AMD GPUs speed up AI text generation with speculative decoding

Original: Speculative Decoding in vLLM on AMD GPUs

Source: vllm.ai ↗