Explanation:
Explanation:
Top-p (nucleus sampling) is a parameter that controls the diversity of generated text by selecting from the smallest set of tokens whose cumulative probability exceeds the top-p value.
For marketing slogans, you want creativity but not complete randomness, so adjusting top-p appropriately helps balance between focused creativity and excessive randomness.
Ultimate access to all questions.
You are configuring a Bedrock model for generating marketing slogans. To ensure the output remains creative but not too random, you adjust the top-p (nucleus sampling) parameter. What effect does increasing the top-p value have?
A
Limits the model to only the top 1% of probable words
B
Allows the model to consider a broader range of likely next words, increasing variability
C
Makes the model more deterministic and repetitive
D
Forces the model to only select the most probable token each time
No comments yet.