Task Statement 3.3: Describe the Training and Fine-Tuning Process for FMs. AIF-C01 Practice Questions (Page 6)

Questions 26–30

1. 26foundation · easy

   How does the size of the fine-tuning dataset affect model performance?

   • ASmaller datasets are always better because they avoid overfitting
   • BLarger datasets always lead to better performance
   • CDataset size has no impact on model performance
   • DSufficiently large datasets with accurate labels generally improve performance
   Reveal Solution Discussion Select an answer first
2. 27foundation · easy

   Which dataset is used for instruction tuning?

   • ALarge unlabeled text corpora
   • BDomain-specific unlabeled text
   • CInstruction-response pairs
   • DHuman preference rankings
   Reveal Solution Discussion Select an answer first
4. 28foundation · easy

   Which activity is part of data curation for fine-tuning?

   • AIncreasing the model's learning rate
   • BCollecting human preference feedback
   • CReducing the number of layers in the model
   • DCleaning and filtering the dataset to remove irrelevant or low-quality examples
   Reveal Solution Discussion Select an answer first
5. 29application · medium

   A financial services firm is fine-tuning a foundation model to summarize regulatory filings. They have gathered a large dataset of filings, but it contains many duplicate documents, outdated templates, and some filings from unrelated industries. The team wants to ensure the model learns the correct style and content. What is the MOST important first step before fine-tuning?

   • ACurate the dataset by removing duplicates, filtering out irrelevant filings, and standardizing the format.
   • BApply RLHF to have financial experts rate the model's summaries.
   • CPerform continuous pre-training on the raw dataset before fine-tuning.
   • DIncrease the dataset size by adding more filings from all industries to improve generalization.
   Reveal Solution Discussion Select an answer first
6. 30foundation · easy

   Why is data governance important in fine-tuning?

   • ATo increase the model's accuracy on all tasks
   • BTo make the model smaller and faster
   • CTo ensure the model is trained on ethically sourced and legally compliant data
   • DTo reduce the computational cost of training
   Reveal Solution Discussion Select an answer first