# Architecture
<a name="architecture"></a>


| **Question** | **Example response** | 
| --- | --- | 
| What type of generative AI model or architecture is being considered? | Transformer, convolutional neural network (CNN), recurrent neural network (RNN), decision trees, and so on. | 
| What is the expected scale or volume of data and computations? | Millions of users, petabytes of data, and so on. | 
| What are the hardware requirements (for example, CPUs or GPUs) for training and inference? | High-end GPUs, CPU clusters, cloud instances, and so on. | 
| How will the generative AI model be updated or retrained over time? | Through continuous learning, periodic retraining, manual updates, and so on. | 
| What are the data preprocessing and feature engineering requirements? | Text cleaning, image augmentation, feature selection, and so on. | 
| How will the generative AI system handle edge cases, outliers, or low-confidence inputs? | Through fallback to human oversight, request clarification, and so on. | 
| What are the latency requirements for the generative AI application? | Real-time, near real time, batch processing, and so on. |