DATA CUT 2026-07-27 116 活跃模型 680 A 类案例

CASE EVIDENCE / A RECORD

lucataco 使用 DeepSeek-V2 处理软件工程任务执行

lucataco 公开的代码代理与软件工程案例,来源为 公开代码库,复核于 2026-06-27T05:47:09Z。

原始记录:lucataco packaged DeepSeek-V2 as a Cog model for Replicate-style deployment

A

Chinese Brief

中文案例导读

lucataco 公开的代码代理与软件工程案例,来源为 公开代码库,复核于 2026-06-27T05:47:09Z。 Model Atlas 将它标记为 A 类证据,因为它同时具备具体使用者、具体任务、公开原始证据和可访问产物。 Model Atlas 不把 benchmark、教程、发布说明或集合页包装成真实案例。

厂商DeepSeek
模型DeepSeek-V2
任务类型代码代理与软件工程
审核状态auto_approved

任务

真实任务背景

这是一个围绕软件工程任务执行的真实任务,公开材料可以回溯到具体使用者和具体产物。 原始资料写作:lucataco created a public Cog implementation of deepseek-ai/DeepSeek-V2 so the model can be built and run with Cog/Replicate-style prediction workflows, including a documented `cog predict -i prompt=...` command.

代码代理与软件工程公开代码库A 类可核验real_case
公开产物

公开材料提供公开代码、README 或项目配置,可用于核验任务结果、项目形态和模型绑定关系。 原始资料写作:The repository contains a working prediction interface (`predict.py`) and Cog build configuration (`cog.yaml`); the README binds the artifact to `deepseek-ai/DeepSeek-V2` and shows a sample prediction prompt for text ge…

模型作用

DeepSeek-V2 在该案例中承担软件工程任务执行相关的生成、分析、编排或实现角色。 原始资料写作:DeepSeek-V2 is the exact model being wrapped and served. The artifact downloads DeepSeek-V2 weights, initializes vLLM with tensor_parallel_size=8 and max_model_len=8192, and streams generated text from user prompts thro…

风险边界

当前判断基于公开材料;若产物下线、仓库变更或模型参与比例仅来自作者自述,需要在引用前重新复核。 原始资料写作:This is an open-source deployment/integration artifact rather than a customer business story. Evidence is strong for exact model binding and a reachable public artifact, but the repository README labels the implementati…