More complete evaluation

wejick / gchain

Composable LLM Application framework inspired by langchain

Apache License 2.0

12 stars 2 forks source link

More complete evaluation #41

Closed wejick closed 9 months ago

wejick commented 9 months ago

QARelevanceEval is grading whether the the answer relevant to the question according to the given fact. CorrectnessEval is an evaluator to evaluate input againts expectation.

Rudimentary test runner