Kevin Di's Highlights on 'PyTorch 2.0'

NLP（二十）：漫谈 KV Cache 优化方法，深度理解 StreamingLLM

zhuanlan.zhihu.com/p/659770503

Oct 11, 2023

疯狂的 H100：现代 GPU 体系结构浅析，从算力焦虑开始聊起

zhuanlan.zhihu.com/p/659738090?utm_psn=1693972042385416192

Oct 9, 2023

FlashAttention2详解（性能比FlashAttention提升200%）

zhuanlan.zhihu.com/p/645376942

Oct 8, 2023

分析transformer模型的参数量、计算量、中间激活、KV cache

zhuanlan.zhihu.com/p/624740065

Sep 12, 2023

图解GPT-2 | The Illustrated GPT-2 (Visualizing Transformer Language Models)_Ann's Blog的博客-CSDN博客

blog.csdn.net/qq_36667170/article/details/125529598?spm=1001.2014.3001.5501

Sep 8, 2023

解析 Transformer 模型 | Way to AGI

blog.waytoagi.com/article/transformer_explained

Sep 7, 2023

PyTorch 2.0

pytorch.org/get-started/pytorch-2.0/

Sep 4, 2023

Google Gemini Eats The World – Gemini Smashes GPT-4 By 5X, The GPU-Poors

www.semianalysis.com/p/google-gemini-eats-the-world-gemini?utm_source=substack&utm_medium=email

Aug 29, 2023

GPU Performance Background User's Guide - NVIDIA Docs --- GPU 性能背景用户指南 - NVIDIA Docs

docs.nvidia.com/deeplearning/performance/dl-performance-gpu-background/index.html

Aug 28, 2023

Best Practices for Building and Deploying Recommender Systems - NVIDIA Docs --- 构建和部署推荐系统的最佳实践 - NVIDIA Docs

docs.nvidia.com/deeplearning/performance/recsys-best-practices/index.html

Aug 24, 2023

Memory-Limited Layers User's Guide - NVIDIA Docs --- 内存限制层用户指南 - NVIDIA Docs

docs.nvidia.com/deeplearning/performance/dl-performance-memory-limited/index.html

Aug 23, 2023

Recurrent Layers User's Guide - NVIDIA Docs --- 循环层用户指南 - NVIDIA Docs

docs.nvidia.com/deeplearning/performance/dl-performance-recurrent/index.html

Aug 23, 2023

Convolutional Layers User's Guide - NVIDIA Docs --- 卷积层用户指南 - NVIDIA Docs

docs.nvidia.com/deeplearning/performance/dl-performance-convolutional/index.html

Aug 22, 2023

Linear/Fully-Connected Layers User's Guide - NVIDIA Docs --- 线性/全连接层用户指南 - NVIDIA Docs

docs.nvidia.com/deeplearning/performance/dl-performance-fully-connected/index.html

Aug 21, 2023

Matrix Multiplication Background User's Guide - NVIDIA Docs --- 矩阵乘法背景用户指南 - NVIDIA Docs

docs.nvidia.com/deeplearning/performance/dl-performance-matrix-multiplication/index.html

Aug 17, 2023

The History of Open-Source LLMs: Better Base Models (Part Two)

cameronrwolfe.substack.com/p/the-history-of-open-source-llms-better?utm_source=substack&utm_medium=email

Aug 1, 2023

The History of Open-Source LLMs: Early Days (Part One)

cameronrwolfe.substack.com/p/the-history-of-open-source-llms-early

Aug 1, 2023

“这是一种战争行为”：解码美国对华芯片封锁行动

cn.nytimes.com/usa/20230713/semiconductor-chips-us-china/?utm_source=news-list&utm_medium=email&utm_campaign=newsletter

Jul 28, 2023

NLP（十八）：LLM 的推理优化技术纵览

zhuanlan.zhihu.com/p/642412124

Jul 17, 2023

NLP（十七）：从 FlashAttention 到 PagedAttention, 如何进一步优化 Attention 性能

zhuanlan.zhihu.com/p/638468472

Jul 14, 2023