On your website you say "Keep your code organized in the way you like" but then we have very little documentation how to pair the actual bench runner command with custom agent logic.
What I'd like to see is the ability to annotate custom files I create (say, with LangChain agents in them) and then have the runner do all the magic is does with the yaml and json notation.
Something like:
@benchllm.custom_test()
or something, and then walk over and make sure those are picked up in the suite as well.
Right now it doesn't behave expectedly when I don't want to use json/yaml.
On your website you say "Keep your code organized in the way you like" but then we have very little documentation how to pair the actual
benchrunner command with custom agent logic.What I'd like to see is the ability to annotate custom files I create (say, with LangChain agents in them) and then have the runner do all the magic is does with the yaml and json notation.
Something like:
@benchllm.custom_test()or something, and then walk over and make sure those are picked up in the suite as well.
Right now it doesn't behave expectedly when I don't want to use json/yaml.