Will there be a GA or Production ready version for Conversational Search as it says experimental

[ramda1234786]

Will there be a GA or Production ready version for Conversational Search as it says experimental ? Also from where i can test the new version as this repository does not have any new code, i see nothing in opensearch-ml comms also apart from this one https://github.com/opensearch-project/ml-commons/blob/main/memory/src/main/java/org/opensearch/ml/memory/ConversationalMemoryHandler.java

Also do we have any documentation to do initial testing on 2.12

Thanks in advance for information RamDa

[HenryL27]

the conversation search feature should be GA for 2.12 - it's in app security / pentesting now.

probably the best way to test it is with the staging docker image. The documentation hasn't been updated yet, so unfortunately the best documentation may be github issues and code in ml-commons... sorry about that. This file has all the memory endpoints in it. The RAG pipeline hasn't (afaik) changed

[HenryL27]

git blame can maybe help find the relevant issues

[ramda1234786]

Thanks @HenryL27 , i was trying the post_process_function, i tried to take help from various sources but no luck. Now i am wondering if there is no changes in RAG files even with GA looks like my solution will not work.

I have tried with langchain but that does not support hybrid search. So unfortunately i am not able to move forward

https://github.com/opensearch-project/ml-commons/issues/1925#issuecomment-1912974517

[mashah]

Version 2.12 is still under development. If you need RAG with OpenSearch my recommendation is to try Sycamore. We know that path works, though it's using 2.11. Once 2.12 is ready, we will have that working with Sycamore as well.

[ramda1234786]

Thanks for your input, my problem is not in data ingestion but using "post_process_function": "\n return params['response'][0].generated_text; \n"

I have this

[
    {
        "generated_text": "Your Generated text"
    }
]

and i want to convert it to this below using post process fuction

   {
        "completion": "Your Generated text"
    }

I have tried this till now

"post_process_function": "\n def json = \"{\" +\n \"\\\"completion\\\":\\\"\" + params['response'][0].generated_text + \"\\\" }\";\n return json;\n "

But it never works with _predict APIs

[ylwu-amzn]

@ramda1234786 Can you cut an issue in ml-commons repo for this process function error https://github.com/opensearch-project/ml-commons/issues

[ramda1234786]

Hi @ylwu-amzn , i have added here https://github.com/opensearch-project/ml-commons/issues/1990