Getting DFP to generate monthly based models

I am trying to figure out how to get a monthly-based autoencoder model .
using the default “production” example on a larger set of log files, when I set DFPFileBatcherStage period=“D” i get a separate model for each day, if i change it to period=“M” I do get a monthly model but i think i am missing the advantage of the daily caching, what am i missing?

Hi Tamir, thanks for your issue. We think we’ve identified a problem in a related stage and the team has created a bug for it here:

Feel free to add your comments or observations in that issue to help us arrive at the correct solution.


A PR was merged to address this: Update DFPRollingWindowStage to emit correct window by efajardo-nv · Pull Request #683 · nv-morpheus/Morpheus · GitHub
In order to test for your use case, please consult the Contributor’s guide for building your own docker image or conda environment:
Morpheus/ at branch-23.03 · nv-morpheus/Morpheus · GitHub