Real-time LLM Inference on Standard GPUs (3k tokens/s per request)blog.kog.ai7 points·morgangiraud··0 commentsOpen articleSaveView on HN