https://promptfoo.dev/ logo
Model errors
# ❓|questions
j
Getting a decent amount of model errors today, less of the throttling one but it is still present. Any advice/tips on this? Is it todo with promptfoo itself, or the model config? I’m using two models and then a third for some llm rubric asserts that I have included in the default test. https://cdn.discordapp.com/attachments/1224783418488000522/1224783418685001798/IMG_1831.jpg?ex=661ebf72&is=660c4a72&hm=49c77acc6e6b455795af8d417e83beb3351e3168e8e28f7b4c5cdd271f0d1401&
i
is this with OpenAI?
my advice would be to add a
--delay
param if you're hitting rate limits. It'll make things slower, but sometimes it's necessary For example, to wait 1 sec between requests
Copy code
promptfoo eval --delay 1000
j
with bedrock, i’ll use the delay thanks