Introduction
Ask-Irfan is a chat/llm engine with Generative, Predictive, and Transformer capabilities and focus on efficient architecture to harness the real potential of AI.
Ask-Irfan is a large language model that allows you to learn a topic, understand universe, how the things work. It can help you schedule your tasks, write emails, strategize to develop your professional activity and assist you with your business activities.
Technical Details
- What is Ask-Irfan?: Ask-Irfan is a conversational assistance based upon GPT architecure capabale of answering your questions from its knowledge base.
- What problem does it solve: Its a combination of AI model fine-tuned to respond to knowledge related questions, as if it is kind of a quick reference which can be accessed through conversation.
- Architecture: Basic model inherits the llama architecture. It's enhansed with System level prompt, fine-tuning, RAG and other technologies. We are evaluating the architectural changes for increasing the performance and usability of the model.
- Models: A combination of models has been used for the framework of Ask-Irfan. The question is analysed for the selection of an Expert. The expert or successive selection of experts are kept informed through a mechanism of context and persistant memory.
- Technologies: Ask-Irfan exploits the technologies like: terminal based UI, API connectivity, MCP, Orchestrating, Contextualization (Hybrid), Persistant VectorMemory or KVM management, Optimization before and after Token generation, RAG, Harnessing, Gaurdrails, Judgemnt, Bias detection, Decision making, Notification of inapropriate prompting, Storage to VectorDB and logging for traceability and audit and finally streaming the response to the user.
Ask-Irfan has its own professional space on the internet. Go to Ask-Irfan