code
PulseAugur coverage of code — every cluster mentioning code across labs, papers, and developer communities, ranked by signal.
2 天有情绪数据
-
Code does not improve LLM math reasoning; structured traces do
A new research paper explores the impact of code on mathematical reasoning in large language models. The study found that while code improves programming abilities, it does not generally enhance mathematical reasoning a…
-
AI code quality hinges on function, not looks, author argues
The author argues that the visual appeal of AI-generated code is secondary to its functional correctness and efficiency. While aesthetically pleasing code might be easier to read, it doesn't guarantee that the underlyin…
-
A-CODE model achieves fully atomic protein co-design with unified diffusion
Researchers have introduced A-CODE, a novel unified diffusion model for protein co-design that simultaneously optimizes atomic types and coordinates. This one-stage approach differs from traditional two-stage methods by…
-
New benchmarks SciMDR and ShredBench evaluate multimodal LLMs on scientific documents and reconstruction
Researchers have introduced ShredBench, a new benchmark designed to evaluate the semantic reasoning abilities of multimodal large language models (MLLMs) in reconstructing documents from shredded fragments. This benchma…
-
Anthropic Pro订阅用户讨论Code功能是否被祖父化
一位Reddit用户正在询问关于Anthropic的“Code”功能的订阅状态。他们想知道重新订阅Pro套餐是否能保留对该功能的访问权限，特别是对于那些之前订阅过然后取消的用户。该用户正在寻求其他有类似订阅经历的人的确认。
-
Replit 研究发现单独的 AI 代码安全扫描不足
Replit 发布的研究表明，仅靠 AI 进行的安全扫描不足以检测代码中的漏洞，尤其是在 Replit 这样的代码生成平台。研究发现，AI 扫描通常是非确定性的，并且对提示措辞敏感，导致诸如硬编码密钥等问题的检测不一致。此外，单独的 AI 在识别依赖项级别的漏洞和供应链风险方面存在困难，因此需要一种结合 AI 推理与传统静态分析和依赖项扫描的混合方法来实现全面的代码安全。