3 ms·Using the Lamborghini of inference engines for serverless Llama 31 points by birdculture 2y agognabgib 2y agoTitle: Serve an interactive language model app with latency-optimized TensorRT-LLM