https://promptfoo.dev/ logo
Can I do this in promptfoo.yaml?
# ❓|questions
v
defaultTest: options: provider: id: file://../../config/model_id.json config: apiHost: file://../../config/model_host.json temperature: 0.7 Is there any alternative to do this? one alternative I know that just put every thing in one .yaml file and use that file for assertion provider sample.yaml file will contain # Learn more about AZURE provider config: https://www.promptfoo.dev/docs/providers/azure#configuration id: "" config: apiHost: "" temperature: 0.7 >>>> I have another question that when should I use file:// to refer to a file? and when should I can use " - ../../config/model_id.json" here difference is that "file keyword" is not used?
i
is the idea just to make swapping models easier across multiple configs?
v
our idea is to print llm model (specific model ) in output.json file which we get when we run the command npx promptfoo eval -o output.json when we are using put everything in the sample.yaml file and use below option defaultTest: options: provider: file://./config/sample.yaml and sample.yaml contains # Learn more about AZURE provider config: https://www.promptfoo.dev/docs/providers/azure#configuration id: azureopenai:chat:gpt-4-1106-Preview config: apiHost: "" temperature: "" but in output.json we are getting provider: file://.config/sample.yaml while it should give id: azureopenai:chat:gpt-4-1106-Preview and other informations whatever are there in sample.yaml file