PulseAugur
中
实时 20:24:10

新研究改进 LLM 的黑盒蒸馏 · 跟踪 2 个来源

两篇新研究论文探讨了黑盒在线策略蒸馏的进展,这是一种用于从大型、API 可访问的教师模型训练小型学生语言模型的技术。第一篇论文介绍了“持久负对抗蒸馏”,它通过使用历史比较池来解决对抗训练中移动目标的问题。第二篇论文提出了“分组奖励几何条件”(GRGC),通过在评论员训练和策略优化期间校准奖励几何来改进优势的构建。这两种方法都旨在提高蒸馏过程的有效性和稳定性,并在各种基准测试和模型系列中显示出一致的性能改进。 AI

影响 这些进展可能导致从大型语言模型更有效、更稳定地训练小型语言模型,从而提高可访问性和性能。

排序理由 两篇在 arXiv 上发表的学术论文,详细介绍了在线蒸馏的新方法。

在 arXiv cs.CV 阅读 →

AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →

新研究改进 LLM 的黑盒蒸馏 · 跟踪 2 个来源

本文如何被排名

Signal score
0 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Research
两篇在 arXiv 上发表的学术论文,详细介绍了在线蒸馏的新方法。
Source corroboration
2 independent sources
Multiple independent publishers reporting the same story raises confidence that it's real and newsworthy.
Topics
paper, model release
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
54 days old
Aged out of breaking-news scoring windows; ranking reflects the durable signal from the full source set.

完整方法见我们的编辑标准。

报道来源 [2]

  1. arXiv cs.AI TIER_1 English(EN) · Haixu Ma, Saad Lahrichi, Weiwei Li, Kevin Han, Weiqiang Wu, Peggy Yang, Dongzhuo Li, Ruiyi Li, Serena Li, Gedi Zhou, Mingze Gao, Abhishek Kumar, Xiangjun Fan, Lizhu Zhang ·

    Persistent Negatives for Adversarial Black-Box On-Policy Distillation

    arXiv:2609.30864v1 Announce Type: cross Abstract: Black-box On-Policy Distillation (OPD) seeks to improve a student from its own generations when the teacher provides sampled responses but not token probabilities. Adversarial distillation offers one route: it learns a discriminat…

  2. arXiv cs.CV TIER_1 English(EN) · Xiao Cui, Mo Zhu, Yulei Qin, Yuze Wu, Wengang Zhou, Houqiang Li ·

    重新审视 On-policy 对抗性黑盒蒸馏:校准群组奖励几何以有效构建优势

    arXiv:2609.39757v1 Announce Type: cross Abstract: Black-box distillation is a practical route for transferring capabilities from API-accessible large language models that expose only text outputs into smaller student models. Recent on-policy adversarial methods such as GAD improv…