Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

MNN version in your repo #1

Closed
hebangwen opened this issue Mar 21, 2024 · 3 comments
Closed

MNN version in your repo #1

hebangwen opened this issue Mar 21, 2024 · 3 comments

Comments

@hebangwen
Copy link

您好,我目前也碰到了 MNN 量化精度完全错误的问题,想请问一下您仓库里能正常推理出结果的 MNN 版本是多少。是否是 MNN/issues/2614 使用的 2.5 版本,还是注释中的 2.6.2 版本

@YingkunZhou
Copy link
Owner

YingkunZhou commented Mar 23, 2024

目前拉取最新代码基本上传统的CNN PTQ int8量化就能work,下面是我实测的精度(取imagenet validation dataset的1/20子集做的快速验证):

Model Top-1 Acc
resnet50 78.3%
mobilenetv3_large_100 54.5%
tf_efficientnetv2_b0 71.8%
tf_efficientnetv2_b1 71.4%
tf_efficientnetv2_b2 72.3%

但是更先进的mobile ViT模型的int8精度就完全不能看了

@YingkunZhou
Copy link
Owner

本仓库用的各种DL engines的版本信息参见:

https://github.com/YingkunZhou/EdgeTransformerBench/releases/tag/v1.0

@hebangwen
Copy link
Author

感谢回复,我这两天基于 2.3.0 测试了 ViT 和 swin transformer,结果还正常,没有出现全部变成一个数值的情况。我再测试一下最新分支的 CNN 网络和 transformer 量化情况。

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants