Preserve per-projection QKV precision in PyTorch quantizers - #2700

Preserve per-projection QKV precision in PyTorch quantizers#2700

Xiaoyu Z (xiaoyu-work) merged 9 commits into

mainmicrosoft/Olive:mainfrom

titaiwang/independent-qkv-quantizationmicrosoft/Olive:titaiwang/independent-qkv-quantizationCopy head branch name to clipboard