Sebastian Raschka releases PyTorch course on building reasoning models
October 26, 2026
This 11-hour technical course covers implementing reasoning models from scratch using PyTorch. It specifically addresses inference-time scaling, reinforcement learning, and distillation techniques starting from a pre-trained LLM.
HOW THIS AFFECTS YOU
●
builderYou can use these implementations to understand how to integrate reasoning capabilities into your own architectures.
●
researcherThis provides a concrete codebase for studying the interplay between RL and inference scaling.