Mapping GPUs to LLMs (and back): A bandwidth-based estimator for local inferencelocalllm-advisor.com2 points·apignotti··0 commentsOpen articleSaveView on HN