Serving Open Source Models 4x faster than vLLM by quantizing with ~no tradeoffsblog.fireworks.ai1 point·kmdupree··0 commentsOpen articleSaveView on HN