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

在全量ali-ccp数据集上训练ESMM模型,在测试集上,cvr和ctcvr的auc=1.0 #56

Closed
Hjh233 opened this issue Oct 30, 2023 · 1 comment

Comments

@Hjh233
Copy link
Contributor

Hjh233 commented Oct 30, 2023

环境

ubuntu 22.04, python=3.9.18, torch=2.1.0, GPU=3090,

预处理

利用preprocess_ali_ccp.py脚本处理得到 train set & test set(没有将test set进一步拆分为validation set和test set)

参数设置

tutorials/Multi_Task.ipynb文件中,将task_types = ["classification", "classification"] 修改为task_types = ["classification", "classification", "classification"](MTLTrainer中第111行,ESMM模型的total_loss = sum(loss_list[1:]),因此训练时一定是要将task_type设置成三个二分类任务(cvr,ctr,ctcvr)),其他超参数设置不变,进行全量模型训练,得到的日志如下

输出日志

train loss: {'task_0:': 0.0009412071586964892, 'task_1:': 0.15991775316040385, 'task_2:': 5.389307421478335e-05}
epoch: 0 validation scores: [1.0, 0.5951526439762982, 1.0]

@Hjh233 Hjh233 closed this as completed Oct 31, 2023
@Hjh233
Copy link
Contributor Author

Hjh233 commented Oct 31, 2023

原始预处理中,sparse feature中含有 ctcvr_label = cvr_label,删除后模型训练正常。

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

1 participant