Cost-Efficient LLM Training with Lifetime-Aware Tensor Offloading via GPUDirect Storage

Open in new window