Google Cloud Generative AI Leader Free Sample Questions

Imagen is Google's text-to-image diffusion model specifically designed to generate high-quality, photorealistic images from text prompts. While Gemini is multimodal, Imagen is the specialized model for pure image generation tasks in this context. Veo is for video, and Gemma is a family of lightweight open models that output text.

Gemma is a family of lightweight, open models built from the same research and technology as Gemini. It is specifically designed for responsible AI development and can be deployed locally on workstations, edge devices, or self-managed infrastructure, meeting strict data residency and isolation requirements.

Data Preparation (often including ingestion) is the stage where raw data is collected, cleaned, labeled, and transformed into a format suitable for training models. This is distinct from training, where the model learns, or deployment, where it is put into production.

Structured data is highly organized and formatted, typically residing in relational databases (RDBMS) with defined schema (rows and columns). Unstructured data refers to text, images, or audio that lacks a pre-defined data model.

False. Foundation models are large models trained on massive amounts of diverse, largely unlabeled data. This gives them a broad understanding and makes them flexible across a wide range of use cases, and they can then be adapted to specialized domains through additional, targeted training (such as fine-tuning). A model trained on a small dataset for a single task is a traditional task-specific model, not a foundation model.

TPUs (Tensor Processing Units) and GPUs (Graphics Processing Units) are hardware accelerators that form the Infrastructure Layer. This layer provides the compute power necessary to train and serve models. The Model layer consists of the algorithms (like Gemini).