Make use of the seed
parameter in the LLM-based evaluators using the OpenAI API
#7713
Labels
seed
parameter in the LLM-based evaluators using the OpenAI API
#7713
Is your feature request related to a problem? Please describe.
The LLM-based evaluators are non-deterministic, and this can be an issue in replicating and comparing evaluation results.
Describe the solution you'd like
One possible solution is to use the
seed
parameter when using the OpenAI-based LLMs: https://platform.openai.com/docs/api-reference/chat/createseed
: integer or nullThe text was updated successfully, but these errors were encountered: