search

irecsys / CARSKit

Java-Based Context-aware Recommendation Library
https://carskit.github.io/
GNU General Public License v3.0
125 stars 53 forks source link

Setting up seed for algorithms with initialization #20

Closed haonanzhang-vv closed 4 years ago

haonanzhang-vv commented 4 years ago

Hi Prof. Zheng,

I saw the CV can specify the random seed, but I think it is for data partition?

I want to reproduce the result (MAP, NDCG), and I use CAMF_CU, whose fitting requires initialisation and gradient descent. I am wondering whether there is a way that I can fix the initialization for the gradient descent? Fix the initial value?

Also, I am wondering how do you use command line code to compile the java scripts into jar file? Would you mind sharing that command in this space.

Thank you for your consideration.

irecsys commented 4 years ago

Hello, thanks for your interests in CARSKit.

CARSKit was built upon LibRec v1.4. The library will create random initializations in the iterative learning methods, such as algorithms based on matrix factorization. If you would like to fix the initializations, you need to download a copy of the source codes, and change the coding.

In terms of the jar file, I used the function in the IDE IntelliJ IDEA, https://www.jetbrains.com/idea/

haonanzhang-vv commented 4 years ago

Dear Prof. Zheng,

Thank you for your quick response.

Yes, I am using Intelli J to compile the file. I am wondering whether we could have a quick zoom chat as I cannot successfully compile the Jar file. I really appreciate your work and hope to have a chat with you if possible.

Thank you for your consideration!

irecsys commented 4 years ago

Problems solved offline. Issue closed.