AIF-C01 Applications of Foundation Models Practice Test 2 — 15 Questions

Question 1

Which TWO factors are MOST important when selecting a foundation model for a text summarization task? (Choose two.)

Accepted Answer

Maximum output length (max tokens). The maximum output length (max tokens) directly determines how much text the model can generate in a single response. For text summarization, you need to ensure the model can produce a summary of the required length—whether it's a single sentence or a multi-paragraph abstract. If the max output is too small, the summary will be truncated or incomplete, failing the task.

Answer

Model creation date

Answer

Model training cost

Answer

Image support

Question 2

Which TWO actions can help reduce bias in a foundation model’s outputs? (Choose two.)

Accepted Answer

Fine-tune the model on a balanced, representative dataset. Fine-tuning on a balanced, representative dataset (Option A) directly addresses bias by ensuring the model learns from data that reflects diverse perspectives and reduces overrepresentation of any group. Careful prompt engineering with neutral wording (Option B) mitigates bias at inference time by avoiding leading or stereotypical language that could skew the model's output. Both actions target the root causes of bias in foundation model outputs.

Answer

Restrict model access to a subset of users

Answer

Increase temperature to add randomness

Answer

Use a larger foundation model

Question 3

Which THREE practices are recommended for responsible AI when deploying foundation models? (Choose three.)

Accepted Answer

Include human review for high-stakes decisions. High-stakes decisions (e.g., medical diagnosis, loan approvals) require human oversight to catch errors, edge cases, and ethical violations that the model cannot self-correct. This aligns with the responsible AI principle of accountability, ensuring a human-in-the-loop can override or validate the model's output before it impacts lives.

Answer

Avoid collecting user feedback to reduce bias

Answer

Use a black box approach to keep model internals secret

Question 4

Refer to the exhibit. This is an Amazon Bedrock invocation request for Claude. What is the purpose of the "stop_sequences" parameter?

Accepted Answer

It tells the model to stop generating when it encounters that sequence. The 'stop_sequences' parameter in Amazon Bedrock's invocation request for Claude tells the model to halt generation as soon as it encounters a specified character sequence. This allows developers to control the output format, such as stopping at a newline or a custom delimiter, ensuring the response ends exactly where intended.

Answer

It specifies a character sequence for the model to include in its response

Answer

It limits the number of tokens in the response

Answer

It controls the randomness of the response

Question 5

Which pricing model does Amazon Bedrock use for foundation model inference?

Accepted Answer

Per-token. Amazon Bedrock charges for foundation model inference based on the number of tokens processed, which includes both input and output tokens. Each model has a specific per-token price, and the total cost is calculated by multiplying the token count by the model's rate. This is the standard pricing model for generative AI inference services like Bedrock.

Answer

Per-request

Answer

Per-hour instance

Answer

Per-GB storage

Question 6

A company wants to adapt a foundation model for a custom domain with very limited labeled data and minimal cost. Which approach is most suitable?

Accepted Answer

Prompt engineering with few-shot examples. Prompt engineering with few-shot examples is the most suitable approach because it allows the company to adapt a foundation model to a custom domain using very limited labeled data and minimal cost. By providing a few input-output examples directly in the prompt, the model can infer the desired task without any weight updates, making it efficient for low-resource scenarios.

Answer

Pre-training from scratch

Answer

Reinforcement learning from human feedback

Answer

Full fine-tuning

Question 7

A developer uses Amazon Bedrock to generate code. Some outputs contain syntax errors. What is the most likely cause?

Accepted Answer

The prompt lacks constraints or examples. Syntax errors in generated code typically arise when the prompt lacks sufficient constraints or examples to guide the model toward producing syntactically valid output. Amazon Bedrock's foundation models rely on clear instructions and few-shot examples to adhere to language syntax rules; without them, the model may generate plausible-looking but incorrect code.

Answer

The max_tokens is too low

Answer

The temperature is too high

Answer

The model lacks knowledge of the language

Question 8

Which AWS service provides a serverless API for accessing foundation models with per-token pricing?

Accepted Answer

Amazon Bedrock. Amazon Bedrock is a fully managed service that provides a serverless API for accessing foundation models (FMs) from providers like AI21 Labs, Anthropic, Cohere, Meta, and Stability AI. It offers per-token pricing, meaning you pay only for the number of tokens processed in both input and output, with no upfront commitments or infrastructure management required.

Answer

Amazon API Gateway

Answer

AWS Lambda

Answer

Amazon SageMaker

Question 9

Which TWO actions are recommended for improving the factual accuracy of a foundation model's responses when using RAG?

Accepted Answer

Include relevant context from the knowledge base in the prompt. A is correct because RAG (Retrieval-Augmented Generation) improves factual accuracy by injecting retrieved, relevant context from an external knowledge base directly into the prompt. This grounds the model's response in verifiable data, reducing hallucinations and ensuring the output is based on the provided sources rather than the model's internal parametric knowledge alone.

Answer

Increase the max_tokens parameter

Answer

Use the largest foundation model available

Answer

Increase the temperature parameter

Question 10

A company operates a customer service platform that uses Amazon Bedrock with a foundation model to generate automated responses. The system has been in production for three months. Recently, customers have reported that responses are becoming repetitive and less relevant over time. The development team notices that the model's performance has degraded, especially for queries about newer products that were added after the initial deployment. The team currently uses a static prompt with a fixed knowledge base that was set up at launch. The model is invoked via the Bedrock API with standard settings. The team wants to improve response quality without incurring high costs or extensive re-engineering. What should the team do?

