AMD's vLLM-ATOM Plugin Supercharges DeepSeek-R1, Kimi-K2, and gpt-oss-120B AI LLM Inference on Instinct MI350 and MI400 Accelerators
… The vLLM-ATOM architecture is broken down into three layers: In terms of model support, the vLLM-ATOM plugin supports both AI LLMs and VLMs through a unified serving pipeline. …