Welcome to the MNLobago repository, which houses the EcoWise chatbot—a project designed to provide users with valuable insights and information on environmental sustainability and climate change. The EcoWise chatbot aims to educate and engage users on pressing environmental issues, promoting awareness and fostering sustainable practices.
-
EcoWise
Code and materials related to the EcoWise chatbot. -
ChatBotEvaluation
Scripts and documentation for assessing the chatbot's performance and effectiveness. -
Data_Collection_Creation
Tools and resources for collecting and creating datasets for training and fine-tuning the chatbot. -
Data_Formats_CSVs
Formatted CSV files ready for use in Phase 1 of the project. -
Data_Quality
Documentation of data quality checks to ensure high-quality datasets. -
Gemma_ModelFineTuning
Scripts for fine-tuning the Gemma-2B language model for environmental topic responses. -
Project Workflow
A diagram illustrating the project's workflow, updated on 2024-11-05. -
App
This application contains the code for deploying my chatbot on Hugging Face Spaces using Gradio.
After determining that Variant 2 was the best-performing model, I took the following steps to deploy it:
- Model Selection: Selected Variant 2 as the optimal model based on performance metrics.
- Hugging Face Spaces: Registered for a Hugging Face account to access the platform's features.
- Integration with Gradio:
- Utilized Gradio to create an interactive web interface for the chatbot.
- The Gradio library allows for easy integration and deployment of machine learning models with minimal code.
- Deployment to Free CPU:
- Deployed the chatbot on Hugging Face Spaces, choosing the free CPU option to minimize costs while testing the deployment.
- Leveraged Hugging Face’s seamless deployment process to quickly bring the chatbot online.
Feel free to explore the code and documentation included in this repository to better understand the deployment process and the functionality of the chatbot.