가입 후 초대 링크를 공유하면 동영상 재생 및 초대 보상을 받을 수 있습니다.

Crusoe
@CrusoeAI
The AI factory company. We are on a mission to accelerate the abundance of energy and intelligence.
가입 January 2019
373 팔로잉 중    19.9K 팬
LoRA freezes the base weights. So why store them in bf16? We keep them in mxfp4 in HBM and dequantize inside the kernel. The optimizer only touches the LoRA adapters. Up to 75% less base-weight memory, with no accuracy loss in our internal evals. ➕ The math:
더 보기