BiXSE trains dense retrievers with binary cross-entropy on LLM-generated graded relevance probabilities, consistently beating InfoNCE and roughly matching pairwise ranking baselines.
Results shown for the best-performing configuration per loss type among BGE-trained models
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.IR 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
BiXSE: Improving Dense Retrieval via Probabilistic Graded Relevance Distillation
BiXSE trains dense retrievers with binary cross-entropy on LLM-generated graded relevance probabilities, consistently beating InfoNCE and roughly matching pairwise ranking baselines.