首页
技术日记
编程
旅游
数码
登录
标签
tune
神经网络编译器的Tensor优化:auto tune和auto schedule
一般情况下,深度神经网络的计算本质上是一对tensor的计算,例如常见的conv2d的计算本质上是一个7层的for循环,那么底层的硬件,例如内存大小,SM的数量,threads和blocks等都会对最终的for循环造成影响。 现存的深度学
神经网络
编译器
tensor
schedule
tune
admin
3月前
48
0
[Ray.Tune] [已解决] Queue objects should only be shared between processes through inheritance
问题错误代码2021-12-24 17:48:01,170WARNING experiment.py:256 -- No name detected on trainable. Using DEFAULT.2021-12-24 17:
Queue
tune
Ray
objects
Inheritance
admin
7月前
121
0
【论文阅读笔记】Tune-An-Ellipse: CLIP Has Potential to Find What You Want
来源:CVPR 2024 源码:GitHub - showlabTune-An-Ellipse: [CVPR 2024] Tune-An-Ellipse: CLIP Has Potential
笔记
论文
tune
Ellipse
potential
admin
8月前
74
0
ChatGPT fine tune微调+prompt介绍
[db:摘要]
Fine
ChatGpt
Prompt
tune
admin
2024-9-14
86
0