The Building Blocks of Today’s and Tomorrow’s Language Models - Sebastian Raschka, RAIR Lab

Description

The Building Blocks of Today’s and Tomorrow’s Language Models - Sebastian Raschka, RAIR Lab In this talk, you’ll learn about the latest trends in large language model (LLM) architectures. We'll look at how the architectural building blocks have evolved this year, with a focus on current transformer-based models (including Llama, GPT-OSS, Gemma, Qwen, and DeepSeek). The talk will also spotlight emerging non-transformer approaches that may signal what comes next for LLM research and development.