Training automatically with a scheduler the Vertex AI search & conversation model

[kbujari]

Hi everyone,

I am searching for more than a week on how to train weekly my  Vertex AI search & conversation model. I have implemented the code that will upload my data weekly to GC Storage, I also implemented the code that transfers the data from GC Storage to the  Vertex AI search & conversation Data Store, now I need the help on how to train this model without human interaction, as the fist two steps are set with a scheduler and the process of data uploading and transferring is completed automatically. Can you help me on the following steps? How can I run the model to train each week after I update my dataset?

Thank you in advance!

[xavidop · Accepted solution]

Hi @kbujari since you are updating your bucket and your datastore, you do not need to do anything to re-train, when a new document is added to a data store, the data store is re-indexed automatically: check the documentation here: https://cloud.google.com/generative-ai-app-builder/docs/refresh-data#discoveryengine_v1_generated_Do...

[kbujari]

Thank you for your help once again. This works!

[LikeADemonicMin]

Hi! 

Could you please share with me the code that transfers the data from GC Storage to the  Vertex AI search & conversation Data Store? I'm new to GCP and struggling with that 😞 I've seen that you've asked the question before and an answer was given, but I didn't undesertand it...

Thanks in advance!