STEVEN AI

StevenAI is a personalized AI chatbot built to answer any questions me. It was originally developed by fine-tuning LLaMA 3.2 with LoRA on a custom dataset of 1,000+ Q&A pairs. However, while this method provided deep personalization, it also surfaced limitations in accuracy, especially when handling rare or nuanced queries.

To address these challenges, StevenAI was upgraded with a retrieval-augmented generation (RAG) layer, which supplements the model's responses with relevant facts from a structured knowledge base. This hybrid architecture blends the strengths of fine-tuning with real-time semantic retrieval, resulting in more accurate, flexible, and context-aware answers.

The system supports multiple base models and retrieval strategies. You can select your preferred combination and ASK A QUESTION NOW!