Who contributed to Hugging Face on July 8, 2026?

3 developers shipped this update, including sayakpaul, zhtmike, and tarekziade.

What were the notable Hugging Face updates?

[sdpa] Allow prefill to use FA kernel with StaticCache (#47094), Add TIPSv2 (#46347), and Fix CLI compatibility with huggingface_hub 1.22 (#47059) (#47064).

@huggingface

Transformers, Datasets, and the open AI-model layer

github ↗

Pick a date

Topics: Python AI / ML Full archive →

The Wire · Showcase

TRANSFORMERS SHIPS FLASH ATTENTION FOR STATIC CACHE, DIFFUSERS 0.39.0 LANDS

By RepoJournal · Filed 06:04 UTC on July 8, 2026 · About Hugging Face

3 people shipped this

tarekziade @tarekziade 2 cited

sayakpaul @sayakpaul 1 cited

zhtmike @zhtmike 1 cited

Transformers unlocked faster prefill inference with Flash Attention kernel support, while diffusers ships a full release and Serge surfaces real error messages instead of pod log dead ends.

The transformers prefill optimization [1] expands Flash Attention kernel conditions to work with StaticCache, removing XPU-specific code paths and dispatching more efficiently through sdpa. That lands alongside TIPSv2 support [2], a new vision-language model that integrates Siglip2 text configs and modular loss computation. CLI users get a critical fix [3] for compatibility with huggingface_hub 1.22, which changed typer_factory's return type and broke every test runner. Diffusers shipped v0.39.0 [5] and added Flash 3 varlen_hub support for sequence parallel workflows [6] with QwenImage pipeline validation. On the ops side, Serge [7] and [8] now surface real crash causes instead of generic "see pod log" messages, and admins can test provider tokens before reviews fail silently. CI memory got a haircut too [4], with tiny Hub repos replacing 342 MB tokenizers in processor tests.

Action items

→ Test your inference provider tokens in Serge admin panel before the next review run huggingface/serge [plan]
→ Upgrade to transformers with the StaticCache prefill optimization if you're serving long-sequence inference huggingface/transformers [plan]
→ Update CLI tests if you're on huggingface_hub 1.22 or later huggingface/transformers [immediate]
→ Review diffusers v0.39.0 release notes for breaking changes before upgrading huggingface/diffusers [monitor]

References

[1] [sdpa] Allow prefill to use FA kernel with StaticCache (#47094) huggingface/transformers
[2] Add TIPSv2 (#46347) huggingface/transformers
[3] Fix CLI compatibility with huggingface_hub 1.22 (#47059) (#47064) huggingface/transformers
[4] processor tests: use tiny Hub repos to reduce CI memory (#47115) huggingface/transformers
[5] post release v0.39.0 ↗ huggingface/diffusers
[6] add SP support for `_flash_3_varlen_hub` backend ↗ huggingface/diffusers
[7] admin: add "Test" button to validate a provider token can run inference ↗ huggingface/serge
[8] review_runner: surface real crash cause instead of "(see pod log)" ↗ huggingface/serge

Quick answers

What shipped in Hugging Face on July 8, 2026?: Transformers unlocked faster prefill inference with Flash Attention kernel support, while diffusers ships a full release and Serge surfaces real error messages instead of pod log dead ends. In total, 30 commits and 30 pull requests landed.
Who contributed to Hugging Face on July 8, 2026?: 3 developers shipped this update, including sayakpaul, zhtmike, and tarekziade.
What were the notable Hugging Face updates?: [sdpa] Allow prefill to use FA kernel with StaticCache (#47094), Add TIPSv2 (#46347), and Fix CLI compatibility with huggingface_hub 1.22 (#47059) (#47064).

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.