Access docs within a loaded model for topic merging?

[jburdo1]

Is there a way to access the documents that went into training a model from within a freshly loaded model? I made a rookie mistake of not setting a seed when taking a random sample of documents before sentence embedding and model training. I saved the model before shutting down the kernel, but now that it's loaded back up and I went through and created a list of lists for manual topic reduction, I realized that I don't have the docs variable to call within 'merge_topics', and I can't retrain on the same docs because of my seed mistake. Not a fatal error, I can retrain after taking another (seeded) random sample and manually groups topics together again, but I thought I'd first ask if there's a better way.

Thanks!

[MaartenGr]

Unfortunately, this is not possible. The documents are not saved within the model in order to keep the model small as saving input data is generally done outside of a ml model.

[jburdo1]

That was my assumption, thought I'd double check. Thanks Maarten, keep up the great work! Cheers, Joe --------------- Joseph Burdo, Ph.D. Senior Scientist, RallyPoint

…

On Fri, Apr 28, 2023 at 2:03 PM Maarten Grootendorst < ***@***.***> wrote: Unfortunately, this is not possible. The documents are not saved within the model in order to keep the model small as saving input data is generally done outside of a ml model. — Reply to this email directly, view it on GitHub <#1225 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AXI6BMDAMY3XWX5WDY2OSU3XDQA7DANCNFSM6AAAAAAXPQG4W4> . You are receiving this because you authored the thread.Message ID: ***@***.***>