Optimal Re-Ranking Depth Varies by Query: LLM-Judged Predictor Cuts Depth 3x
_reachsumit · x · 2026-11-06
- Paper Optimal Re-Ranking Depth (arXiv:2611.05908, Huo, Parry, Ganguly, Clarke) challenges the fixed re-ranking depth assumption in two-stage neural rankers.
- Using LLM-based relevance judgments, the authors pinpoint per-query optimal depths and show many queries have a well-defined depth beyond which effectiveness stagnates or degrades.
- Standard query performance predictors fail at this task, but their IR-DCG@10 predictor — derived from LLM-assessed first-stage ranking quality — cuts average re-ranking depth by up to 3x while preserving effectiveness; oracle depth selection can improve retrieval by over 7%.
More from Research
- ExPert paper inverts perturbation models to trace cell states back to causal causes — anshulkundaje · 2026-11-06
- Follow-up: gradient flow lets residual states encode information for far-future predictions — mlpowered · 2026-11-06
- Gautam Kamath on AI review: fun science experiment, but false positives can mislead people at scale — thegautamkamath · 2026-11-06
- UCLA's FlowPilot uses human preferences to cut sidewalk robot interventions by 40% — zhoubolei · 2026-11-06
- diffTLM: NeurIPS 2026 Spotlight differentiable vocal tract simulator reconstructs shape from voice — sainingxie · 2026-11-06
- AI model learns thousands of words from one child's audio — only with nightly replay, Science reports — jpillowtime · 2026-11-06