feat: run:ai gpu_direct model loading with quantization-aware device maps - #635

feat: run:ai gpu_direct model loading with quantization-aware device maps#635

khaiwang wants to merge 5 commits into

devndif-team/nnsight:devfrom

zikai/runai-gpu-directndif-team/nnsight:zikai/runai-gpu-directCopy head branch name to clipboard