Which output parser should Dana use to convert an LLM response into CSV-like comma-separated values?

❌ JSONOutputParser
❌ PandaDataFrameParser
❌ XMLParser
✅ CommaSeparatedListOutputParser

Explanation:
CommaSeparatedListOutputParser converts output into comma-separated formats, suitable for CSV-like data.

---

Question 2

Correct sequence for creating an LCEL pattern:

❌ Option 1
❌ Option 2
❌ Option 3
✅ Define a template with variables → Create a PromptTemplate → Build a chain using the pipe operator → Invoke with input values

Explanation:
LCEL workflow:

1. Write template with {variables}

2. Create PromptTemplate

3. Chain using |

4. Run with inputs

---

Question 3

How does an agent integrate with external tools?

❌ Stores inputs
❌ Runs predefined commands only
❌ Creates a new tool per request
✅ Uses LLM to decide actions, then queries databases/websites via tools

Explanation:
Agents reason using the LLM and then call tools to fetch real data (APIs, DBs, search, etc.).

---

Question 4

Why is text splitting important for long documents?

❌ Aesthetics
❌ Remove duplicates
❌ Reduce token cost (side effect, not main reason)
✅ Break documents into chunks that fit within model context window

Explanation:
Models have context limits; splitting ensures each chunk fits and can be processed properly.

---

Question 5

Purpose of FewShotPromptTemplate:

❌ Store conversation history
❌ Visualize data
❌ Execute without examples
✅ Provide examples (“shots”) to guide the LLM output

Explanation:
Few-shot templates give sample inputs and outputs to steer the model’s behavior.

---

Question 6

How does LangChain use memory?

❌ Stores only final output
❌ Uses “sophisticated algorithms”
❌ Alters devices
✅ Reads and writes memory to maintain continuity across interactions

Explanation:
Memory tracks past messages so the LLM behaves consistently in a conversation.

---

Question 7

Correct LCEL syntax for chaining components:

❌ >>
✅ prompt_template | llm | output_parser
❌ .connect()
❌ Array notation

Explanation:
LCEL uses the pipe operator (|) to link components in a readable chain.

---

🧾 Summary Table