DART adapts a scoring matrix at inference time via gradient updates on pseudo-labels from top/bottom documents to gain +2.1% mean NDCG@10 on six BEIR benchmarks with under 10ms added latency.
arXiv preprint arXiv:2212.08841 , year=
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
years
2026 2verdicts
UNVERDICTED 2representative citing papers
CoHyDE iteratively co-trains an LLM rewriter and dense encoder using InfoNCE and DPO, improving NDCG@5 by 2.5-6.3 points over baselines on a 10k tool catalog, with larger gains on vague queries.
citing papers explorer
-
CoHyDE: Iterative Co-Training of LLM Rewriter & Dense Encoder for Tool Retrieval
CoHyDE iteratively co-trains an LLM rewriter and dense encoder using InfoNCE and DPO, improving NDCG@5 by 2.5-6.3 points over baselines on a 10k tool catalog, with larger gains on vague queries.