Evolution of LLMs and AI Agent: From Transformer to DeepSeek-R1 to Manus AI
Date:
In this talk, I present an overview of the evolving landscape of Large Language Models (LLMs), tracing their progression from the introduction of the Transformer in 2017 to the latest developments like DeepSeek-R1 in 2025. The talk reflects on this evolution, particularly in terms of the types of training datasets and approaches, including pretraining, Supervised/Instruction Fine-Tuning (SFT), and Reinforcement Learning from Human Feedback (RLHF). It also delves into the rise of Agentic AI during 2025, featuring key technologies such as Manus AI and MCP. The presentation includes hands-on demonstrations showcasing the creation of basic AI agents using GPTs, Claude+MCP, and n8n.
