Who contributed to PyTorch on July 2, 2026?

5 developers shipped this update, including SaoirseARM, Sebastian-Larsson, irtrukhina, calebmkim, and atalman.

What were the notable PyTorch updates?

Fix misleading error for complex order in torch.linalg.cond (#188591), [CD] Refactor macOS wheel build scripts as a Python pipeline (#187944), and Arm backend: Add TOSA dialect unary elementwise node visitors.

@pytorch

PyTorch and the broader machine-learning ecosystem

github ↗

Pick a date

Topics: Python AI / ML Full archive →

The Wire · Showcase

PYTORCH FIXES CRYPTIC LINALG ERROR, ACCELERATES EXECUTORCH BACKENDS

By RepoJournal · Filed 06:03 UTC on July 2, 2026 · About PyTorch

5 people shipped this

calebmkim @calebmkim 3 cited

SaoirseARM @SaoirseARM 1 cited

Sebastian-Larsson @Sebastian-Larsson 1 cited

irtrukhina @irtrukhina 1 cited

atalman @atalman 1 cited

PyTorch shipped a fix for a misleading error in torch.linalg.cond that now properly rejects complex-valued norm orders [ref:1], while ExecutorTorch's Arm backend gained TOSA dialect support and optimized lowering passes to cut graph rebuild overhead [ref:6] [ref:7].

The linalg fix [1] replaces a confusing overflow error with a clear message when developers pass complex values to cond's order parameter, fixing a long-standing issue where the actual problem was buried under type conversion noise. Separately, ExecutorTorch shipped three consecutive wins: TOSA unary elementwise operators [3] expand the Arm dialect coverage, smart pass-skipping [4] prevents rebuilding graphs when no relevant nodes exist (cutting placeholder, rank, and decomposition pass overhead), and MLPerfTiny profiling tests [5] validate real-world model performance. The PyTorch build pipeline also tightened: macOS wheel builds now follow the same Python pipeline structure as Linux and Windows [2], unifying CD across all platforms, while test-infra added ClickHouse alerting for nightly binary failures [9] [10]. Helion's autotuner entered stage 2, composing matmul facts with reduction epilogues [6] and specializing backward kernel block sizes via occupancy tracking [7], after stage 1 tightened the reduction seed heuristic [8].

Action items

→ Review the macOS CD pipeline refactor if you maintain cross-platform wheel builds pytorch/pytorch [plan]
→ Monitor ExecutorTech Arm backend PRs if you deploy models to edge devices pytorch/executorch [monitor]
→ Watch Helion's backward kernel autotuner improvements for LLM inference gains pytorch/helion [monitor]

References

[1] Fix misleading error for complex order in torch.linalg.cond (#188591) pytorch/pytorch
[2] [CD] Refactor macOS wheel build scripts as a Python pipeline (#187944) pytorch/pytorch
[3] Arm backend: Add TOSA dialect unary elementwise node visitors ↗ pytorch/executorch
[4] Arm backend: Skip irrelevant lowering passes ↗ pytorch/executorch
[5] [EIEX-947] Add profiling tests for MLPerfTiny models ↗ pytorch/executorch
[6] [autotuner] composed-fact seed for fused matmul + reduction-epilogue ↗ pytorch/helion
[7] [autotuner] specialize the M-reduction seeds via per_feature_accumulator (occupancy + byte-cap) ↗ pytorch/helion
[8] [autotuner] reduction seed: budgeted r_block + liveness-aware persistent/looped decision ↗ pytorch/helion
[9] Add ClickHouse query for nightly binary pipeline failure alert (#8231) pytorch/test-infra
[10] Add ClickHouse query for nightly binary pipeline failure alert ↗ pytorch/test-infra

Quick answers

What shipped in PyTorch on July 2, 2026?: PyTorch shipped a fix for a misleading error in torch.linalg.cond that now properly rejects complex-valued norm orders [ref:1], while ExecutorTorch's Arm backend gained TOSA dialect support and optimized lowering passes to cut graph rebuild overhead [ref:6] [ref:7]. In total, 83 commits and 34 pull requests landed.
Who contributed to PyTorch on July 2, 2026?: 5 developers shipped this update, including SaoirseARM, Sebastian-Larsson, irtrukhina, calebmkim, and atalman.
What were the notable PyTorch updates?: Fix misleading error for complex order in torch.linalg.cond (#188591), [CD] Refactor macOS wheel build scripts as a Python pipeline (#187944), and Arm backend: Add TOSA dialect unary elementwise node visitors.

EXECUTORCH ROUTES REASONING SEPARATELY, PYTORCH MODERNIZES COMPILER STACK

ExecutorTorch's shared OpenAI adapter now exposes optional reasoning content as a first-class output channel, while PyTorch core moves to clang-21 and strips dead C++ macros.

python 20 shipped 1-min read

@pytorch 2 days ago

DYNAMO FIXES GRAPH-BREAKING SDPA PARAMS, PAT SVD LEAK BREAKS DOWNSTREAM TESTS

PyTorch shipped fixes for two critical compilation and module patching bugs overnight that were silently breaking downstream workflows.

python 19 shipped 1-min read

@pytorch 3 days ago

CRITICAL OIDC INJECTION IN DOCS PREVIEW WORKFLOW PATCHED

PyTorch's docs-preview CI trusted fork-controlled artifacts in a context with token-write permissions, exposing the entire build pipeline to code injection.

python 66 shipped 1-min read

@pytorch 4 days ago

PYTORCH AUTOGRAD GETS 7% FASTER, AOTI FIXES SILENT FAILURES

Interned attribute names in autograd.Function shaved microseconds off the hot path while AOTI's scatter operations now properly report errors instead of silently corrupting results.

python 64 shipped 1-min read

Elsewhere on the wire

AI Agents about 21 hours ago

CLINE SHIPS FIVE CRITICAL AGENT FIXES AFTER END-TO-END TESTING BLITZ

Saoudrizwan just landed a cascade of fixes that prevent the agent loop from running wild, stop file reads from resolving to the wrong directory, and plug the compaction pipeline.

ai-agents 16 shipped 2-min read

Local LLMs about 21 hours ago

LLAMA.CPP SHIPS MINIMAX-M3 SPARSE ATTENTION, VLLM UNLOCKS HETEROGENEOUS PREFILL-DECODE

The inference stack gets a new model architecture and a fundamental shift in how prefill and decode can run on different hardware configurations.

+10

llms 125 shipped 1-min read

@CachyOS about 21 hours ago

GLIBC 2.43 REBUILD SHIPS, INSTALLER JUMPS TO C++26

CachyOS pushed glibc 2.43 through the pipeline while the CLI installer upgraded its C++ standard to C++26, signaling a major refresh across core infrastructure.

infra 97 shipped 1-min read

Elixir & Phoenix about 21 hours ago

ECTO FIXES SUBQUERY PLANNING IN JOINS, ELIXIR PATCHES IEX PARSING

Ecto's query planner now handles subqueries correctly in interpolated join filters, while Elixir fixes a syntax error that broke 'not in' as a continuation operator in IEx.