Free NVIDIA-Certified Associate: Generative AI with LLMs (NCA-GENL) Practice Questions

Sample NCA-GENL Practice Questions

Browse all 15 free NVIDIA-Certified Associate: Generative AI with LLMs practice questions below.

1. In a transformer model, what is the primary purpose of the self-attention mechanism?

   • To prevent the model from overfitting during training
   • To compress the input embedding dimension and reduce memory usage
   • To reduce the computational complexity of the model by processing tokens in parallel
   • To allow each token in a sequence to attend to all other tokens and capture contextual relationships

2. Which component of the transformer architecture is responsible for preserving the sequential order information of input tokens?

   • Feed-forward neural network
   • Multi-head attention
   • Layer normalization
   • Positional encoding

3. In multi-head attention, why does the transformer use multiple attention heads instead of a single attention mechanism?

   • To reduce the total number of parameters in the model
   • To speed up inference time by distributing computation across multiple GPUs
   • To allow the model to attend to information from different representation subspaces at different positions
   • To prevent gradient vanishing during backpropagation

4. What is the key difference between the encoder and decoder components in an encoder-decoder transformer architecture?

   • The encoder has more layers than the decoder to better process input
   • The encoder uses self-attention while the decoder uses cross-attention only
   • The decoder uses masked self-attention to prevent attending to future tokens during training, while the encoder uses bidirectional attention
   • The decoder uses larger embedding dimensions for better generation quality

5. In the attention mechanism, what mathematical operation is performed on the Query (Q) and Key (K) matrices to compute attention scores?

   • Element-wise multiplication followed by summation
   • Matrix multiplication (dot product) followed by scaling and softmax
   • Convolution operation to capture local patterns
   • Concatenation followed by a linear transformation

6. What is the primary advantage of FlashAttention in modern transformer implementations?

   • It reduces memory usage by computing attention incrementally using tiling and recomputation techniques
   • It replaces multi-head attention with a single attention head to reduce parameters
   • It uses quantization to reduce attention weights to 8-bit precision
   • It eliminates the need for softmax normalization in the attention mechanism

7. In decoder-only transformer architectures like GPT, what is the purpose of the KV (Key-Value) cache during inference?

   • To enable batch processing of multiple sequences simultaneously
   • To compress the model's parameters during deployment
   • To store previously computed key and value vectors from past tokens, avoiding redundant computation during autoregressive generation
   • To implement attention masking for causal language modeling

8. What is Multi-Query Attention (MQA) and how does it differ from standard multi-head attention?

   • MQA uses a single set of key and value projection matrices shared across all attention heads, while maintaining separate query projections per head
   • MQA processes multiple user queries simultaneously in a single forward pass
   • MQA eliminates the query projection entirely and computes attention using only keys and values
   • MQA increases the number of attention heads to improve model capacity

9. Why do decoder-only architectures like GPT use causal (masked) self-attention?

   • To compress the KV cache memory footprint during inference
   • To reduce the computational complexity of the attention mechanism by half
   • To enable bidirectional attention patterns for better context understanding
   • To prevent the model from attending to future tokens during training, ensuring each position only depends on previous positions

10. What happens during the token generation process when a decoder-only LLM produces an output?

    • The model generates all tokens in the sequence simultaneously in a single forward pass
    • The model outputs embeddings that are directly used as the final text representation
    • The model outputs logits (raw scores) for each vocabulary token, which are converted to probabilities via softmax and then sampled or selected
    • The model directly outputs text characters without any intermediate probability computation

11. In a sequence-to-sequence (seq2seq) model for machine translation, what role does the encoder play in the overall architecture?

    • It processes the input sequence and creates a fixed-length or variable-length context representation that captures the semantic meaning
    • It applies attention mechanisms to align input and output sequences during training
    • It tokenizes and normalizes the input text before processing
    • It generates the target language output tokens one at a time in an autoregressive manner

12. When computing cross-entropy loss for a language model with a vocabulary of 50,000 tokens, what does the model output before the loss calculation?

    • A probability distribution over all 50,000 vocabulary tokens, typically produced by a softmax layer
    • An embedding vector representing the semantic meaning of the predicted token
    • A single integer representing the predicted token ID
    • A binary classification result indicating whether the prediction is correct

13. In the context of training recurrent neural networks (RNNs) for sequence modeling, what is 'backpropagation through time' (BPTT)?

    • A technique that processes sequences in reverse order during the forward pass to capture backward dependencies
    • An algorithm that unfolds the RNN across time steps and applies standard backpropagation to compute gradients with respect to parameters shared across all timesteps
    • A technique for initializing RNN weights to prevent gradient problems
    • A method for scheduling learning rates that decreases them over training epochs

14. Why is Xavier (Glorot) initialization commonly used for initializing weights in deep neural networks for NLP tasks?

    • It initializes all weights to zero to ensure symmetric learning across neurons
    • It sets all weights to small constant values like 0.01 to prevent gradient explosion
    • It sets weights proportional to the number of input and output connections, helping maintain consistent variance of activations and gradients across layers
    • It initializes weights randomly from a uniform distribution between -1 and 1

15. What is the purpose of weight decay (L2 regularization) when training large language models?

    • To decrease the batch size as training progresses
    • To reduce the learning rate exponentially over time
    • To gradually remove neurons from the network during training
    • To penalize large weight values and encourage the model to learn simpler patterns, reducing overfitting