English(EN) Draft-KV: Learning Useful Latent Communication Between Language Models

Draft-KV 实现冻结语言模型之间的潜在通信

作者 PulseAugur 编辑部 · [1 个来源] · 2026-09-28 00:00

研究人员推出了一种名为 Draft-KV 的新方法，该方法能够让语言模型通过其内部状态而非仅通过解码文本来传递有用信息。这种方法允许接收模型直接利用共享模型在起草响应时生成的键值（KV）状态。与以往通信增益可能不依赖于内容的旧方法不同，Draft-KV 促进了基于信息的协作，并显示出显著的改进。一个冻结的 Qwen2.5-0.5B-Instruct 接收模型在与 Qwen3-8B 共享模型配对时，在 MMLU-Redux 上取得了 78.04% 的分数。 AI

影响这种方法可能带来更高效、更有效的专业化 AI 模型之间的协作。

排序理由该集群包含一篇详细介绍语言模型通信新方法的论文。[lever_c_demoted from research: ic=1 ai=1.0]

在 Hugging Face Daily Papers 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: 0 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 该集群包含一篇详细介绍语言模型通信新方法的论文。[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: 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: 2 days old; Coverage has settled into its steady-state source set.

完整方法见我们的编辑标准。

报道来源 [1]

Hugging Face Daily Papers TIER_1 English(EN) · 2026-09-28 00:00

Draft-KV: 学习语言模型间有用的潜在通信

Latent communication passes internal states between language models instead of decoded text, but higher receiver accuracy does not show that the receiver used the message content. Across five method-dataset pairs, replacing each message with one from an unrelated question changes…

本文如何被排名

报道来源 [1]

Draft-KV: 学习语言模型间有用的潜在通信

相关实体

相关话题