search

kbeaugrand / SemanticKernel.Connectors.Memory.SqlServer

SQL Server connector for Semantic Kernel plugin and Kernel Memory
MIT License
46 stars 10 forks source link

Using Kernel Memory and not getting anything out #140

Closed RRoss-Carollo closed 5 months ago

RRoss-Carollo commented 5 months ago

Hello,

I have a basic import function that takes in my employee handbook for my company and when I use ask or search I get back one result with no useful information.

Import is very basic, I'm using ImportTextAsync with chunks of data, setting the document id and a group.

For the request I am trying to use Search but I have also been testing with Ask.

Search code: SearchResult searchResult = await sqlMemory.SearchAsync(ask, "general") MemoryAnswer askResult = await sqlMemory.AskAsync(ask, "general")

Any ask results in the same replies: Search will give me back a result with one partition, and that is just the first page of text. Ask returns "INFO NOT FOUND" without the period. In my messing around I was able to get it to return a result of "INFO NOT FOUND." with a period but I can't seem to replicate it now and it was the only difference I have been able to get.

I have tried deleting the database and re-embedding the file, adding a minimum relevance, and changing my index in the hopes of getting any result, all with no luck.

Any help or direction would be greatly appreciated.

RRoss-Carollo commented 5 months ago

I solved this,

I was not changing the document id and overwriting the same item over and over.