What is an example of a pre-trained model?

• A. bert-base-uncased
• B. google/gemma-2-2b
• C. gpt-3.5-turbo
• D. resnet50

Answer: (B)

A pre-trained model is one that already has learned weights from training on large data, so you can load it and use it or fine-tune it for your task without starting from scratch. The example google/gemma-2-2b fits this idea because it is a specific model with pre-trained weights published by Google and available to load directly from the Hugging Face Hub. You can use it out of the box or adapt it for your needs, illustrating what it means to be pre-trained. The other options either reference models accessed only via API or are not as clearly presented as a ready-to-load, pre-trained weight example in this context.