score compares n-gram overlaps between predicted and true sequences.
    

Precision vs Recall tradeoff with concrete examples

In sequence generation, precision means how many predicted tokens are correct, recall means how many true tokens were predicted. For example, in chatbot replies, high precision means the reply is mostly correct words, high recall means the reply covers most of the expected content.

Sometimes, a model might generate safe but short replies (high precision, low recall). Other times, it might generate longer replies with more content but some errors (higher recall, lower precision). Balancing these helps make replies both accurate and informative.

What "good" vs "bad" metric values look like for sequence-to-sequence

Good: Token accuracy above 80%, BLEU score above 0.5 usually means the model generates sequences close to the target. This means the model understands the task well.

Bad: Token accuracy below 50%, BLEU score near 0 means the model predictions are mostly wrong or random. The model might be guessing or not learning the sequence patterns.

Common pitfalls in sequence-to-sequence metrics

• Ignoring sequence length: Short predictions can have high accuracy but miss important content.
• Overfitting: High training accuracy but low validation BLEU means the model memorizes training sequences.
• Data leakage: If test sequences appear in training, metrics look falsely good.
• Using only accuracy: It misses sequence quality; BLEU or ROUGE give better insight.

Self-check question

Your sequence-to-sequence model has 98% token accuracy but BLEU score of 0.1 on test data. Is it good for production?

Answer: No. High token accuracy but very low BLEU means the model predicts many tokens correctly but fails to produce meaningful full sequences. It might predict common words but not the right order or context. You should improve sequence-level learning before production.

Key Result

Sequence-to-sequence models require both token-level accuracy and sequence-level BLEU score to evaluate meaningful predictions.

Practice

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

easy

A. To calculate the loss during training

B. To process the input sequence and create a summary representation

C. To generate the output sequence directly

D. To shuffle the input data randomly

Solution

1. Step 1: Understand the encoder's function

   The encoder reads the input sequence and compresses its information into a fixed-size context vector.

2. Step 2: Differentiate encoder from decoder

   The decoder uses this context vector to generate the output sequence, so the encoder does not generate output directly.

3. Final Answer:

   To process the input sequence and create a summary representation -> Option B

4. Quick Check:

   Encoder = input summary [OK]

Hint: Encoder summarizes input; decoder creates output [OK]

Common Mistakes:

• Confusing encoder with decoder
• Thinking encoder generates output sequence
• Assuming encoder shuffles data

2. Which TensorFlow layer is commonly used to build the encoder and decoder in sequence-to-sequence models?

easy

A. Dense

B. Dropout

C. LSTM

D. Conv2D

Solution

1. Step 1: Identify sequence processing layers

   LSTM layers are designed to handle sequences by remembering information over time steps.

2. Step 2: Exclude other layers

   Dense layers are fully connected but not specialized for sequences; Conv2D is for images; Dropout is for regularization.

3. Final Answer:

   LSTM -> Option C

4. Quick Check:

   LSTM = sequence layer [OK]

Hint: LSTM handles sequences best in TensorFlow [OK]

Common Mistakes:

• Choosing Dense for sequence memory
• Confusing Conv2D with sequence layers
• Using Dropout as main sequence layer

3. What will be the output shape of the decoder LSTM(128, return_sequences=True) layer if the input shape is (batch_size, 10, 64)?

medium

A. (batch_size, 10, 128)

B. (batch_size, 128)

C. (batch_size, 10, 64)

D. (batch_size, 64)

Solution

1. Step 1: Understand LSTM output with return_sequences=True

   When return_sequences=True, LSTM returns output for each time step, so output shape is (batch_size, time_steps, units).

2. Step 2: Match units and input shape

   The LSTM has 128 units and input time steps are 10, output shape is (batch_size, 10, 128).

3. Final Answer:

   (batch_size, 10, 128) -> Option A

4. Quick Check:

   return_sequences=True keeps time steps [OK]

Hint: return_sequences=True keeps time steps in output [OK]

Common Mistakes:

• Ignoring return_sequences effect
• Confusing units with input feature size
• Assuming output shape drops time steps

4. Identify the error in this TensorFlow sequence-to-sequence decoder snippet:

decoder_lstm = tf.keras.layers.LSTM(64, return_sequences=True, return_state=True)
outputs, state_h = decoder_lstm(decoder_inputs, initial_state=encoder_states)

medium

A. initial_state should be None

B. LSTM units should be 128, not 64

C. return_sequences should be False for decoder

D. Missing unpacking of the second state variable

Solution

1. Step 1: Check LSTM return values

   When return_state=True, LSTM returns three outputs: output, state_h, and state_c.

2. Step 2: Verify unpacking variables

   The code unpacks only two variables, missing the cell state (state_c), causing an error.

3. Final Answer:

   Missing unpacking of the second state variable -> Option D

4. Quick Check:

   return_state=True returns 3 outputs [OK]

Hint: return_state=True returns 3 outputs; unpack all [OK]

Common Mistakes:

• Unpacking only two outputs instead of three
• Changing units without reason
• Misusing return_sequences parameter

5. You want to build a sequence-to-sequence model in TensorFlow to translate sentences. Which approach correctly connects encoder and decoder?

hard

A. Use encoder LSTM to get states, then pass these states as initial_state to decoder LSTM

B. Feed decoder inputs directly to encoder without states

C. Use two separate LSTMs without sharing states or outputs

D. Only use Dense layers for both encoder and decoder

Solution

1. Step 1: Understand encoder-decoder connection

   The encoder processes input and produces states that summarize the input sequence.

2. Step 2: Use encoder states to initialize decoder

   Passing encoder states as initial_state to decoder LSTM allows the decoder to start generating output based on input context.

3. Final Answer:

   Use encoder LSTM to get states, then pass these states as initial_state to decoder LSTM -> Option A

4. Quick Check:

   Encoder states initialize decoder [OK]

Hint: Pass encoder states to decoder initial_state [OK]

Common Mistakes:

• Ignoring encoder states in decoder
• Feeding decoder inputs to encoder
• Using only Dense layers for sequences