Who contributed to Hugging Face on July 13, 2026?

4 developers shipped this update, including githubnemo, Fabien, 3outeille, and Tarek Ziade.

What were the notable Hugging Face updates?

Shard on read Dtensor aware, Shard on read Dtensor aware (#46717), and Add reachy_mini mock-deps registry entry (#807).

@huggingface

Transformers, Datasets, and the open AI-model layer

github ↗

Pick a date

Topics: Python AI / ML Full archive →

The Wire · Showcase

TRANSFORMERS CUTS DISTRIBUTED LOAD TIME IN HALF WITH SHARD-ON-READ

By RepoJournal · Filed 06:58 UTC on July 13, 2026 · About Hugging Face

4 people shipped this

githubnemo @githubnemo 2 cited

Fabien @FabienDanieau 2 cited

3outeille @3outeille 2 cited

Tarek Ziade @tarekziade 1 cited

Transformers shipped native DTensor shard-on-read support, letting each rank load its own checkpoint slice instead of materializing the full model on every device first.

The big win here is inference speed on distributed setups. Instead of every rank reading the entire checkpoint and then sharding it down, each rank now reads only its slice directly [1] [2]. This cuts checkpoint load time roughly in half on multi-GPU and multi-node deployments. The PR also migrates FSDP2 wrapping to a dedicated distributed module while keeping backward compatibility [2], cleaning up the integration layer for anyone already on FSDP. On the docs side, doc-builder now handles the reachy_mini robotics library through mock deps instead of custom Docker images [3] [4], which means faster builds and easier maintenance. The PEFT repo fixed two quiet but annoying bugs: task switching was broken from leftover copy-paste code, and the UI was showing empty legend items in highlight mode [5] [6]. Transformers-ci is porting daily jobs to OTEL for better observability [7].

Action items

→ If you're using distributed training or inference on transformers, test shard-on-read on your next checkpoint load huggingface/transformers [plan]
→ Update FSDP imports to use transformers.distributed.fsdp if you're on custom wrapping huggingface/transformers [plan]
→ If you maintain docs on HF, check if your repo benefits from mock-deps registry entries huggingface/doc-builder [monitor]

References

[1] Shard on read Dtensor aware ↗ huggingface/transformers
[2] Shard on read Dtensor aware (#46717) huggingface/transformers
[3] Add reachy_mini mock-deps registry entry (#807) huggingface/doc-builder
[4] Add reachy_mini mock-deps registry entry ↗ huggingface/doc-builder
[5] Minor app embed improvements (#3322) huggingface/peft
[6] Minor app embed improvements ↗ huggingface/peft
[7] port over the daily ci jobs so we can OTEL them huggingface/transformers-ci

Quick answers

What shipped in Hugging Face on July 13, 2026?: Transformers shipped native DTensor shard-on-read support, letting each rank load its own checkpoint slice instead of materializing the full model on every device first. In total, 4 commits and 3 pull requests landed.
Who contributed to Hugging Face on July 13, 2026?: 4 developers shipped this update, including githubnemo, Fabien, 3outeille, and Tarek Ziade.
What were the notable Hugging Face updates?: Shard on read Dtensor aware, Shard on read Dtensor aware (#46717), and Add reachy_mini mock-deps registry entry (#807).

TRANSFORMERS-CI SHIPS INTELLIGENT TEST TRIAGE, CUTS LLM NOISE 100X

Hugging Face's serge verify loop now distills 100KB failure tracebacks down to the actual assertion, runs every test 5 times to catch flakiness, and dispatches each failure category with a custom instruction before spinning GPU or consuming a single LLM token.

python 45 shipped 2-min read

@huggingface 2 days ago

TRL CI BURNS, BITSANDBYTES PINNED BELOW 0.50.0

Hugging Face's TRL training library is firefighting two separate CI failures overnight, forcing temporary pins on bitsandbytes and xfailing critical tests against a broken transformers dev build.

python 7 shipped 1-min read

@huggingface 3 days ago

TRANSFORMERS OVERHAULS LINEAR ATTENTION WHILE DEPRECATING LEGACY RESPONSE SCHEMA

The transformers library is retiring its fragile response_schema prototype in favor of streaming-compatible parsing, while simultaneously refactoring every linear attention model to use standardized convolution patterns.

python 70 shipped 2-min read

@huggingface 4 days ago

TRANSFORMERS SHIPS FSDP DISTRIBUTED TRAINING STACK, HUB LIBRARY PLUGS REDOS HOLE

Hugging Face landed distributed training orchestration in transformers while plugging a regex vulnerability that could stall untrusted card parsing for minutes.

+10

python 61 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.