PyTorch 2.13.0 Release

PyTorch 2.13.0 Release pytorch / pytorch Public Notifications You must be signed in to change notification settings Fork 30.5k Star 103k PyTorch 2.13.0 Release angelayi released this 08 Jul 17:39 · 5583 commits to main since this release v2.13.0 cf30153 This commit was created on GitHub.com and signed with GitHub’s verified signature . GPG key ID: B5690EEEBB952194 Verified Learn about vigilant mode .

This HardwareLaunch is relevant to the technology intelligence record because it involves Apple, Intel, GitHub, Meta. The source article should remain the factual reference for follow-up coverage.

• pytorch / pytorch Public Notifications You must be signed in to change notification settings Fork 30.5k Star 103k PyTorch 2.13.0 Release angelayi released this 08 Jul 17:39 · 5583 commits to main since this release v2.13.0 cf30153 This commit was created on GitHub.com and signed with GitHub’s verified signature .
• GPG key ID: B5690EEEBB952194 Verified Learn about vigilant mode .
• PyTorch 2.13.0 Release Notes Highlights Backwards Incompatible Changes Deprecations New Features Improvements Bug fixes Performance Documentation Developers Highlights FlexAttention lands on Apple Silicon (MPS), with up to ~12x speedup over SDPA on sparse patterns, and gains a deterministic backward path on CUDA for reproducible gradient computation.
• CuTeDSL "Native DSL" backend gives Inductor a second high-performance code path (alongside Triton) for key GPU operations, with faster compilation.
• [Prototype] nn.LinearCrossEntropyLoss combines the final prediction and loss computation to cut peak GPU memory by up to 4x for large-vocabulary language model training.
• torchcomms , a new communications backend for PyTorch Distributed, improves fault tolerance, scalability, and debuggability for large-cluster training.