AI代理全流程生成深度学习系统，从代码到验证均自动完成。
VibeTensor: System Software for Deep Learning, Fully Generated by AI Agents
- 用大模型代理在人类指导下生成代码，通过自动化构建测试验证
- 实现类PyTorch张量库，支持CPU/CUDA、Python/Node.js接口和自动微分
- 首个端到端由AI生成的深度学习运行时，适合关注AI编程的开发者
VIBETENSOR 是一个开源的深度学习系统软件栈，完全由大语言模型驱动的编码代理在高层人类指导下发生成。此处“完全生成”指代码溯源：所有修改均由代理提出补丁并自动应用；验证依赖代理执行的构建、测试与差异检查，无需逐条人工审查补丁。系统实现类PyTorch的即时张量库，核心为C++20（含CPU+CUDA），通过nanobind提供类torch的Python接口，并实验性支持Node.js/TypeScript。不同于轻量绑定，VIBETENSOR自包含张量存储系统、无模式调度器、反向模式自动微分、CUDA运行时（流/事件/图）、流有序缓存分配器及诊断功能，以及稳定C ABI用于动态加载算子插件。本发布被视为AI辅助软件工程的里程碑：证明编码代理可生成覆盖语言绑定至CUDA内存管理的完整深度学习运行时，主要通过构建与测试验证。我们描述架构，总结生成与验证流程，评估成果。报告仓库规模与测试集构成，汇总配套生成内核套件的可复现微基准测试，包括融合注意力与PyTorch SDPA/FlashAttention对比。还报告在NVIDIA H100（Hopper, SM90）和Blackwell GPU上3个小型工作负载（序列反转、ViT、miniGPT）的端到端训练验证；多卡结果仅限Blackwell，使用基于CUTLASS的环形归约插件，需CUDA 13+和sm103a工具链支持。最后讨论生成系统软件的失败模式，如‘弗兰肯斯坦’组合效应——局部正确模块交互导致全局性能不佳。
原文摘要 · Abstract (English)
VIBETENSOR is an open-source research system software stack for deep learning, generated by LLM-powered coding agents under high-level human guidance. In this paper, "fully generated" refers to code provenance: implementation changes were produced and applied as agent-proposed diffs; validation relied on agent-run builds, tests, and differential checks, without per-change manual diff review. It implements a PyTorch-style eager tensor library with a C++20 core (CPU+CUDA), a torch-like Python overlay via nanobind, and an experimental Node.js/TypeScript interface. Unlike thin bindings, VIBETENSOR includes its own tensor/storage system, schema-lite dispatcher, reverse-mode autograd, CUDA runtime (streams/events/graphs), a stream-ordered caching allocator with diagnostics, and a stable C ABI for dynamically loaded operator plugins. We view this release as a milestone for AI-assisted software engineering: it shows coding agents can generate a coherent deep learning runtime spanning language bindings down to CUDA memory management, validated primarily by builds and tests. We describe the architecture, summarize the workflow used to produce and validate the system, and evaluate the artifact. We report repository scale and test-suite composition, and summarize reproducible microbenchmarks from an accompanying AI-generated kernel suite, including fused attention versus PyTorch SDPA/FlashAttention. We also report end-to-end training sanity checks on 3 small workloads (sequence reversal, ViT, miniGPT) on NVIDIA H100 (Hopper, SM90) and Blackwell-class GPUs; multi-GPU results are Blackwell-only and use an optional CUTLASS-based ring-allreduce plugin gated on CUDA 13+ and sm103a toolchain support. Finally, we discuss failure modes in generated system software, including a "Frankenstein" composition effect where locally correct subsystems interact to yield globally suboptimal performance.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。