Gemini Pro error

[Manish7]

I'm testing Gemini Pro & Bison to generate a HTML script given a description of what the HTML script would do. Sometimes it gives correct response, but most of the times it starts generating the HTML script and stops in between, final response I get:

The model response was blocked because of a quality issue or a parameter setting. Try increasing the temperature, top-k, or top-p parameters to generate a different response. (error code: {errorCodes})

errorCode -> 210 in case of Bison

I tried the the advised thing in the message but the error persists.

Please assist me through this issue.

[Cedart]

Hello,

I have the same error on the GCP prompt interface. And when I use the stream API I got a partial  or empty answer . I think the interface crash due to a empty answer part.

  const streamingResp = await generativeModel.generateContentStream(req);

  for await (const item of streamingResp.stream) {

    process.stdout.write('stream chunk: ' + item.candidates[0].content.parts[0].text); // crash parts[0] is undefined
  }​

Regards,

Cedric

[Manish7]

Yup, same here.

[GeneAlchemist]

Were you trying to submit a request that would output a high token count? I was getting this on requests, and was surprised at how limited the token response was. I kept reducing the request complexity until it responded.

[AndrewB]

Did you try adjusting the safety filters? 

https://cloud.google.com/vertex-ai/generative-ai/docs/multimodal/configure-safety-attributes#how_to_...