StartLux Open-Sources Prism, an Inference Engine Built for Decision Models With Bit-Exact Outputs

T2Covered by 1 source · first reported by China: SCMP Tech / TechNode / Pandaily / KrASIA / Sixth Tone at 24 Oct, 02:00 UTC

StartLux open-sourced Prism, an inference engine rebuilt for decision models. On one H200, it reports up to 5.3x faster multi-question runs, 41 ms average latency and bit-identical outputs.

ai

Sources

• T2StartLux Open-Sources Prism, an Inference Engine Built for Decision Models With Bit-Exact OutputsChina: SCMP Tech / TechNode / Pandaily / KrASIA / Sixth Tone24 Oct, 02:00