AI updates
PyTorch 2.5: cuDNN SDPA backend, FlexAttention, regional compilation
PyTorch 2.5 added a cuDNN backend for scaled dot-product attention (speedups by default on H100-class GPUs), regional compilation to cut torch.compile cold-start time, and a FlexAttention prototype.
- Regional compilation compiles one repeated block (like a Transformer layer) once, instead of the whole model.
- FlexAttention (prototype) lets you write custom attention variants that still compile to fast kernels.