search

yusanshi / news-recommendation

Implementations of some methods in news recommendation.
MIT License
241 stars 50 forks source link

测试集 #32

Closed whxhx closed 2 years ago

whxhx commented 2 years ago

你好，MIND large测试集是没有标签吗？那评估的时候只能用验证集评估吗？需要负采样什么的吗？我直接预测算的准确度很低很低

yusanshi commented 2 years ago

你好请参考 https://github.com/yusanshi/news-recommendation/issues/29#issuecomment-1070034286

whxhx commented 2 years ago

非常感谢您~~那可以用train作为训练集，dev作为测试集吗，不设置验证集。或者干脆自己划分数据集可以吗？

yusanshi commented 2 years ago

哈哈当然可以，比如 train set 中划出来一部分做 val set，然后现在的 val set 作为 test set。

yusanshi commented 2 years ago

只不过这样的话，毕竟没有用官方发布的 test set，没法打榜罢了

whxhx commented 2 years ago

好的，感谢您的回答~非常非常非常感谢~~

whxhx commented 2 years ago

您好，还有一个问题，我看large数据集上train set, val set, test set是有重叠用户历史纪录的，而small数据集train set和val set没有重叠用户，训练模型的话需要保证是相同的用户吗？

yusanshi commented 2 years ago

你好请看下#35~

https://hackernoon.com/45-github-issues-dos-and-donts-dfec9ab4b612

whxhx commented 2 years ago

好的，明白啦~太感谢您啦，可以怎么支持您呢？

yusanshi commented 1 year ago

哈哈没关系，感谢支持😄

On Wed, Apr 20, 2022 at 11:27 Shannon @.***> wrote:

认真看了您发的链接，不好意思，之前不正确地提交issue，之后会按照规范reporting issue的~

Shannon 手机：15050591110 @.***

在2022年4月20日 @.***> 写道：

你好请看下#35~

https://hackernoon.com/45-github-issues-dos-and-donts-dfec9ab4b612

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you modified the open/close state.Message ID: @.***>

— Reply to this email directly, view it on GitHub https://github.com/yusanshi/news-recommendation/issues/32#issuecomment-1103432952, or unsubscribe https://github.com/notifications/unsubscribe-auth/AIUV5BW7B33JDHVJAZCEIH3VF52RBANCNFSM5SR267PQ . You are receiving this because you modified the open/close state.Message ID: @.***>