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

PaddleX 3.0beta1,无法正常训练多标签分类模型(PP-HGNetV2-B4_ML.yaml) #3290

Open
188080501 opened this issue Nov 8, 2024 · 1 comment
Assignees

Comments

@188080501
Copy link

188080501 commented Nov 8, 2024

我的环境是
PaddleX 3.0-beta1
paddlepaddle-gpu 3.0.0b2
Windows10 ltsc 2019
Python 3.10
CUDA 11.8
4090显卡

PaddleClas是通过插件安装的,命令如下:
paddlex --install PaddleClas

训练命令:
python main.py -c paddlex/configs/multilabel_classification/PP-HGNetV2-B4_ML.yaml -o Global.mode=train -o Global.dataset_dir=./dataset/mlcls_nus_examples -o Global.device=gpu:0

问题描述:
对于mlcls_nus_examples数据集,MultiLabelMAP不管怎么训练都是0.1左右,而且最后训练完的模型拿去推理,可以得到结果数据,但是以不管什么图得到的结果都是几乎一样的,也不正确

同样的训练数据,训练CLIP_vit_base_patch16_448_ML模型,一切正常,推理得到的结果也是正常

训练时部分输出的log如下:
MultiLabelAsymmetricLoss: 68.95149, loss: 68.95149, MultiLabelMAP(integral): 0.093
MultiLabelAsymmetricLoss: 78.03179, loss: 78.03179, MultiLabelMAP(integral): 0.099
其中MultiLabelMAP不管怎么训练都是0.1左右的一个值

@liuhongen1234567
Copy link
Collaborator

感谢您的反馈,我们这边已经复现出来这个问题,后续会及时对其进行修复

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