PyTorch 2.0 released (torch.compile in beta)
Released PyTorch 2.0 (torch.compile in beta) — works on 93% of 165 open-source models, 36% faster on average with AMP. Adopted Triton as the GPU backend
Why it matters
Compiler-based optimization became a framework default
Source — Primary sources first; secondary reports are labelled “Press”
Officialpytorch.orghttps://pytorch.org/blog/pytorch-2.0-release/Added in Week 35, 2026 · See that week