Kimi K3 Draft Collection: three open-sourced draft models trained with TorchSpec and vLLM on GB200
AccBalanced · x · 2026-09-04
lightseekorg released the Kimi K3 Draft Collection — three draft models (EAGLE-3, DFlash2, DSpark) trained with TorchSpec and vLLM on NVIDIA GB200, with data recipes open sourced. The vLLM team highlighted it as a model example of training-serving collaboration. These speculative-decoding draft models can accelerate K3 inference, and the released recipes let the community reproduce similar work.
Related event: Three Open-Source Draft Models for Kimi K3 Released(3 posts)→
More from Infra
- CXMT's 3D DRAM roadmap: risk production in 2028, mass production in 2029 — zephyr_z9 · 2026-09-04
- Musk calls for a 'construction army' to staff xAI's Memphis supercomputer buildout — elonmusk · 2026-09-04
- Salesforce research: Random eviction of reasoning tokens matches selective KV cache compression — Salesforce · 2026-09-04
- Red Teamer Seeks Advice on Local LLM Harness for Authorized Security Exercises — CivilBug4007 · 2026-09-04
- Musk: winning AI means owning both chips and models — 'we must win here' — r0ck3t23 · 2026-09-04
- Gimlet Labs raises $300M at $3B valuation to split AI workloads across chip types — dinabass · 2026-09-04