English(EN) Lost in the Latent Space: How Code Embeddings Represent Functional Intent and Implementation Fidelity

代码嵌入模型在功能和形式之间表现出张力 · arXiv 研究

作者 PulseAugur 编辑部 · [1 个来源] · 2026-10-29 07:17

arXiv 上发表的一项新研究调查了代码嵌入模型的内部表示，这对于软件工程应用至关重要。研究发现，在保留功能抽象和实现保真度之间存在一种张力，这意味着擅长其中一项的模型通常在另一项上表现不佳。这表明单一、通用的代码表示尚未实现，凸显了对这些模型进行更细致评估和选择的必要性。 AI

影响强调了当前代码嵌入模型的局限性，表明需要改进软件工程中的评估方法。

排序理由该项目是发表在 arXiv 上的研究论文，讨论代码嵌入模型。[lever_c_demoted from research: ic=1 ai=1.0]

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: 2 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 该项目是发表在 arXiv 上的研究论文，讨论代码嵌入模型。[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, other; 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: 1 days old; Coverage has settled into its steady-state source set.

完整方法见我们的编辑标准。

arXiv cs.IR (Information Retrieval) TIER_1 English(EN) · Zhijie Wang · 2026-10-29 07:17

迷失在潜在空间：代码嵌入如何表示功能意图和实现保真度

Code embedding models have become a key building block for many software engineering applications, yet their internal representations remain poorly understood. Existing evaluations tend to focus on downstream task performance, offering limited insight into whether a single embedd…