RepoJournal
PyTorch

@pytorch

PyTorch and the broader machine-learning ecosystem

Keep up with PyTorch in about 3 minutes a day: what actually shipped — the commits, pull requests, releases, and security advisories that matter.

One email a day. Unsubscribe in one click.

Pick a date

  1. Sun 14 jun
  2. Mon 15 jun
  3. Thu 18 jun
  4. Sat 27 jun
  5. Sun 28 jun
  6. Mon 29 jun
  7. Tue 30 jun
  8. Wed 1 jul
  9. Thu 2 jul
  10. Fri 3 jul
  11. Sat 4 jul
  12. Sun 5 jul
  13. Mon 6 jul
  14. Tue 7 jul
  15. Wed 8 jul
  16. Thu 9 jul
  17. Fri 10 jul
  18. Sat 11 jul
  19. Sun 12 jul
  20. Mon 13 jul
  21. Tue 14 jul
  22. Thu 16 jul
  23. Sat 18 jul
  24. Sun 19 jul
  25. Mon 20 jul
  26. Tue 21 jul
  27. Fri 24 jul
  28. Sat 25 jul
  29. Sun 26 jul
  30. Mon 27 jul
Topics: Python AI / ML Full archive →

The Wire · Showcase

TORCHTITAN SHIPS FLEXATTENTION INDUCTOR BOOST, GRAPH TRAINER UNLOCKS CPU OFFLOADING

By RepoJournal · Filed · About PyTorch

aditvenk mlazos shuhuayu pzhan9
4 people shipped this
aditvenk @aditvenk 1 cited mlazos @mlazos 1 cited shuhuayu @shuhuayu 1 cited pzhan9 @pzhan9 1 cited

FlexAttention now compiles through Inductor when using aot_eager backend, cutting Step 1 training loss mismatch significantly while the graph trainer gains view replay for CPU-offloaded activations.

The regional_inductor context manager [1] wraps FlexAttention ops to trigger Inductor compilation instead of falling back to eager, validated on RL workloads where Step 1 loss variance dropped measurably. This pairs with a major fix in graph_trainer [2] that replays view operations (transpose, reshape, permute) during backward, finally enabling CPU activation offloading for tensors whose consumers reach them through view chains. Qwen3.5 evolution [3] shipped with hybrid attention architecture (75% GatedDeltaNet linear + 25% full attention) and head-sharded TP on GatedDeltaNet projections, marking a significant architecture jump from Qwen3-VL. The RL infrastructure expanded with a GeneratorRouter [4] supporting round-robin and least-loaded routing across multiple generators for large-scale training, plus weight sync modes for hot-swap deployment. On the PyTorch core side, the build system fixed a critical bug in build_with_debinfo.py [5] that broke targeted debug builds with CONFIGURE_DEPENDS globbing, while Dynamo now serializes higher-order-op subgraphs correctly [6] so fx_graph_runnable repros work for cond/while_loop branches. Inductor's assertion removal [7] [8] continues hardening error handling across fx_passes.

One email a day. Unsubscribe in one click.

Action items

References

  1. [1] [RL] Enable regional_inductor in FlexAttention ↗ pytorch/torchtitan
  2. [2] [graph_trainer] Add view replay for CPU activation offloading ↗ pytorch/torchtitan
  3. [3] [qwen3_5] evolve qwen3_vl to qwen3_5 ↗ pytorch/torchtitan
  4. [4] Add a router for multiple generators ↗ pytorch/torchtitan
  5. [5] Fix build_with_debinfo.py broken by CONFIGURE_DEPENDS globbing (#186780) pytorch/pytorch
  6. [6] [dynamo] Serialize higher-order-op subgraphs in NNModuleToString.convert (#186804) pytorch/pytorch
  7. [7] remove plain assertions in remaining torch/_inductor top-level files (#186392) pytorch/pytorch
  8. [8] remove plain assertions in torch/_inductor/fx_passes (#186391) pytorch/pytorch

Quick answers

What shipped in PyTorch on June 11, 2026?
FlexAttention now compiles through Inductor when using aot_eager backend, cutting Step 1 training loss mismatch significantly while the graph trainer gains view replay for CPU-offloaded activations. In total, 99 commits and 40 pull requests landed.
Who contributed to PyTorch on June 11, 2026?
4 developers shipped this update, including aditvenk, mlazos, shuhuayu, and pzhan9.
What were the notable PyTorch updates?
[RL] Enable regional_inductor in FlexAttention, [graph_trainer] Add view replay for CPU activation offloading, and [qwen3_5] evolve qwen3_vl to qwen3_5.

More from @pytorch

PyTorch @pytorch

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.

Gasoonjia Nikita Shulga Richard Barnes PyTorch UpdateBot +3
python 20 shipped 1-min read
PyTorch @pytorch

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.

Guan-Ming (Wesley) Chiu Denis Boyda Jim Meyering drisspg +1
python 19 shipped 1-min read
PyTorch @pytorch

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.

atalman Cynthia Yang Michael Lazos Akshay Madavalappil Ramesh +5
python 66 shipped 1-min read
PyTorch @pytorch

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.

kurt739910017 Richard Zou Bin Bao PyTorch MergeBot +5
python 64 shipped 1-min read

Daily updates, in your inbox

Follow PyTorch

Keep up with PyTorch in about 3 minutes a day: what actually shipped — the commits, pull requests, releases, and security advisories that matter.

or

One email a day. Unsubscribe in one click. Read a past issue →

Elsewhere on the wire

AI Agents AI Agents

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.

langchain-oss-model-profiles langchain-oss-model-profiles[bot] saoudrizwan
ai-agents 16 shipped 2-min read
Local LLMs Local LLMs

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.

github-actions[bot] dhiltgen timkhronos Pascal +10
llms 125 shipped 1-min read
CachyOS @CachyOS

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.

Piotr Gorski github-actions Vladislav Nepogodin ptr1337 +7
infra 97 shipped 1-min read
Elixir & Phoenix Elixir & Phoenix

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.

Benjamin Milde Alexander Gubarev lukaszsamson makoto-developer
elixir 10 shipped 1-min read
Want every project, not just this one?

We use privacy-friendly analytics (Google Analytics, IP-anonymized) to see which pages help readers. No ads, and we never sell your data. See our Privacy Policy.