How to Dynamically set Top_k Value

[amalsankar2022]

Hai team,
I am working on a questionnaire application using haystack. I am facing an issue in some cases when the top_k value is 10 for pipe.run (pipe.run(query="xxxxxxxxxxxxxx?", params={"Retriever": {"top_k": 10}, "Reader": {"top_k": 10}})) .it run infinitely .No time out and not getting an answer. For the same case I assign top_K values to 2 .Then I got an answer . How can I choose the correct top k value without getting an infinite timeout?

[bglearning]

Hi @amalsankar2022 ,

That seems odd. For the top_k=10 case does the process dies quietly or it keeps running but without any output?
Also what haystack version are you on?

[amalsankar2022]

when the top_k value is 10 for pipe.run it run infinitely. Not geting output.
I used 1.8 version

[bglearning]

Hi,
Some more questions: what is your pipeline? And what is the final output before it goes dark? Also could you keep an eye on resource usage to distinguish whether it just keeps running without outputting anything or it dies silently.

[amalsankar2022]

Hai, i am used ExtractiveQAPipeline

[bglearning]

Could you provide a bit more details? Specifically which DocumentStore, Retriever, Reader you are using in the pipeline (is it same as from some particular tutorial?) Does the pipeline run with say top_k=8? And also the full printout before it goes silent would be great