A Docker Compose to run a local ChatGPT-like application using Ollama, Ollama Web UI, Mistral NeMo & DeepSeek R1.

Benchmarking Mistral NeMo for Italian Cultural Alignment using ITALIC benchmark

Psych_RAG is a Retrieval Augmented Generation chatbot that answers psychology-related questions using Mistral Nemo Instruct 2407, Pinecone for vector search, LangChain for integration, FastAPI for the backend, and Gradio for the interface.

GraphLex AI: A production-grade Agentic RAG for Zero-Shot Document Intelligence. Features Invariant Extraction, Unified Reasoning Nodes, and secure local LLM orchestration.

Fine-tuning the Mistral NeMo 12B Instruct model to transform natural language questions into executable SQL queries. We adopted a two-phase incremental learning strategy to build a robust model capable of handling both syntax and complex relational logic

This project is about making two AI models debate with each other. One AI is told to support a topic and the other AI is told to oppose it. They take turns replying to each other for a few rounds, like a real debate. The program keeps track of everything they say and saves the full debate to a text file.

Godot 4 custom build to run Mistral-Nemo-Instruct-2407 and SDXL finetunes.

The Mistral-Nemo-12b model has been fine-tuned for text generation tasks. This fine-tuning was performed using the Unsloth optimization framework, which significantly accelerates the training process, achieving a 2x faster fine-tuning time compared to conventional methods.

AI-Driven OLED Technical Assistant - PhD-level RAG system