AWS Certified Cloud Practitioner

Explanation

Correct Answer: A - Both control creativity and can be tuned together for optimal output diversity

Why this is correct:

In Amazon Bedrock (and generally in language models), temperature and top-p (nucleus sampling) are both parameters that control the randomness and creativity of model outputs:

1. Temperature: Controls the randomness of predictions by scaling the logits before applying softmax. Higher temperature (e.g., 0.8-1.0) makes the output more diverse and creative, while lower temperature (e.g., 0.1-0.3) makes it more deterministic and focused.

2. Top-p (Nucleus Sampling): Controls the cumulative probability threshold for token selection. The model considers only the smallest set of tokens whose cumulative probability exceeds the top-p value (e.g., 0.9 means considering tokens until their cumulative probability reaches 90%).

How they work together:

• These parameters can be tuned together to achieve optimal balance between creativity and coherence

• Temperature affects the entire probability distribution, while top-p focuses on the high-probability tokens

• Using them together allows fine-grained control over output diversity

Why other options are incorrect:

• B: Incorrect - Top-p does not affect latency only; it affects token selection and output diversity

• C: Incorrect - Temperature doesn't control grammar, and top-p doesn't control sentence length

• D: Incorrect - These parameters don't reduce vocabulary size; they control how tokens are selected from the existing vocabulary

Best Practice: When using Amazon Bedrock models, experiment with different combinations of temperature and top-p values to achieve the desired balance between creativity and coherence for your specific use case.