MCQ Generator

I need to get the public YT video transcription as an agent in LangChain.

As a second agent, I will create MCQ questions based on the first agent's transcribed output

To achieve this, you can create two agents in LangChain. Here's how you can approach it:

Agent 1: Transcription Agent

1. Fetch Video Transcription:

   • Use the YouTube Transcript API to fetch the transcription of the public YouTube video.

   • Install the library:

     pip install youtube-transcript-api

   • Fetch transcription:

     from youtube_transcript_api import YouTubeTranscriptApi
     
     def fetch_transcription(video_url):
         video_id = video_url.split("v=")[1].split("&")[0]  # Extract video ID
         transcript = YouTubeTranscriptApi.get_transcript(video_id)
         return " ".join([t['text'] for t in transcript])
     
     video_url = "https://youtu.be/bq1Plo2RhYI?si=ALig8huv6lfTn_Mb"
     transcription = fetch_transcription(video_url)
     print(transcription)

2. Integrate with LangChain:

   • Create an agent in LangChain to fetch the transcription.

   from langchain.agents import initialize_agent, Tool
   from langchain.tools import tool
   from langchain.chat_models import ChatOpenAI
   
   @tool
   def get_youtube_transcription(video_url: str) -> str:
       return fetch_transcription(video_url)
   
   tools = [Tool(name="TranscriptionTool", func=get_youtube_transcription)]
   llm = ChatOpenAI(temperature=0)
   agent = initialize_agent(tools, llm, agent="zero-shot-react-description")

Agent 2: MCQ Generator Agent

1. Generate MCQs:

   • Use the transcription as input to generate MCQs.

   • Integrate a prompt that uses a generative LLM like OpenAI GPT to create questions.

   from langchain.prompts import PromptTemplate
   
   mcq_prompt_template = """
   You are an expert at generating multiple-choice questions (MCQs). Based on the transcription provided, generate 5 MCQs.
   Transcription:
   {transcription}
   
   Example format:
   Q: [Question]
   A. [Option 1]
   B. [Option 2]
   C. [Option 3]
   D. [Option 4]
   Correct Answer: [Correct Answer]
   """
   
   mcq_prompt = PromptTemplate(template=mcq_prompt_template, input_variables=["transcription"])

2. Set up the MCQ generation agent:

   @tool
   def generate_mcq(transcription: str) -> str:
       prompt = mcq_prompt.format(transcription=transcription)
       response = llm(prompt)
       return response
   
   mcq_tools = [Tool(name="MCQGeneratorTool", func=generate_mcq)]
   mcq_agent = initialize_agent(mcq_tools, llm, agent="zero-shot-react-description")

Integration Workflow

To run both agents sequentially:

# Step 1: Fetch Transcription
transcription = agent.run(video_url)

# Step 2: Generate MCQs
mcq_output = mcq_agent.run(transcription)

print("MCQs Generated:")
print(mcq_output)

Results

• The first agent fetches the transcription from the YouTube video.

• The second agent uses the transcription to generate MCQs.

You can further refine the LLM prompts and add post-processing if needed.