https://community.databricks.com/t5/get-started-discussions/using-model-serving-from-databricks-privacy-issue/m-p/77996#M7705 <P>When using Databrick's model serving to query Llama3, I noticed the endpoint URL is my databricks instance. Does that still mean data is sent to databricks process by databricks?&nbsp;&nbsp;<BR />If so, does databricks keep/use any of the data sent to model serving end point?</P> Tue, 09 Jul 2024 22:54:35 GMT gmu77113355 2024-07-09T22:54:35Z https://community.databricks.com/t5/get-started-discussions/using-model-serving-from-databricks-privacy-issue/m-p/77996#M7705 <P>When using Databrick's model serving to query Llama3, I noticed the endpoint URL is my databricks instance. Does that still mean data is sent to databricks process by databricks?&nbsp;&nbsp;<BR />If so, does databricks keep/use any of the data sent to model serving end point?</P> Tue, 09 Jul 2024 22:54:35 GMT https://community.databricks.com/t5/get-started-discussions/using-model-serving-from-databricks-privacy-issue/m-p/77996#M7705 gmu77113355 2024-07-09T22:54:35Z https://community.databricks.com/t5/get-started-discussions/using-model-serving-from-databricks-privacy-issue/m-p/78392#M7707 <P>Related question, when databricks process requests for foundational model, I noticed the latency is pretty small, wondering what kind of processing power is used on databricks side? I am interested in hosting model ourselves so wondering what type of processing power does databricks use.</P> Thu, 11 Jul 2024 16:23:08 GMT https://community.databricks.com/t5/get-started-discussions/using-model-serving-from-databricks-privacy-issue/m-p/78392#M7707 gmu77113355 2024-07-11T16:23:08Z