Streaming in production in LangChain - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to enable streaming output in a LangChain LLM call.

LangChain

from langchain.llms import OpenAI
llm = OpenAI(streaming=[1])
response = llm("Hello, how are you?")

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

ANone

BFalse

C"yes"

DTrue

Attempts:

3 left

💡 Hint

Common Mistakes

Using a string instead of a boolean for streaming parameter.

Leaving streaming as False or None, which disables streaming.

Setting streaming=True enables the LLM to stream tokens as they are generated.

2fill in blank

medium

Complete the code to define a callback handler for streaming tokens in LangChain.

LangChain

from langchain.callbacks.base import BaseCallbackHandler

class StreamHandler(BaseCallbackHandler):
    def on_llm_new_token(self, token: str, **kwargs):
        print([1])

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

Akwargs

Bself

Ctoken

DNone

Attempts:

3 left

💡 Hint

Common Mistakes

Printing self or kwargs instead of the token string.

Forgetting to print anything inside the method.

The on_llm_new_token method receives each token as a string, which we print.

3fill in blank

hard

Fix the error in attaching the streaming callback handler to the OpenAI LLM.

LangChain

llm = OpenAI(streaming=True, callbacks=[[1]])

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

AStreamHandler()

BStreamHandler

CStreamHandler.on_llm_new_token

DStreamHandler.callback

Attempts:

3 left

💡 Hint

Common Mistakes

Passing the class name without parentheses.

Passing a method instead of an instance.

You must pass an instance of the handler class, not the class itself.

4fill in blank

hard

Fill both blanks to create a streaming LLM with a callback handler and run a prompt.

LangChain

llm = OpenAI(streaming=[1], callbacks=[[2]])
result = llm("Tell me a joke.")

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

ATrue

BFalse

CStreamHandler()

DStreamHandler

Attempts:

3 left

💡 Hint

Common Mistakes

Setting streaming to False disables streaming.

Passing the class name instead of an instance for callbacks.

Streaming must be True and the callback must be an instance of StreamHandler.

5fill in blank

hard

Fill all three blanks to define a streaming callback handler that collects tokens, then print the full response.

LangChain

class CollectHandler(BaseCallbackHandler):
    def __init__(self):
        self.text = ""
    def on_llm_new_token(self, token: str, **kwargs):
        self.text += [1]

handler = CollectHandler()
llm = OpenAI(streaming=[2], callbacks=[[3]])
response = llm("Say something nice.")
print(handler.text)

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

Atoken

B" "

Chandler

DTrue

Attempts:

3 left

💡 Hint

Common Mistakes

Adding a space string instead of the token.

Passing the class name instead of the instance as callback.

Setting streaming to False.

Add each token to self.text, enable streaming=True, and pass the handler instance.