Who contributed to Hugging Face on July 7, 2026?

3 developers shipped this update, including qgallouedec, albertvillanova, and jaygala223.

What were the notable Hugging Face updates?

Make doc builds faster (#47099), Remove the PAPO trainer, and Simplify tokenization [1/N]: Remove redundant is_vlm parameter (#6298).

@huggingface

Transformers, Datasets, and the open AI-model layer

github ↗

Pick a date

Topics: Python AI / ML Full archive →

The Wire · Showcase

TRANSFORMERS CUTS DOC BUILD TIME, TRL DUMPS ZERO-USE PAPO TRAINER

By RepoJournal · Filed 06:04 UTC on July 7, 2026 · About Hugging Face

3 people shipped this

qgallouedec @qgallouedec 1 cited

albertvillanova @albertvillanova 1 cited

jaygala223 @jaygala223 1 cited

Hugging Face shipped doc infrastructure speedups across transformers while quietly removing an experimental trainer that hasn't seen production use in six weeks.

The transformers team optimized their doc build workflow [1] by switching to light installs and dropping custom container overhead, directly addressing the velocity tax on their documentation pipeline. In parallel, TRL is killing the PAPO trainer [2] after discovering it had zero runs in the last six weeks despite shipping in October 2025; the trainer was burning maintenance cycles by subclassing GRPOTrainer and overriding large portions of logic that's been rewritten multiple times since. TRL is also simplifying its tokenization layer in a multi-part refactor [3][4], removing redundant is_vlm parameters and converting private _tokenize methods into module-level functions to reduce coupling and improve testability. Over in kernels, the team added a vendor-neutral Triton skill [5][6] that covers portable patterns for NVIDIA and AMD GPUs, filling a gap between raw CUDA and backend-specific implementations. LeRobot shipped v0.6.0 [7], a breaking change that decouples optional dependencies (add `[training]` for training support) and fixes import paths, with v0.6.1 already rolling out for hotfixes.

Action items

→ Review TRL tokenization refactor before next training run; is_vlm parameter removal may affect custom trainer subclasses huggingface/trl [plan]
→ Upgrade to LeRobot v0.6.1 if on 0.6.0; fix import paths to use canonical public entry points huggingface/lerobot [plan]
→ Adopt vendor-neutral Triton skill as base for custom kernel development across NVIDIA and AMD targets huggingface/kernels [monitor]

References

[1] Make doc builds faster (#47099) huggingface/transformers
[2] Remove the PAPO trainer ↗ huggingface/trl
[3] Simplify tokenization [1/N]: Remove redundant is_vlm parameter (#6298) huggingface/trl
[4] Simplify tokenization [2/N]: Make _tokenize a module-level function ↗ huggingface/trl
[5] triton-kernels: add vendor-neutral Triton skill ↗ huggingface/kernels
[6] triton-kernels: add vendor-neutral Triton skill (#675) huggingface/kernels
[7] Release v0.6.0 ↗ huggingface/lerobot

Quick answers

What shipped in Hugging Face on July 7, 2026?: Hugging Face shipped doc infrastructure speedups across transformers while quietly removing an experimental trainer that hasn't seen production use in six weeks. In total, 21 commits, 20 pull requests, and 2 releases landed.
Who contributed to Hugging Face on July 7, 2026?: 3 developers shipped this update, including qgallouedec, albertvillanova, and jaygala223.
What were the notable Hugging Face updates?: Make doc builds faster (#47099), Remove the PAPO trainer, and Simplify tokenization [1/N]: Remove redundant is_vlm parameter (#6298).

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.