Tags: neo/ragas
Tags
docs: langchain v3 integration tutorial (explodinggradients#1818) - This PR adds a integration tutorial for evaluating a simple question-answering bot using LangChain. - fixed the make file
fix: remove reproducibility arg (explodinggradients#1790) Remove reproducibility arg in light on alignment feature which better solves the same problem. fixes: explodinggradients#1711
docs: demo related fixes (explodinggradients#1750) Co-authored-by: jjmachan <jamesjithin97@gmail.com>
feat: Test generation for non-english corpus (explodinggradients#1734)
Fixes ToolCallAccuracy raises ZeroDivisionError when called without a… …ny arguments. (explodinggradients#1685) - Fixes: explodinggradients#1684 1. Check for both empty dictionaries: - If both refs and preds are empty, return 1.0 as you specified in your logic. 2. Check for empty refs: - If refs is empty but preds is not, return 0.0 since there's nothing to compare. Co-authored-by: ikka <shahules786@gmail.com>
Bug fixing recall in _factual_correctness.py (explodinggradients#1650) Bug fixing recall in factual correctness: recall is equal to: tp / (tp + fn) but was calculated exactly like the precision. Reference: https://docs.ragas.io/en/latest/concepts/metrics/available_metrics/factual_correctness/#factual-correctness
feat: improvements in test synthesization (explodinggradients#1621) PR 2 of improvements in test generation --------- Co-authored-by: Jin Lin Tham <jltham18@gmail.com>
docs: make llms and embeddings explicit (explodinggradients#1553)
PreviousNext