AION
Open-source releaseTraining & Scaling · Hardware & Compute · MLOps, Tooling & Infrastructure1 source · Mar 23, 2026

pytorch/pytorch v2.11.0: PyTorch 2.11.0 Release

<strong>FlexAttention</strong> now has a <strong>FlashAttention-4</strong> backend on <strong>Hopper</strong> and <strong>Blackwell</strong> GPUs

PyTorchAppleNVIDIA BlackwellFlashAttention
← Back to brief

Key points

  • Added Support for <strong>Differentiable Collectives</strong> for Distributed Training

Sources (1)

  • [1]pytorch/pytorch v2.11.0: PyTorch 2.11.0 Release
    GitHub: pytorch/pytorch · Mar 23, 06:38 PM
    <strong>FlexAttention</strong> now has a <strong>FlashAttention-4</strong> backend on <strong>Hopper</strong> and <strong>Blackwell</strong> GPUs
    Added Support for <strong>Differentiable Collectives</strong> for Distributed Training

Extractive summary: sentences quoted from the sources.