Question 1

You are working with a Large Language Model (LLM) trained on general-purpose data, but now you need it to perform well on legal document processing. After fine-tuning the model on a legal dataset, it still struggles with accurately interpreting certain legal terminologies and produces inconsistent outputs. What would be the most effective next step to improve the model's performance?

Options :

A : Apply data augmentation techniques on the general-purpose dataset.

B : Fine-tune the model on a larger legal dataset.

C : Integrate domain-specific embeddings for legal terms.

D : Add more general-purpose data to the training set.

Answer: C

Question 2

You have trained two different generative AI models, Model X and Model Y, on the same dataset. To compare their performance, you calculate the Mean Squared Error (MSE) and the proportion of explained variance (R²). Model X has a lower MSE but also a lower R² compared to Model Y. Which model is likely better at explaining the variability in the data?

Options :

A : Model X, because it has a lower MSE

B : Model Y, because it has a higher MSE

C : Model Y, because it has a higher R²

D : Model X, because a lower MSE always indicates a better model

Answer: C

Question 3

You are working on deploying a Large Language Model (LLM) for a financial institution where it needs to generate detailed reports based on user inputs. The LLM occasionally generates responses that are logically correct but irrelevant to the user's query. The senior team member suggests a modification in the model’s attention mechanism to improve relevance. What adjustment to the attention mechanism would most likely improve the relevance of the LLM’s responses?

Options :

A : Implement a sparse attention mechanism to focus on fewer but more relevant tokens.

B : Increase the number of attention heads during the model-s fine-tuning process.

C : Decrease the sequence length to reduce the complexity of the attention mechanism.

D : Use a multi-task learning approach to train the attention mechanism on multiple related tasks.

Answer: A

Question 4

In the evaluation of Natural Language Processing (NLP) systems, what do ‘validity’ and ‘reliability’ imply regarding the selection of evaluation metrics?

Options :

A :

Validity involves the metric’s ability to predict future trends in data, and reliability refers to its capacity to integrate with multiple data sources.

B :

Validity ensures the metric accurately reflects the intended property to measure, while reliability ensures consistent results over repeated measurements.

C :

Validity is concerned with the metric’s computational cost, while reliability is about its applicability across different NLP platforms.

D :

Validity refers to the speed of metric computation, whereas reliability pertains to the metric’s performance in high-volume data processing

Answer: B

Question 5

You are analyzing a large dataset of sales transactions from a retail company to identify factors that most significantly impact customer purchase behavior. The dataset contains missing values, outliers, and categorical features. Which sequence of actions would be most appropriate to clean and prepare the data for modeling?

Options :

A : Remove all rows with missing values, normalize all features using Min-Max scaling, and label encode categorical features.

B : Fill missing values with zeros, convert categorical features to numeric using ASCII encoding, and leave outliers as they are to avoid data manipulation.

C : Impute missing values using the median for numerical features, encode categorical features using one-hot encoding, and remove outliers based on the interquartile range (IQR).

D : Drop all columns with any missing values, scale numerical features with StandardScaler, and manually encode categorical variables using domain knowledge.

Answer: C

Confidently Practice Online with Free NCA-GENL Exam Cram

Buying Options: