Fine-Tuning vs Feature Extraction

Hard

neural-networks OpenAI Google Anthropic

When applying a pre-trained model to a new task, you can either fine-tune the entire network or use it as a fixed feature extractor. Under what conditions would you prefer feature extraction over full fine-tuning?

Multiple Choice

When the target dataset is very small and similar to the pre-training domain, since fine-tuning would risk severe overfitting with so few examples When the target dataset is very large and from a completely different domain, since pre-trained features would be irrelevant When the pre-trained model uses a different architecture than what is needed for the target task When real-time inference speed is critical, since frozen layers always execute faster than trainable ones

Correct!

Incorrect — try again next time!

Upgrade to Pro to unlock solutions and explanations

Fine-Tuning vs Feature Extraction

Multiple Choice

Related problems

Deep-dive guides