issues
search
gomate-community
/
rageval
Evaluation tools for Retrieval-augmented Generation (RAG) methods.
Apache License 2.0
81
stars
9
forks
source link
update answer nli groundedness metric
#60
Closed
QianHaosheng
closed
4 months ago
QianHaosheng
commented
4 months ago
59 In the evaluation of ALCE's ELI5 dataset, claim recall is calculated by decomposing gt_answer into claims, and then using the NLI model to determine whether the answer generated by LLM can contain the claim.
59 In the evaluation of ALCE's ELI5 dataset, claim recall is calculated by decomposing gt_answer into claims, and then using the NLI model to determine whether the answer generated by LLM can contain the claim.