Foundation Models Quiz

What is a foundation model?

• A model trained for a single, narrow task

• A small model used for initialization

• A large pre-trained model adaptable to multiple tasks

• A model specifically for reinforcement learning

Which of the following is a foundation model?

• SVM

• ResNet-18

• GPT-4

• k-NN

Foundation models are usually trained using:

• Supervised learning only

• Unsupervised or self-supervised learning on massive data

• Reinforcement learning

• Small labeled datasets

What enables foundation models to generalize across tasks?

• Task-specific optimization

• Handcrafted features

• Large-scale training on diverse data

• Shallow architecture

Which of the following is a challenge in foundation models?

• Lack of scalability

• Inability to pre-train

• Hallucination and bias

• Too few parameters

What is fine-tuning in the context of foundation models?

• Reducing model parameters

• Training from scratch

• Adapting a pre-trained model to a specific task

• Testing the model performance

Which of the following foundation models is designed for both vision and language tasks?

• YOLO

• BERT

• CLIP

• DeepSpeech

What is a key capability of foundation models that makes them effective across multiple tasks?

• Fixed architecture and parameters

• Extensive use of labeled datasets only

• Pre-training on broad data followed by task-specific adaptation

• Limitation to supervised learning methods

Which of the following is a key benefit of using foundation models?

• Require labeled data only

• Cannot generalize

• Reusability across domains

• Low accuracy on NLP tasks

A foundation model like GPT is primarily based on which architecture?

• CNN

• RNN

• Decision Trees

• Transformer