NCA-GENL Question 176
Single answerYou are monitoring the data collection process for training a large language model. During the process, you notice a significant drop in the volume of incoming data from one of the primary sources. What is the best immediate action to take to ensure the data collection process remains effective?
- A
Pause the entire data collection pipeline until the issue is resolved.
- B
Investigate the primary data source for connectivity or availability issues.
- C
Proceed with training using the reduced dataset to save time.
- D
Switch to a pre-configured backup data source to maintain continuity.
Show answer and explanation
Correct answer: B
Explanation
Monitoring the functioning of data collection processes involves promptly identifying and addressing issues that arise. Investigating the primary data source for connectivity or availability issues is the correct course of action, as it allows you to diagnose and resolve the root cause of the problem without unnecessarily disrupting the pipeline or compromising data quality.
- A. Incorrect.
Pausing the entire pipeline may cause unnecessary delays, especially if the issue is isolated to one source.
- B. Correct.
Investigating the primary data source is the most logical and immediate step to identify and resolve the drop in data volume.
- C. Incorrect.
Proceeding with training using incomplete data risks introducing biases or reducing the model's performance due to insufficient or unbalanced data.
- D. Incorrect.
Switching to a backup data source without understanding the root cause of the issue could lead to similar problems or inconsistencies.