Who contributed to Hugging Face on July 1, 2026?

4 developers shipped this update, including IlyasMoutawwakil, douglas-reid, qgallouedec, and s1lent4gnt.

What were the notable Hugging Face updates?

🚨 Modeling changes for export, compile, and hybrid-attention standardization, 🚨 [gemma 3/4] Fix bidirectional attention masking crossing sliding window boundaries, and Drop vLLM 0.14 support (#6209).

@huggingface

Transformers, Datasets, and the open AI-model layer

github ↗

Pick a date

Topics: Python AI / ML Full archive →

The Wire · Showcase

TRANSFORMERS SHIPPING EXPORT STANDARDIZATION, GEMMA ATTENTION FIXES LAND

By RepoJournal · Filed 06:04 UTC on July 1, 2026 · About Hugging Face

4 people shipped this

IlyasMoutawwakil @IlyasMoutawwakil 1 cited

douglas-reid @douglas-reid 1 cited

qgallouedec @qgallouedec 1 cited

s1lent4gnt @s1lent4gnt 1 cited

Transformers just locked in broad modeling changes that make ONNX, torch.export, and ExecuTorch export work cleanly across a huge set of models, while Gemma 3/4 gets critical attention masking corrections.

The modeling standardization PR [1] reshapes how models handle compilation and export paths, eliminating one of the sharpest pain points for production deployments. That lands alongside targeted fixes for Gemma 3/4 attention masking at sliding window boundaries [2], which patches a behavioral drift in local layer attention that would have broken inference consistency. TRL is also housekeeping: dropping vLLM 0.14 support [3], removing the defunct sft_video_llm.py script [4], and integrating the new response parsing API [5]. Over in robotics, LeRobot shipped configurable MIT control mode for ReBot [6], letting users swap between position-velocity and torque-based control per joint with tunable stiffness parameters. The physics-intern-skills repo continues internal iteration on Codex integration [7], workspace bootstrap flow [8], and plugin distribution [9], with documentation catching up to match.

Action items

→ Review transformers modeling export changes for your deployment targets (ONNX/torch.export/ExecuTorch) huggingface/transformers [plan]
→ If you're running Gemma 3/4 in production, pull the attention masking fix before next inference deployment huggingface/transformers [immediate]
→ TRL users on vLLM 0.14: upgrade vLLM before next training run huggingface/trl [plan]

References

[1] 🚨 Modeling changes for export, compile, and hybrid-attention standardization ↗ huggingface/transformers
[2] 🚨 [gemma 3/4] Fix bidirectional attention masking crossing sliding window boundaries ↗ huggingface/transformers
[3] Drop vLLM 0.14 support (#6209) huggingface/trl
[4] Remove sft_video_llm.py script (#6193) huggingface/trl
[5] Integrate the new response parsing API ↗ huggingface/trl
[6] Feat(robot): add MIT control mode to ReBot ↗ huggingface/lerobot
[7] Codex host: migrate sub-agent glue to auto-discovered .codex/agents/ huggingface/physics-intern-skills
[8] Drop the problem one-liner / start-research bootstrap step huggingface/physics-intern-skills
[9] Add Codex CLI plugin: $init-physics-intern + build-codex-plugin.sh huggingface/physics-intern-skills

Quick answers

What shipped in Hugging Face on July 1, 2026?: Transformers just locked in broad modeling changes that make ONNX, torch.export, and ExecuTorch export work cleanly across a huge set of models, while Gemma 3/4 gets critical attention masking corrections. In total, 34 commits and 24 pull requests landed.
Who contributed to Hugging Face on July 1, 2026?: 4 developers shipped this update, including IlyasMoutawwakil, douglas-reid, qgallouedec, and s1lent4gnt.
What were the notable Hugging Face updates?: 🚨 Modeling changes for export, compile, and hybrid-attention standardization, 🚨 [gemma 3/4] Fix bidirectional attention masking crossing sliding window boundaries, and Drop vLLM 0.14 support (#6209).

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.