memory-mapping
A complete inference runtime for open-weight large language models, enabling efficient execution through streaming weights, quantization, and memory-aware scheduling
