In a small Earth-science reranking dataset, DPO-trained LLMs rank best and SHAP attribution scores help a general LLM explain why items were selected, but the explanation claim rests on only two examples.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
How Reliable are LLMs for Reasoning on the Re-ranking task?
In a small Earth-science reranking dataset, DPO-trained LLMs rank best and SHAP attribution scores help a general LLM explain why items were selected, but the explanation claim rests on only two examples.