MicodeGPT is a chatbot application that simulates a conversation with the french content creator "Micode". The chatbot provides valuable IT info/advice to users, drawing from Micode's insights shared on both his Youtube channels "Micode" & "UnderScore_". It also has access to transcripts of all Micode's videos, which are used to provide context and support for the chatbot's responses.
- Engage in a conversation with a chatbot that emulates Micode's communication style.
- Receive focused, practical, and direct IT info/advices.
- Access relevant snippets from transcripts of Micode's videos to support the chatbot's responses.
- Utilize semantic search to find relevant content from the transcripts.
- Python 3.7 or higher
- OpenAI API key
- Pinecone API key and environment details
- Clone the repository:
git clone https://github.com/your-repo-url/MicodeGPT.git
- Change to the project directory:
cd MicodeGPT
- Install the required dependencies:
pip install -r requirements.txt
- Set up the environment variables:
OPENAI_API_KEY
: Your OpenAI API keyPINECONE_API_KEY
: Your Pinecone API keyPINECONE_ENVIRONMENT
: Your Pinecone environment detailsPINECONE_INDEX_NAME
: Your Pinecone index detailsPINECONE_ENDPOINT
: Your Pinecone endpoint
- Run the Streamlit app:
streamlit run app.py
- Open the app in your web browser and enter your prompt to start the conversation with the chatbot. By default, Streamlit app will run on PORT 8501:
- Local URL: http://localhost:8501
- Network URL: http://26.26.26.1:8501
If you have any questions or need assistance, please feel free to reach out to me directly. You can contact me via email.
- Micode & the Micorp Team for creating valuable IT videos ! (Please, don't sue me ๐ฑ)
- OpenAI for their language models and embeddings.
- AssemblyAI for providing a great Whisper API DX.
- Langchain for their text processing tools.
- Pinecone for their semantic search capabilities.
- Streamlit & its amazing community for developing such great framework.
- Python for being amazing โค๏ธ