# The tensor engine was first implemented inside SN3 (before it was called Lush) in 1992 at Bell Labs ...

Canonical URL: https://www.traeai.com/articles/b25dae67-19e9-43c9-b94c-d51ee3fcf0df
Original source: https://x.com/ylecun/status/2045606881125642536
Source name: Yann LeCun(@ylecun)
Content type: tweet
Language: 英文
Score: 7.5
Reading time: 2 分钟
Published: 2026-04-18T20:54:20+00:00
Tags: 深度学习, PyTorch, 张量计算, AI历史, 贝尔实验室

## Summary

Yann LeCun回顾1992年在贝尔实验室与Léon Bottou开发的SN3（后称Lush）中首次实现张量引擎，其命名规范延续至PyTorch等现代深度学习库。

## Key Takeaways

- 张量引擎最早于1992年在贝尔实验室的SN3/Lush系统中实现
- PyTorch等库沿用了该早期系统的张量操作命名规范
- EBlearn、Torch5/7等项目继承并传播了这一技术传统

## Citation Guidance

When citing this item, prefer the canonical traeai article URL for the AI-readable summary and include the original source URL when discussing the underlying source material.