When training models involving region-related data (such as language models, recommendation systems, advertising, etc.), how can we effectively test the performance of the models in different regions? For example, the recommendation system may need to adjust the recommendation strategy based on the user’s geographic location, and direct access to real data in the target region may be limited. We want to try to use the proxy IP pool to simulate global requests, but we don’t know if it will work. We are currently using pia unlimited IP proxy, and hope to solve it as soon as possible.
How can we efficiently simulate user requests from different regions to verify the generalization ability of the model?
I don’t know if it can be completed successfully.
Perhaps gathering data from all regions and training it those data…
1 Like
@CloudHop
Per the Code of Conduct, you are only allowed one post that advertises a service.
Thank you, I will abide by the community agreement.