PulseAugur
EN
LIVE 04:16:12

SGLang bug causes endless repetition in FP8 lm_head models

By PulseAugur Editorial · [1 sources] ·

A bug in SGLang versions prior to commit 5375babb causes endless repetition and empty responses when serving models with FP8 lm_head configurations, such as unsloth/Qwen3.8-27B-NVFP4. This issue arises because SGLang incorrectly handles the lm_head component, leading to corrupted logit rankings. The problem is silent at server startup, with the only indication being a specific warning in the server load logs. A fix was merged into SGLang's main branch on August 19, 2026, but has not yet been included in a released version. AI

IMPACT This bug fix addresses a specific serving issue for certain model configurations, potentially improving reliability for users of SGLang.

RANK_REASON This is a technical bug report and fix for a specific software library, not a new model release or significant industry event.

Read on dev.to — LLM tag →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

SGLang bug causes endless repetition in FP8 lm_head models

How we ranked this

Signal score
40 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
This is a technical bug report and fix for a specific software library, not a new model release or significant industry event.
Source corroboration
Single-source cluster
Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics
infra
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
Breaking (< 6h)
Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

Full methodology in our editorial standards.

COVERAGE [1]

  1. dev.to — LLM tag TIER_1 English(EN) · Jahn ·

    SGLang outputs endless repetition on NVFP4 models: the FP8 lm_head bug

    <p>Your NVFP4 model serves fine on vLLM but outputs an endlessly repeated phrase on SGLang, from the very first token, even on a trivial prompt. The response <code>content</code> comes back empty, every request ends with <code>finish_reason: length</code>, and reasoning traces lo…

RELATED ENTITIES

RELATED TOPICS