Skip to content

Transformers

v5.8.1 Bugfix

This release fixes issues for SREs watching stability and regressions.

Published 22d LLM Frameworks
✓ No known CVEs patched
Read the diff → Tool health → What is this tool? →

✓ No known CVEs patched in this version

Topics

audio machine-learning deepseek gemma glm llm
+11 more
model-hub natural-language-processing nlp pretrained-models python pytorch pytorch-transformers qwen speech-recognition transformer vlm

ReleasePort's take

Moderate signal
editorial:auto 13d

Release v5.8.1 fixes several bugs impacting the DeepSeek V4 integration and related components.

Why it matters: Patch to v5.8.1 immediately if your system uses ContinuousBatchingManager, WeightConverter, or DeepSeek V4; these bugfixes resolve fatal errors, regex mismatches, and CSA mask collapse issues.

Summary

AI summary

Fixed multiple issues affecting the DeepSeek V4 integration.

Changes in this release

Bugfix Medium

Add fatal_error to ContinuousBatchingManager for error handling

Add fatal_error to ContinuousBatchingManager for error handling

Source: llm_adapter@2026-05-21

Confidence: high

Bugfix Medium

Fix WeightConverter regex incorrectly matching shared_experts as experts

Fix WeightConverter regex incorrectly matching shared_experts as experts

Source: llm_adapter@2026-05-21

Confidence: high

Bugfix Medium

Fix Deepseek V4 integration issues

Fix Deepseek V4 integration issues

Source: llm_adapter@2026-05-21

Confidence: high

Bugfix Medium

Fix Deepseek V4 CSA mask collapse

Fix Deepseek V4 CSA mask collapse

Source: llm_adapter@2026-05-21

Confidence: high

Full changelog

Patch release v5.8.1

This release is mainly to fix the Deepseek V4 integration!!!

  • [fix] Add fatal_error to ContinuousBatchingManager so the serving... by @qgallouedec, @remi-or
  • Fix WeightConverter regex incorrectly matching shared_experts as experts by @silencelamb, @claude
  • Fix deepseek v4 by @ArthurZucker (#45892)
  • Deepseek v4 csa mask collapse by @ArthurZucker, @Sawyer117 (#45928)

Weekly OSS security release digest.

The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.

No spam, unsubscribe anytime.

Share this release

Track Transformers

Get notified when new releases ship.

Sign up free

About Transformers

All releases →

Related context

Earlier breaking changes

  • v5.9.0 `text_embeds` input for SAM3, EdgeTAM, and SAM3-Lite-Text now expects full text embeddings instead of pooler outputs.

Beta — feedback welcome: [email protected]