Who contributed to PyTorch on July 14, 2026?

7 developers shipped this update, including Tristan Rice, Frederik Gossen, Limin Tang, Shangdi Yu, vmoens, theap06, and Benson Ma.

What were the notable PyTorch updates?

[inductor] Rewind block_ptr across outer loop in nested reductions (#188771) (#188771), [c10d] Use _coalescing_manager for abort instead of hardcoded NCCL group calls (#189770), and [Feature] Unify TorchRL checkpointing.

@pytorch

PyTorch and the broader machine-learning ecosystem

github ↗

Pick a date

Topics: Python AI / ML Full archive →

The Wire · Showcase

INDUCTOR FIXES NESTED REDUCTION BLOCK POINTER BUG, TORCHRL UNIFIES CHECKPOINTING

By RepoJournal · Filed 06:59 UTC on July 14, 2026 · About PyTorch

7 people shipped this

Tristan Rice @d4l3k 2 cited

vmoens @vmoens 2 cited

Limin Tang @limintang 1 cited

Shangdi Yu @yushangdi 1 cited

Frederik Gossen @frgossen

theap06 @theap06

Benson Ma @q10

Inductor's Triton codegen was silently corrupting fused broadcasts and reductions by losing track of block pointer advancement across outer loops, now patched [ref:3].

This wasn't a rare edge case: any `mean(a * b)` style operation where shapes broadcast then reduce would have produced wrong numerics in compiled kernels [1]. The fix tracks per-level pointer advancement correctly through nested loop structures. On the distributed training front, c10d's abort handler got a needed refactor to use the generic coalescing_manager instead of hardcoded NCCL calls [2], making it work cleanly across all backends. Meanwhile, TorchRL shipped a massive checkpointing unification [3] that consolidates legacy versioned formats into one manifest-driven API supporting both directory and ZIP containers, atomic replacement, and partial loading across all eight RL algorithms. The tensordict team landed single-file zip archives for memory-mapped tensors [4], letting you ship memmap data as a single file without a new format. Inductor also got smarter about int32 versus int64 indexing by analyzing actual addressing expressions instead of guessing from buffer sizes [5]. Three additional Inductor and c10d polish commits landed for NCCL review feedback and nested reduction bookkeeping [6].

Action items

→ Recompile any kernels using fused broadcast-reduction patterns to pick up the Inductor block pointer fix pytorch/pytorch [immediate]
→ Upgrade TorchRL if using checkpointing; the new unified API is production-ready and handles version migration automatically pytorch/rl [plan]
→ Monitor FBGEMM group_index_select backward on ROCm after the correctness and fake kernel fixes landed pytorch/FBGEMM [monitor]

References

[1] [inductor] Rewind block_ptr across outer loop in nested reductions (#188771) (#188771) pytorch/pytorch
[2] [c10d] Use _coalescing_manager for abort instead of hardcoded NCCL group calls (#189770) pytorch/pytorch
[3] [Feature] Unify TorchRL checkpointing ↗ pytorch/rl
[4] [Feature] Single-file zip archives for memory-mapped TensorDicts ↗ pytorch/tensordict
[5] [Inductor] Bound exact index expressions for int32 indexing eligibility (#189108) pytorch/pytorch
[6] [c10d][nccl2] Address stack review comments (#188701) pytorch/pytorch

Quick answers

What shipped in PyTorch on July 14, 2026?: Inductor's Triton codegen was silently corrupting fused broadcasts and reductions by losing track of block pointer advancement across outer loops, now patched [ref:3]. In total, 71 commits and 6 pull requests landed.
Who contributed to PyTorch on July 14, 2026?: 7 developers shipped this update, including Tristan Rice, Frederik Gossen, Limin Tang, Shangdi Yu, vmoens, theap06, and Benson Ma.
What were the notable PyTorch updates?: [inductor] Rewind block_ptr across outer loop in nested reductions (#188771) (#188771), [c10d] Use _coalescing_manager for abort instead of hardcoded NCCL group calls (#189770), and [Feature] Unify TorchRL checkpointing.

PYTORCH FIXES CRITICAL MPS MATH BUG WHILE TAMING CI GHOSTS

PyTorch's Metal backend shipped an accurate erfc function that eliminates 100% relative error on upper tail computations, fixing a cascade of broken numerics in GELU and special functions.

python 14 shipped 1-min read

@pytorch 2 days ago

PYTORCH FIXES H100 FLASH ATTENTION CI BREAKAGE, HELION CONSOLIDATES PALLAS LAUNCHERS

PyTorch patched a critical H100 test failure caused by upstream flash-attention directory churn, while Helion merged a major Pallas launcher unification that simplifies the runtime API.

python 20 shipped 1-min read

@pytorch 3 days ago

DYNAMO FIXES ITERATOR CORRUPTION AND FAKE INT ARITHMETIC

Dynamo's deque iteration was silently failing to detect mutations and reversing into the wrong iterator type, while custom hash functions using id() arithmetic crashed the compiler.

python 56 shipped 1-min read

@pytorch 4 days ago

TORCHRL CHECKPOINTS NOW MEMORY-MAPPED BY DEFAULT, FIXING WINDOWS FILE-LOCK CORRUPTION

TorchRL's trainer checkpoint loader was materializing entire files into non-evictable RAM and aliasing external state into live components, causing permission errors on Windows and potential corruption if checkpoints were rewritten externally.

python 104 shipped 2-min read

— or follow the whole beat:

Elsewhere on the wire

@cachyos about 4 hours ago

CACHYOS LINUX 7.2-RC3 SHIPS WITH HARDWARE FIXES AND EXPANDED LANGUAGE SUPPORT

CachyOS Linux 7.2-rc3-1 is live with Razer Blade speaker fixes and v4l2loopback updates, while the package installer now speaks Turkish and Azerbaijani.

infra 63 shipped 1-min read

Elixir & Phoenix about 4 hours ago

PHOENIX LIVE VIEW PATCHES SCHEME VALIDATION BYPASS

Phoenix LiveView 1.2.7 ships an emergency security fix for a scheme validation bypass that could let attackers inject malicious URLs into your application.

elixir 6 shipped 1-min read

@dotnet about 4 hours ago

RUNTIME FIXES DECIMAL AND BIGINTEGER ROUNDING BUGS THAT LOST PRECISION

Three critical numeric conversion bugs that silently corrupted precision in decimal, BigInteger, and floating-point math are now fixed across the stack.

dotnet 122 shipped 1-min read

@spring-projects about 4 hours ago

SPRING BOOT PLUGS FILE HANDLE LEAK IN IMAGE BUILDPACK

Spring Boot's image buildpack was leaking file handles and orphaning temp files when resolving missing builder images, but a fix just landed to clean up both the streams and the intermediate artifacts.

java 16 shipped 1-min read