T
traeai
Sign in

人物

Sara Hooker

别名:Dr. Sara Hooker

AI 学者,曾指出 DNN 主要由矩阵乘法构成,该观点被引用以强调 matmul 的重要性。

相关材料

已收录 2 条与 Sara Hooker 相关的内容,按评分排序。

Profiling in PyTorch (Part 1): A Beginner's Guide to torch.profiler

Profiling in PyTorch (Part 1): A Beginner's Guide to torch.profiler

Hugging Face Blog5278 字 (约 22 分钟)
87

This beginner-friendly guide walks through using torch.profiler to analyze a matrix multiplication + addition operation, revealing CPU-GPU coordination patterns and how torch.compile fuses operations to reduce kernel launch overhead.

入选理由:使用 `torch.profiler.profile` + `record_function` 可轻松捕获 CPU/GPU 事件与内核调用链,生成可交互 trace 文件

FeaturedArticle#PyTorch#profiler#performance#CUDA#torch.compile英文
On the slow death of Scaling (birth of Adaption Labs) | Sara Hooker | HF ML Club India EP2

On the slow death of Scaling (birth of Adaption Labs)

Hugging Face14376 字 (约 58 分钟)
75

Current AI over-relies on scaling; Adaption Labs focuses on adaptive intelligence to address real-time learning and environmental adaptation challenges

入选理由:当前AI系统依赖大规模数据和计算,但缺乏适应动态环境的能力(如实时学习与场景迁移)

FeaturedVideo#Adaption Labs#AI#Machine Learning#Adaptive Intelligence#Deep Learning英文

跨材料问答 · Sara Hooker

回答基于:Sara Hooker 相关 2 条材料
    0 / 500

    AI may generate inaccurate information. Please verify important content.