PulseAugur
实时 00:19:43
English(EN) Solar Open 2 Technical Report

Solar Open 2 语言模型拥有 100 万 token 上下文窗口和强大的代理能力

研究人员推出了 Solar Open 2,这是一个拥有 2500 亿参数的混合专家(Mixture-of-Experts)语言模型,专为长时序代理任务而设计。该模型通过混合注意力机制实现了 100 万 token 的上下文窗口,并利用更强的起点和精选的高价值数据进行高效训练。Solar Open 2 在多项英语基准测试中表现领先,包括 MMLU-ProAPEX-Agents 套件,并与其他顶级开放权重模型取得了有竞争力的结果。它还在韩语基准测试和内部韩语办公代理基准测试中取得了最高分,性能优于更大的模型。 AI

影响 在代理基准测试上设定了新的 SOTA(State-of-the-Art),并展示了大型上下文窗口的高效训练。

排序理由 技术报告,详细介绍了具有基准测试结果的新语言模型发布。[lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.CL 阅读 →

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

Solar Open 2 语言模型拥有 100 万 token 上下文窗口和强大的代理能力

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
技术报告,详细介绍了具有基准测试结果的新语言模型发布。[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, paper
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
48 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 [1]

  1. arXiv cs.CL TIER_1 English(EN) · Sungrae Park (University of Seoul), Sanghoon Kim (University of Seoul), Gyoungjin Gim (University of Seoul), Jungho Cho (University of Seoul), Hyunwoong Ko (University of Seoul), Minbyul Jeong (University of Seoul), Minjeong Kim (University of Seoul), Ke… ·

    Solar Open 2 技术报告

    arXiv:2607.20062v1 Announce Type: new Abstract: We present Solar Open 2, a 250B-A15B Mixture-of-Experts language model built for long-horizon agentic tasks, scaled up from Solar Open 1 (Solar Open 100B). To hold entire agent trajectories in a single context, Solar Open 2 reaches …