pytorch/pytorch v2.3.0: PyTorch 2.3: User-Defined Triton Kernels in torch.compile, Tensor Parallelism in Distributed
PyTorch 2.3 offers support for user-defined Triton kernels in torch.compile, allowing for users to migrate their own Triton kernels from eager without experiencing performance complications or graph breaks.
Key points
- We are excited to announce the release of PyTorch® 2.3!
- As well, Tensor Parallelism improves the experience for training Large Language Models using native PyTorch functions, which has been validated on training runs for 100B parameter models.
- This release is composed of 3393 commits and 426 contributors since PyTorch 2.2.
- As always, we encourage you to try these out and report any issues as we improve 2.3.
Sources (1)
- [1]pytorch/pytorch v2.3.0: PyTorch 2.3: User-Defined Triton Kernels in torch.compile, Tensor Parallelism in DistributedGitHub: pytorch/pytorch · Apr 24, 04:12 PM
PyTorch 2.3 offers support for user-defined Triton kernels in torch.compile, allowing for users to migrate their own Triton kernels from eager without experiencing performance complications or graph breaks.
We are excited to announce the release of PyTorch® 2.3!
Extractive summary: sentences quoted from the sources.