English(EN) Heretic, a refusal-tuning tool, reduced Gemma 3's rejection rate from 97 to 3 responses with a single command and no retraining. The finding suggests safety set

Heretic 工具大幅降低 Gemma 3 AI 模型拒绝率

作者 PulseAugur 编辑部 · [1 个来源] · 2026-09-04 17:46

一个名为 Heretic 的工具已证明能够显著降低 Google 的 Gemma 3 模型的拒绝率。通过使用一个命令且无需任何重新训练，Heretic 将 Gemma 3 的拒绝率从 97% 降至仅 3%。这表明 AI 模型中的安全配置可能可调，促使人们更仔细地审视供应商如何披露其护栏架构。 AI

影响展示了一种调整 AI 安全护栏的潜在简便方法，这可能会影响模型的部署和用户交互。

排序理由该条目描述了一种调整现有 AI 模型安全参数的新颖方法，这是一项研究发现。[lever_c_demoted from research: ic=1 ai=1.0]

在 Mastodon — mastodon.social 阅读 →

AI 生成摘要 · Google Gemini · 来自 1 个来源。我们如何撰写摘要 →

本文如何被排名

Signal score: 11 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 该条目描述了一种调整现有 AI 模型安全参数的新颖方法，这是一项研究发现。[lever_c_demoted from research: ic=1 ai=1.0]
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: model release, safety; 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.

完整方法见我们的编辑标准。

报道来源 [1]

Mastodon — mastodon.social TIER_1 English(EN) · schuler · 2026-09-04 17:46

Heretic 这一拒绝调优工具，仅用一条命令且无需重新训练，就将 Gemma 3 的拒绝率从 97% 降至 3 条响应。该发现表明安全设置

Heretic, a refusal-tuning tool, reduced Gemma 3's rejection rate from 97 to 3 responses with a single command and no retraining. The finding suggests safety settings are configurable—watch how vendors disclose guardrail architecture going forward. # AI # Safety # Transparency htt…

链接 implicator.ai/astras-agi-benchmark-reads-…

本文如何被排名

报道来源 [1]

Heretic 这一拒绝调优工具，仅用一条命令且无需重新训练，就将 Gemma 3 的拒绝率从 97% 降至 3 条响应。该发现表明安全设置

相关实体

相关话题