A company uses foundation models (FMs) to automate daily tasks. An AI practitioner at the company is creating system instructions that include context that is relevant to the tasks. The AI practitioner wants to save and reuse the instructions in daily interactions with FMs in Amazon Bedrock.Which Amazon Bedrock solution will meet these requirements?
Answer(s): D
Prompt management in Amazon Bedrock lets you create, save, version, and reuse prompts, including system instructions and task context, across repeated interactions.
A company uses an Amazon Bedrock foundation model (FM) to summarize documents for an internal use case. The company trained a custom model in Amazon Bedrock to improve the quality of the model's summarizations. The company needs a solution to use the customized model on Amazon Bedrock.Which solution will meet this requirement?
Answer(s): A
Customized foundation models in Amazon Bedrock require Provisioned Throughput to be purchased so the model can be deployed and used for inference, enabling the application to access and run the custom summarization model.
In which stage of the generative AI model lifecycle are tests performed to examine the model's accuracy?
The evaluation stage is when you run tests and benchmarks, such as accuracy, precision, and other performance metrics, to measure how well the generative AI model performs on hold-out or validation data before moving on to deployment.
A company is developing a new foundation model (FM) for multilingual text generation. The model must work across multiple languages and dialects. The model must adhere to the principles of responsible AI.Which model training process will meet these requirements?
Answer(s): B
Continued pre-training on a diverse dataset that represents multiple languages and dialects helps the foundation model learn broad linguistic patterns while reducing representation gaps. Using sufficiently diverse and representative training data supports responsible AI principles by improving inclusiveness and reducing the risk of systematic bias across language groups.
HOTSPOT (Drag and Drop is not supported)Select the correct prompt injection attack from the following list for each definition. Select each prompt injection attack one time.Hot Area:
Fake completion supplies pre-filled output to steer subsequent model behavior. Extracting the conversation history attempts to expose prior messages or sensitive context. Ignoring the prompt template explicitly instructs the model to disregard existing template rules. Extracting the prompt template attempts to exploit or imitate template structure to reveal or manipulate the underlying prompt instructions.
Post your Comments and Discuss Amazon AIF-C01 exam prep with other Community members: