AI For Beginners 4 — Questions and Answers
Question 1: What does the acronym 'GPU' stand for, and why is it important for AI training?
- General Processing Unit; it handles all OS tasks
- Graphics Processing Unit; its massively parallel architecture accelerates matrix operations (Correct answer)
- Grid Processing Unit; it manages distributed clusters
- Gradient Processing Unit; it computes backpropagation automatically
Correct answer: Graphics Processing Unit; its massively parallel architecture accelerates matrix operations
GPUs contain thousands of small cores designed for parallel computation, making them far faster than CPUs for the matrix math in deep learning.
Question 2: Which loss function is typically used for multi-class classification problems?
- Mean Squared Error
- Binary Cross-Entropy
- Categorical Cross-Entropy (Correct answer)
- Huber Loss
Correct answer: Categorical Cross-Entropy
Categorical cross-entropy measures the divergence between predicted class probabilities and the one-hot true labels across multiple classes.
Question 3: What is the purpose of the softmax function in the output layer of a neural network?
- To clip values between -1 and 1
- To convert raw logits into a probability distribution summing to 1 (Correct answer)
- To apply dropout regularization
- To normalize input features
Correct answer: To convert raw logits into a probability distribution summing to 1
Softmax exponentiates each logit and divides by the sum, producing non-negative class probabilities that sum to exactly 1.
Question 4: In natural language processing, what does 'tokenization' refer to?
- Encrypting text for secure transmission
- Splitting raw text into smaller units such as words or subwords (Correct answer)
- Converting audio speech into text
- Compressing a document into a fixed-length vector
Correct answer: Splitting raw text into smaller units such as words or subwords
Tokenization breaks text into tokens (words, subwords, or characters) that serve as the discrete input units for NLP models.
Question 5: What is the k-nearest neighbors (k-NN) algorithm?
- A deep learning architecture with k hidden layers
- A non-parametric method that classifies a point based on the majority label of its k closest training examples (Correct answer)
- An optimization technique that updates k parameters per step
- A clustering algorithm that partitions data into exactly k groups
Correct answer: A non-parametric method that classifies a point based on the majority label of its k closest training examples
k-NN assigns the label voted by the k training samples nearest to a query point, with no explicit model learned.
Question 6: Which of the following is a key advantage of using a pre-trained large language model (LLM) over training from scratch?
- LLMs require no data for fine-tuning
- LLMs eliminate the need for any compute resources
- LLMs provide rich prior knowledge, drastically reducing data and compute needed for a target task (Correct answer)
- LLMs cannot be fine-tuned on domain-specific data
Correct answer: LLMs provide rich prior knowledge, drastically reducing data and compute needed for a target task
Pre-trained LLMs encode broad world knowledge from massive corpora, so fine-tuning them needs far fewer labeled examples and compute than training from scratch.
Question 7: What does 'epoch' mean in the context of training a machine learning model?
- A single forward pass through one training sample
- One complete pass through the entire training dataset (Correct answer)
- The time taken to complete one batch update
- The number of layers in the model
Correct answer: One complete pass through the entire training dataset
One epoch means the model has seen every training example exactly once; multiple epochs repeat this process to improve learning.
AI For Beginners 4
What does the acronym 'GPU' stand for, and why is it important for AI training?