DeepDiscovery框架增强AI对工业代码库的理解能力

作者 PulseAugur 编辑部 · [1 个来源] · 2026-09-15 04:00

研究人员开发了DeepDiscovery，一个旨在增强大型工业代码库在软件工程任务中理解能力的新型框架。这个两阶段的定位推理系统旨在识别与任务相关的代码片段及其更广泛的上下文，即使在计算受限的情况下也能实现。在方法级别任务、内部工业场景和SWE-bench Verified基准测试上的评估表明，DeepDiscovery在改进文件恢复和提升AI编码系统的性能方面卓有成效。 AI

影响该框架有望显著提升AI编码助手在复杂软件工程任务中的能力。

排序理由该集群包含一篇详细介绍代码理解新框架的研究论文。[lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.AI 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: 17 / 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代码理解新框架的研究论文。[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, product; 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]

arXiv cs.AI TIER_1 English(EN) · Jiawei He, Weisong Sun, Mengyu Shi, Jie Jia, Tong Bian, Xikai Yang, Dong Sun · 2026-09-15 04:00

DeepDiscovery：用于任务级存储库理解的定位推理框架

arXiv:2606.22906v2 Announce Type: replace-cross Abstract: Large language models have shown strong performance on software engineering (SE) tasks, yet understanding large industrial repositories remains challenging. Existing methods often retrieve only local fragments and fail to …

本文如何被排名

报道来源 [1]

DeepDiscovery：用于任务级存储库理解的定位推理框架

相关实体

相关话题