issues
search
Marker-Inc-Korea
/
RAGchain
Extension of Langchain for RAG. Easy benchmarking, multiple retrievals, reranker, time-aware RAG, and so on...
Apache License 2.0
274
stars
28
forks
source link
Add BLEU score + refactor __calculate_metrics
#315
Closed
vkehfdl1
closed
9 months ago
vkehfdl1
commented
9 months ago
Now we have BLEU score (answer evaluation with ground truths)
You can see ground truth values at EvaluateResult.each_results
We classify each metrics easily make using_metrics and its child classes.
change answer metric solution input for multiple ground truth values.