Generators response text limit

[mohduvesh043]

I am utilizing Gemini Pro Models to generate text for the job description. However, the generated output is not consistently complete, as there is often some missing information. Is there a limit to the response generated by the generator?

@xavidop 

[xavidop]

It is a limit of 4k characters: https://cloud.google.com/dialogflow/quotas#length_limits

Best,

Xavi

[mohduvesh043]

But I always generate incomplete text(I want to generate job descriptions using a generator and Gemini models). I don't know if it an issue with a generator or Dialogflow cx.

[xavidop]

did you check the number of characters? because it could be incomplete because you are reaching 4K characters

[mohduvesh043]

 it is around always 90 to 95 characters

[xavidop]

that is weird, I cannot help you more with this topic. Not sure why this is happening

[mohduvesh043 · Accepted solution]

Hi,

I found solution.  It is due to  their is token limit inital is set .

[xavidop]

ooooh!! good catch!

[serranodf]

This only works if it is done from Vertex IA, I have the same problems from Dialog Flow CX and I cannot find any section where the ML configuration can be done for the Token limit and temperature configuration

[xavidop]

You can modify the LLM config in any Gen AI feature on Dialogflow CX

[VishalBulbule]

Gemini specifications given here
https://ai.google.dev/models/gemini

[AICoachJeffK]

I am looking for a similar but opposite use case. Looking for a prompt that will enforce the response does NOT exceed 105 characters. Attempted stating this requirement in multiple places but still getting some responses over 130.