Real-time LLM Inference on Standard GPUs: 3k tokens/s per request

(blog.kog.ai)

110 points | by NicoConstant 5 hours ago ago

54 comments