TensorFlow NLP Developer Skills Assessment

1. What is the purpose of the 'out_of_vocabulary_token' parameter in Tokenizer?

Handles words not seen during tokenizer training

Removes rare words permanently

Increases vocabulary size

Normalizes word frequencies

2. Which metric is commonly used to evaluate text classification performance?

Accuracy, precision, recall, and F1-score

Mean squared error only

Perplexity alone

Token count

3. What is 'sparse_categorical_crossentropy' used for?

Multi-class classification when labels are integers

Binary classification with one-hot encoding

Regression tasks with continuous values

Clustering text documents

4. In sequence-to-sequence models, what is the purpose of the attention mechanism?

Allows decoder to focus on relevant encoder outputs

Increases batch normalization

Reduces vocabulary size

Improves tokenization speed

5. What does the 'char_level' parameter in Tokenizer do?

Tokenizes at character level instead of word level

Converts uppercase to lowercase

Removes special characters

Splits text into sentences

6. Which loss function is appropriate for multi-class text classification?

Categorical crossentropy

Mean squared error

Hinge loss

Huber loss

7. What is the role of Dropout in NLP models?

Reduces overfitting by randomly deactivating neurons

Removes stopwords from sequences

Increases model capacity

Speeds up tokenization

8. In sentiment analysis, why is it important to handle class imbalance?

Prevents model bias toward the majority class

Reduces training time

Increases embedding dimension

Simplifies tokenization

9. What does 'sequence_length' define in pad_sequences?

Target length for all padded sequences

Minimum vocabulary size

Number of embedding dimensions

Training batch size

10. Which TensorFlow API is used for text classification with pre-trained models?

TensorFlow Hub

TensorFlow Lite

TensorFlow.js

TensorFlow Probability

11. What does the Tokenizer class in tf.keras.preprocessing.text do?

Converts text into numerical sequences

Applies sentiment analysis to documents

Removes stopwords automatically

Performs machine translation

12. In a Bidirectional LSTM (BiLSTM), what information does each output layer utilize?

Context from both past and future tokens

Only previous tokens

Only future tokens

Token frequency statistics

13. What is the main advantage of using pre-trained word embeddings like GloVe?

Captures semantic relationships learned from large corpora

Eliminates the need for training data

Automatically detects spam

Increases model size

14. Which activation function is typically used in the output layer for binary text classification?

Sigmoid

ReLU

Tanh

Linear

15. What does tf.keras.preprocessing.sequence.pad_sequences accomplish?

Truncates or pads sequences to a fixed length

Removes punctuation from text

Applies word embeddings

Splits data into train/test sets

16. In NLP, what does 'num_words' parameter typically control in Tokenizer?

Maximum vocabulary size to keep

Number of sequences to process

Embedding dimension

Training epochs

17. The tf.keras.preprocessing.text.Tokenizer converts words to indices. What determines the index assigned?

Word frequency in the training data

Alphabetical order

Random assignment

Word length

18. What is padding in the context of sequence processing?

Adding zeros to make sequences equal length

Removing rare words from vocabulary

Increasing model parameters

Smoothing probability distributions

19. Which layer is commonly used to process sequences in NLP models?

LSTM or GRU

Batch Normalization

Max Pooling

Dropout only

20. In TensorFlow, what is the purpose of an Embedding layer?

Maps discrete tokens to dense vectors in continuous space

Normalizes text input values

Splits sentences into words

Encodes one-hot vectors

TensorFlow Developer Natural Language Processing Quiz