MindMenders/Mental_Health_Chatbot

Mental Health AI Chatbot

Overview

This AI-powered chatbot provides emotional support and mental health guidance through context-aware conversations. It is fine-tuned on mental health counseling conversations to ensure compassionate and relevant responses.

Features

• Emotion Detection & Sentiment Analysis → Adjusts responses based on detected user emotions.
• Context-Aware Conversations → Remembers past interactions for coherent discussions.
• Journal Mood Pattern Analysis → Tracks mood trends over time.
• Moderation & Toxicity Filtering → Ensures safe and ethical interactions.
• Emergency Contact Redirection → Provides immediate help when distress signals are detected.
• Optimized Model Performance → Uses LoRA and 4-bit quantization for efficiency.
• Hosted on RTX 4090 → Faster checkpoint downloads and improved inference speed.

Model Training Details

• Base Model → Fine-tuned LLaMA-2 for mental health counseling.
• Training Framework → PyTorch, Hugging Face Transformers.
• Optimization Techniques → LoRA, 4-bit quantization.
• Dataset Used → Custom dataset of mental health counseling conversations.
• Training Environment → Google Colab with GPU support & RTX 4090 desktop for inference acceleration.

How to Run on Google Colab

• Open Google Colab: Go to Google Colab

  • Upload the Notebook:
  • Click on File → Upload Notebook
  • Select GoatedModel_MHB.ipynb from your local storage. (It can be downloaded from the 'Files and Versions' tab of our Hugging Face repo.)

• Enable GPU:

  • Click on Runtime → Change runtime type
  • Set Hardware accelerator to GPU.

• Run the Notebook:

  • Execute all cells sequentially.
  • The model will be downloaded, set up, and ready for interaction.

Usage

• The chatbot runs interactively in the Colab notebook.
• Users can enter queries, and the model will generate responses accordingly.
• Outputs can be analyzed for emotional insights and response quality.

Dependencies

• The notebook automatically installs required dependencies, but ensure the following packages are installed if needed:
• !pip install torch transformers unsloth langchain llama-index

Future Enhancements

• Adding voice-based interactions.
• Integrating mindfulness and relaxation exercises.
• AI-powered therapy mode selection (CBT, mindfulness, etc.).

Notes

• This chatbot is not a substitute for professional mental health support.
• Ensure ethical usage and refrain from using the bot for harmful purposes.

Repository Links

• Hugging Face Model → [https://huggingface.co/MindMenders/Mental_Health_Chatbot]

Acknowledgments

• Built with Hugging Face Transformers and PyTorch.
• Developed by Siddharth Amdal/Swapnil Patidar with optimization for efficient AI-based counseling.

---

For any issues or contributions, feel free to open a pull request or contact us!