Issues

• ## [Proposal] Expand GPU Support for DeepInfer + RedKnot: H100, H200, and H800

  rednote-machine-learning/RedKnot#28 · Filter by author windystirfry opened on Oct 28, 2026

• [RFC] Reduce MoE computation through head-aware token selection and cached-state reuse

  rednote-machine-learning/RedKnot#27 · Filter by author aarav-infra-Oo opened on Oct 26, 2026

• [Feature Request] Add DeepInfer support for head-aware reuse and efficient long-context serving

  rednote-machine-learning/RedKnot#26 · Filter by author wickedatten opened on Oct 26, 2026

• [Feature Request] Introduce a hardware-agnostic runtime and backend layer

  rednote-machine-learning/RedKnot#23 · Filter by author Shutian-Luo opened on Oct 23, 2026

• How does RedKnot handle cumulative errors caused by MoE routing changes after KV-cache concatenation?

  rednote-machine-learning/RedKnot#21 · Filter by author Jaehong-Min-prog opened on Oct 22, 2026

• [Feature Request] Support plan for Qwen3.8-27B?

  rednote-machine-learning/RedKnot#20 · Filter by author skeleton-bear opened on Oct 22, 2026