AI End

The cosine similarity between two non-zero vectors, A and B, is calculated using the following formuia:
cosine simitarity =
Where:
  A - B is the dot product (or inner product) of the vectors.
  ||A|| and |B are the magnitudes (or Euclidean norms/lengths) of the vectors. •
What is the cosine similarity between two embedding vectors u = (3, 0, 4) and v = (2, 2, 1) representing the context
of texts of a two company reports. What is the similarity between these two reports?

• 3/4

• 2/3

• 3/5

• 1/2

• 1/3

A classifier's output layer changes from 10 classes to 4 classes. Each neuron receives 512 inputs plus bias. How many fewer parameters does the new layer contain?

• A. 2,560 fewer parameters overall

• B. 3,078 fewer parameters overall

• C. 3,520 fewer parameters overall

• D. 4,096 fewer parameters overall

• E. 4,800 fewer parameters overall

Which of these is a key advantage of random forests over individual decision trees?

• A. Lower computational complexity

• B. Better interpretability for end-users

• C. More accurate predictions most of the time due to ensemble averaging

• D. Faster training time compared to single decision trees

• E. Simpler implementation in most software tools

When modifying the output layer of a classification neural network to support a new set of target classes, which factor most directly determines how the output layer must be redesigned?

• A. The dimensionality of the internal feature embeddings produced by the model

• B. The batch size used during the final epoch of model fine-tuning
  

• C. The number of unique labels that the model must predict at inference

• D. The dropout probability applied in the preceding hidden layers
  

• E. The type of optimizer used during the training procedure

  
  

What is tokenization in NLP?

• A Assigning probables to class labels
  

• B. Splitting text into units such as words or subwords

• C. Actregating all text data into a singl epresentation

• D. Removing irrelevant information from text

• E. Reducing text to its base form
  

What is the purpose of identifying stop words in NLP tasks?

• A. To emphasize the most frequent words
  in text

• B. To improve efficiency by removing commonly used words that might add little meaning

• C. To replace rare words with similar high-
  frequency words

• D. To reduce overfitting during model training

• E. To identify important topic-specific terms

Why are word embeddings significant in modern NLP models?

• A. They simplify tokenization processes
  

• B. They represent words as vectors that capture semantic meaning
  

• C. They are exclusively used for supervised
  

• D. They remove the need for pretraining in LLMs
  learning tasks

• E. They ensure that each token has a unique representation
  

What is the role of the attention mechanism in NLP?

• A. To eliminate stop words dynamically

• B. To assign importance to different parts of input data for context understanding

• C. To tokenize input data into smaller units
  

• D. To generate embeddings for sequences
  

• E. To reduce the dimensionality of input data
  

What does zero-shot learning enable in language models?

• A. Learning tasks with no specific training data for those tasks
  

• B. Adapting to unseen data using a small labeled dataset
  

• C. Using reinforcement learning for improved accuracy
  

• D. Generating text using unsupervised methods exclusively

• E. Predicting sequences without embeddings
  

What is transfer learning in NLP?

• Using pre-trained knowledge from one task to enhance model performance for another lask

• Training all word embenddings from randomly initialized state exclusively on the data specific to the intendes target task

• Transfering generative models from unsupervised text generation and iterative quality refinement

• Applying clustering algorithms to a large dataset of text to apply clusters to a smaller dataset.

• Integrating a reinforcement learning policy with a language model to guide text generation based on human