Red Hat AI(@RedHat_AI):Kimi K3 is a 2.8T-parameter model. We trained a DSpark speculator for it, and the speedup holds up. Single-stream math reasoning goes from ~110 to ~435 tokens/sec per user. Under concurrent load, up to ~3.5x higher output throughput at matched interactivity. The drafter is a 5B model proposing 8 tokens a step, and on math it gets ~6.4 of them accepted per round. Training a drafter for a model this big meant going multi-node: Speculators plus a new Mooncake connector streaming hidden states between vLLM inference and training over RDMA. Two GB300 nodes to serve the target, one to train.

Red Hat AI

@RedHat_AI

Accelerating AI innovation with open platforms and community. The future of AI is open.

参加 May 2018

2.1K フォロー中 12.6K ファン

Red Hat AI@RedHat_AI

2026.09.17 14:36

Kimi K3 is a 2.8T-parameter model. We trained a DSpark speculator for it, and the speedup holds up. Single-stream math reasoning goes from ~110 to ~435 tokens/sec per user. Under concurrent load, up to ~3.5x higher output throughput at matched interactivity. The drafter is a 5B model proposing 8 tokens a step, and on math it gets ~6.4 of them accepted per round. Training a drafter for a model this big meant going multi-node: Speculators plus a new Mooncake connector streaming hidden states between vLLM inference and training over RDMA. Two GB300 nodes to serve the target, one to train.