Accepted Answer

Implement a feedback loop to periodically update the knowledge base with new product information and use a dynamic prompt that includes recent interactions.. The core issue is that the static knowledge base and prompt have become stale as new products were added. Implementing a feedback loop to periodically update the knowledge base with new product information and using a dynamic prompt that includes recent interactions directly addresses the root cause of degradation—outdated context—without requiring costly fine-tuning or model swaps. This approach leverages Amazon Bedrock's native capabilities for retrieval-augmented generation (RAG) to keep responses relevant and non-repetitive.

Answer

Increase the temperature parameter to 0.9 to introduce more randomness and reduce repetition.

Answer

Fine-tune the model every week on the latest customer interactions using Amazon SageMaker.

Answer

Switch to a larger foundation model to handle the increased complexity of new products.

Question 11

A company needs to summarize thousands of customer reviews daily using a foundation model. The solution must minimize latency and cost while handling variable traffic. Which AWS service should they use?

Accepted Answer

Amazon Bedrock with on-demand mode. Amazon Bedrock with on-demand mode is correct because it provides serverless access to foundation models (FMs) with pay-per-use pricing, which minimizes cost for variable traffic and eliminates the need to provision infrastructure. The on-demand mode handles thousands of daily summarization requests with low latency by leveraging AWS's scalable inference infrastructure, making it ideal for variable workloads without upfront commitments.

Answer

Amazon SageMaker real-time endpoint

Answer

Amazon Comprehend

Answer

Amazon Lex

Question 12

A development team uses a foundation model via Amazon Bedrock to generate code snippets. They notice that the model sometimes produces code with security vulnerabilities, such as SQL injection. The team wants to reduce these occurrences without delaying project timelines. What should they do?

Accepted Answer

Fine-tune the model on a dataset of secure code examples.. Fine-tuning the foundation model on a curated dataset of secure code examples directly modifies the model's weights to reduce the likelihood of generating vulnerable patterns like SQL injection. This approach integrates security improvements into the model itself, avoiding the need for manual review or post-processing, and can be completed within project timelines using Amazon Bedrock's fine-tuning capabilities.

Answer

Manually review all generated code before deployment.

Answer

Switch to a smaller, specialized code generation model.

Answer

Use Amazon Bedrock's guardrails to filter insecure code.

Question 13

A legal firm wants to use a foundation model to extract key clauses from thousands of contracts. Accuracy is critical, and the model must not hallucinate or fabricate information. The firm has a large internal database of labeled contracts. Which approach should they take?

Accepted Answer

Implement Retrieval-Augmented Generation (RAG) using Amazon Bedrock and a vector store of contract clauses.. Retrieval-Augmented Generation (RAG) grounds the model's output in actual contract clauses stored in a vector store, eliminating the risk of hallucination. By retrieving relevant clauses from the firm's labeled database and passing them as context to the foundation model via Amazon Bedrock, the model generates responses based solely on verified data, not on its parametric memory. This approach ensures high accuracy and factual reliability, which is critical for legal document analysis where fabrication is unacceptable.

Answer

Use a smaller model specifically designed for legal text.

Answer

Fine-tune the model on the labeled contracts using Amazon Bedrock's fine-tuning capability.

Answer

Use a pre-trained model with detailed prompts and few-shot examples.

Question 14

A company wants to build a customer service chatbot using a foundation model. The chatbot must respond in under 2 seconds and handle high throughput. Which model deployment option should they choose?

Accepted Answer

Amazon Bedrock provisioned throughput. Amazon Bedrock provisioned throughput is the correct choice because it guarantees low-latency inference (sub-2-second responses) and high throughput by reserving dedicated model capacity for the foundation model. This option eliminates contention from other customers, ensuring consistent performance under load, which is critical for a real-time chatbot.

Answer

Amazon Bedrock on-demand

Answer

Amazon SageMaker real-time endpoint hosting a foundation model

Answer

Amazon Lex with a pre-built foundation model

Question 15

A company uses Amazon Bedrock to automatically generate product descriptions for their e-commerce website. They use a prompt that includes product attributes and a short description as a starting point. Recently, the generated descriptions have become overly verbose, including irrelevant details and sometimes even incorrect product specifications. The team has tried simplifying the prompt and reducing the max tokens, but the issue persists. The descriptions must be concise and accurate. What is the most effective next step to address this problem?

Accepted Answer

Decrease the top_p parameter to 0.1 and keep max tokens low.. Decreasing the top_p parameter to 0.1 forces the model to sample from a smaller, more probable set of tokens, making the output more focused and less likely to include irrelevant details. Keeping max tokens low enforces conciseness. Option A (switching to a larger model) may increase verbosity and cost without guarantee of improvement. Option B (increasing temperature to 0.9) would increase randomness (since higher temperature makes the model less deterministic), potentially worsening the issue. Option D (using a negative prompt) might help but is less reliable than tuning top_p and max tokens.

Answer

Switch to a larger foundation model that handles details better.

Answer

Increase the temperature parameter to 0.9 to make the model more deterministic.

Answer

Use a negative prompt specifying 'do not include unnecessary details'.