Combining retrieved context with LLM in Prompt Engineering / GenAI - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to add the retrieved context to the prompt before sending it to the LLM.

Prompt Engineering / GenAI

prompt = "Answer the question based on the context: " + [1] + "\nQuestion: " + question

Drag options to blanks, or click blank then click option'

Auser_input

Bquestion

Cllm_response

Dretrieved_context

Attempts:

3 left

💡 Hint

Common Mistakes

Adding the question twice instead of the context.

Using the LLM response before generating it.

The retrieved context should be added to the prompt so the LLM can use it to answer the question.

2fill in blank

medium

Complete the code to combine the retrieved context and the user question into a single input for the LLM.

Prompt Engineering / GenAI

llm_input = f"Context: [1]\nQuestion: {question}"

Drag options to blanks, or click blank then click option'

Auser_input

Bllm_output

Cretrieved_context

Dsystem_message

Attempts:

3 left

💡 Hint

Common Mistakes

Using the user input instead of the retrieved context.

Forgetting to include the context in the input string.

The LLM input should include the retrieved context so it can generate an informed answer.

3fill in blank

hard

Fix the error in the code that sends the combined context and question to the LLM for completion.

Prompt Engineering / GenAI

response = llm.complete(prompt=[1])

Drag options to blanks, or click blank then click option'

Allm_input

Bquestion

Cretrieved_context

Duser_input

Attempts:

3 left

💡 Hint

Common Mistakes

Passing only the question or only the context to the LLM.

Using a variable that does not contain the full prompt.

The LLM expects the combined input (context and question) as the prompt, which is stored in llm_input.

4fill in blank

hard

Fill both blanks to create a dictionary that maps each document to its text and filters documents with length greater than 100.

Prompt Engineering / GenAI

filtered_docs = {doc.id: doc.[1] for doc in documents if len(doc.[2]) > 100}

Drag options to blanks, or click blank then click option'

Atext

Bcontent

Cmetadata

Dsummary

Attempts:

3 left

💡 Hint

Common Mistakes

Using different attributes for the dictionary value and length check.

Using metadata or summary which may not contain full text.

The document's text attribute holds the content, and filtering is done by checking the length of the text.

5fill in blank

hard

Fill all three blanks to create a prompt dictionary with keys 'context', 'question', and 'max_tokens' for the LLM call.

Prompt Engineering / GenAI

prompt_data = {"context": [1], "question": [2], "max_tokens": [3]

Drag options to blanks, or click blank then click option'

Aretrieved_context

Buser_question

C150

Dllm_response

Attempts:

3 left

💡 Hint

Common Mistakes

Using the LLM response instead of the question or context.

Setting max_tokens to a string instead of a number.

The prompt dictionary should include the retrieved context, the user's question, and a token limit for the LLM.