NCA-GENL Question 134
Single answerYou are working on a generative AI project and need to visualize the performance metrics of your LLM, such as perplexity and accuracy over multiple training epochs. Which of the following tools or libraries would be most appropriate for creating graphs or charts to effectively convey these results?
- A
Matplotlib
- B
Pandas
- C
scikit-learn
- D
TensorBoard
Show answer and explanation
Correct answer: A
Explanation
Matplotlib is the most appropriate library for creating detailed and customizable graphs or charts to convey the results of data analysis. While tools like TensorBoard are useful for tracking metrics during training, they are not general-purpose visualization libraries. Pandas and scikit-learn serve different roles and are not ideal for this specific task.
- A. Correct.
Matplotlib is a widely used library for creating graphs and charts in Python. It is well-suited for visualizing performance metrics like perplexity and accuracy over training epochs.
- B. Incorrect.
Pandas is primarily used for data manipulation and analysis. While it can generate simple plots, it is not as robust as Matplotlib for creating detailed visualizations.
- C. Incorrect.
scikit-learn is a machine learning library focused on training and evaluating models. It does not specialize in creating visualizations.
- D. Incorrect.
TensorBoard is useful for monitoring model training and visualizing metrics in real-time, but it is not a general-purpose graphing library like Matplotlib.