PyTorch 2.13 Released: FlexAttention Hits Apple Silicon

PyTorch 2.13 has been officially released, introducing key performance optimizations. The update notably brings FlexAttention support to Apple Silicon and significantly reduces VRAM usage for large-vocabulary models by up to 4x.

2026-08-01 ~ 2026-08-01 · 2 related posts

• PyTorch 2.13 Brings FlexAttention to Apple Silicon, Cuts Peak Memory by 4× — PyTorch · 2026-08-01

1 near-duplicate retellings: PyTorch