ExLlamaV3

ExLlamaV3: High-Performance LLM Inference Engine

Running large language models on consumer hardware requires efficient inference engines that squeeze every drop of performance from available GPU...