PromptEngineering

Question 1

The act of developing, designing and optimizing prompts to enhance the output of foundation models is called ____.

Answer 1

prompt engineering

Question 2

Improved prompting techniques consists of what four areas?

Answer 2

• instructions: as task for the model to do
• context: external information to guide the model
• input data: the input for which you want a response
• output indicator: the output type or format

Question 3

____ prompting is a technique where you explicitly instruct the model on what not to include or do in its response.

Answer 3

Negative

Question 4

Negative prompting help to do what three things?

Answer 4

avoid unwanted content, maintain focus, enhance clarity

Question 5

The model’s ____ determine how the model should behace and reply.

Answer 5

system prompts

Question 6

A model’s ____ setting controls the creativity of the model’s output and can range from 0 to 1.

Answer 6

temperature; the lower the setting, the outputs are more conservative, repetitve, focused on most likely responses; the higher the setting, the more diverse, createive and unpredictable and maybe less coherent.

Question 7

A model’s ____ setting determines which likely words will be selected (if .25 for example, it will consider th e 25% most likely words).

Answer 7

Top P; a low number will make a more coherent response, a high number considers a broad range of possible words and possible more creative an diverse output

Question 8

A model’s ____ setting limits the number of probable words to select from.

Answer 8

Top K; a low number leads to more coherent response and less probable words. A high number has more probable words and more diverse and creative output.

Question 9

A model’s ____ sets a maximum length on the answer.

Answer 9

length

Question 10

A model’s ____ are tokens that signal the model to stop generating output.

Answer 10

stop sequences

Question 11

A model’s prompt ____ is how fast the model responds.

Answer 11

latency

Question 12

True/False: Prompt latency is not impacted by Top P, Top K or Temperature.

Answer 12

t

Question 13

List 4 parameters that can impact prompt latency.

Answer 13

model size, model type, number of tokens in the input, number of tokens in the output

Question 14

____ prompting is to present a task to the model without providing examples or explicit training for that specific task.

Answer 14

Zero-Shot

Question 15

With zero-shot prompting, you fully rely on the model’s ____.

Answer 15

general knowledge

Question 16

When using ____ prompting, the larger and more capable the FM, the more likely you’ll get good results.

Answer 16

zero-shot

Question 17

____ prompting provides examples of a task to the model to guide its output.

Answer 17

Few-shots (if only one example is provided, its called one-shot or single-shot)

Question 18

With ____ prompting, we divide the task into a sequence of reasoning steps, leading to more structure and coherence.

Answer 18

Chain of Thought

Question 19

When you want to solve a problem that usually requires several steps, its helpful to use the ____ prompting technique by including a sentence like “think step by step”.

Answer 19

Chain of Thought

Question 20

Prompt ____ can be used to simplify and standardize the process of generating prompts.

Answer 20

templates

Question 21

Prompt templates can help with :

a) processes user input text and output prompts from FMs
b) orchestrates between the FM, action groups, and knowledge bases
c) formats and returns responses to the user

Question 22

True/False: Prompt templates can be used with Bedrock Agents.

Answer 22

t

Question 23

You can protect your prompt templates against ____ by adding explicit instructions to ignore any unrelated or potential malicious content.

Answer 23

prompt injections