With a multi-pass teacher label construction and a position-weighted loss, a fine-tuned long-context LLM can rank 100 passages in one pass, beating sliding-window rerankers in both NDCG@10 and latency.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.IR 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Sliding Windows Are Not the End: Exploring Full Ranking with Long-Context Large Language Models
With a multi-pass teacher label construction and a position-weighted loss, a fine-tuned long-context LLM can rank 100 passages in one pass, beating sliding-window rerankers in both NDCG@10 and latency.