Technical Work
This section separates technical background from user task guides. Use the Inference and Fine-Tuning sections when you want to run a model. Use Technical Work when you want to understand the system ideas, implementation direction, or public talks behind KTransformers.
Topics
| Topic | Page |
|---|---|
| CPU-GPU heterogeneous MoE inference | Heterogeneous Inference |
| Workstation LoRA SFT direction | Local Fine-Tuning |
| Public talks and slide decks | Talks and Slides |
| GitHub docs reading map | GitHub Docs Reading Map |
Technical pages may discuss ideas before they become polished product workflows. To run a model directly, prefer pages that include package versions, commands, hardware configurations, and validation notes.