AWS Certified Cloud Practitioner

Explanation

Correct Answer: B - Increase top-p to 0.95

Why this is correct:

1. Top-P (Nucleus Sampling) controls the diversity of generated text by limiting the vocabulary to the top P% of probable tokens. Increasing top-p (e.g., to 0.95) allows the model to consider a wider range of possible tokens, making the output more diverse and creative.

2. For ad copy and marketing content, creativity and imagination are crucial. A higher top-p value encourages the model to explore more varied word choices and sentence structures, reducing the "robotic" feel.

Why other options are incorrect:

• A. Lower temperature to 0.3: Temperature controls randomness - lower temperature makes output more deterministic and focused on high-probability tokens, which would actually make the text MORE robotic, not less.

• C. Reduce max-tokens: This only limits the length of the generated text, not its creativity or style. It might truncate responses but won't make them more imaginative.

• D. Add stop-sequence after each sentence: This controls when generation stops, not the creativity of the content. It might create shorter outputs but won't improve imagination.

Key Concept:

In AWS Bedrock and other LLMs, top-p (nucleus sampling) is a key parameter for controlling creativity. Higher values (closer to 1.0) allow more diverse outputs, while lower values make outputs more focused and predictable. For creative tasks like ad copy, increasing top-p is the most effective way to reduce robotic-sounding text.