Codebasics Q&A: Question and Answer System Based on Google Palm LLM and Langchain for E-learning company

This is an end to end LLM project based on Google Palm and Langchain. We are building a Q&A system for an e-learning company called codebasics (website: codebasics.io). Codebasics sells data related courses and bootcamps. They have thousands of learners who uses discord server or email to ask questions. This system will provide a streamlit based user interface for students where they can ask questions and get answers.

YouTube

Project Highlights

• Use a real CSV file of FAQs that Codebasics company is using right now.
• Their human staff will use this file to assist their course learners.
• We will build an LLM based question and answer system that can reduce the workload of their human staff.
• Students should be able to use this system to ask questions directly and get answers within seconds

You will learn following,

• Langchain + Google Palm: LLM based Q&A
• Streamlit: UI
• Huggingface instructor embeddings: Text embeddings
• FAISS: Vector databse

Installation

0. using WSL

1.Clone this repository to your local machine using:

  git clone https://github.com/codebasics/langchain.git
  pip install google-generativeai

2.Navigate to the project directory:

  cd 3_project_codebasics_q_and_a

3. Install the required dependencies using pip:

  pip install -r requirements.txt

4.Acquire an api key through makersuite.google.com and put it in .env file

  GOOGLE_API_KEY="your_api_key_here"

Usage

1. Run the Streamlit app by executing:

streamlit run main.py

2.The web app will open in your browser.

• To create a knowledebase of FAQs, click on Create Knolwedge Base button. It will take some time before knowledgebase is created so please wait.

• Once knowledge base is created you will see a directory called faiss_index in your current folder

• Now you are ready to ask questions. Type your question in Question box and hit Enter

Sample Questions

• Do you guys provide internship and also do you offer EMI payments?
• Do you have javascript course?
• Should I learn power bi or tableau?
• I've a MAC computer. Can I use powerbi on it?
• I don't see power pivot. how can I enable it?

Project Structure

• main.py: The main Streamlit application script.
• langchain_helper.py: This has all the langchain code
• requirements.txt: A list of required Python packages for the project.
• .env: Configuration file for storing your Google API key.

Google MakerSuite

https://makersuite.google.com

Issues

ModuleNotFoundError: No module named 'InstructorEmbedding'

2023-11-26 09:46:41.978 Uncaught app exception Traceback (most recent call last): File "/home/gong/projects/st/lib/python3.9/site-packages/langchain/embeddings/huggingface.py", line 148, in init from InstructorEmbedding import INSTRUCTOR ModuleNotFoundError: No module named 'InstructorEmbedding'

The above exception was the direct cause of the following exception:

Traceback (most recent call last): File "/home/gong/projects/st/lib/python3.9/site-packages/streamlit/runtime/scriptrunner/script_runner.py", line 565, in _run_script exec(code, module.dict) File "/home/gong/projects/AI/langchain/3_project_codebasics_q_and_a/main.py", line 2, in from langchain_helper import get_qa_chain, create_vector_db File "/home/gong/projects/AI/langchain/3_project_codebasics_q_and_a/langchain_helper.py", line 20, in instructor_embeddings = HuggingFaceInstructEmbeddings(model_name="hkunlp/instructor-large") File "/home/gong/projects/st/lib/python3.9/site-packages/langchain/embeddings/huggingface.py", line 154, in init raise ImportError("Dependencies for InstructorEmbedding not found.") from e ImportError: Dependencies for InstructorEmbedding not found.

Solution: pip install InstructorEmbedding

venv not working in WSL

https://stackoverflow.com/questions/70519432/python-virtual-env-succesfully-activated-via-wsl-but-not-working

Solution:

conda deactivate  # get out of base

pip install jupyter

cd ~/projects
source st/bin/activate # enter st

cd ~/projects/AI/langchain/3_project_codebasics_q_and_a

streamlit run main.py

load INSTRUCTOR_Transformer
max_seq_length  512

Data issue

UnicodeDecodeError: 'utf-8' codec can't decode byte 0x92 in position 1203: invalid start byte

CSVLoader on 'codebasics_faqs.csv' failed

because "�" is not invalid unicode, replace it with "'"