A curated reading list for DeltaNet, FlashKDA, vLLM serving and MoE

austinvhuang · x · 2026-08-04

A curated list of explanations and deep dives for several AI engineering and research topics, including Songlin Yang’s design intuition posts, a chunkwise DeltaNet algorithm section, FlashKDA, vLLM serving, AttnRes training/inference explanations, and Jianlin Su’s MoE series.

The post also points to newer work such as Kimi Linear, LatentMoE, and the Kimi K3 technical report, and notes that prompting Codex to explain FlashKDA made the deep dive finally click.

Original post →

More from coding & agent

coding & agent channel →