Fusing three cross-encoders buys +0.014 nDCG at 6x the cost
CShorten30 · x · 2026-11-16
Tweet 12/17: does fusing cross-encoders cash in on their complementary failures? Barely. Equal-weight RSF at k=2,000 reaches 0.438 nDCG@10 vs zerank-2's 0.424, at $0.205/query — about 6x the singleton. Fusion's clearest edge is only in Recall@20 (0.566 vs 0.560).
Related event: Reranking Study Finds Two-Stage Pipeline Beats Fusion and Deeper Windows(13 posts)→
More from Research
- BeyondBinding challenge launches: AI-designed antibodies get wet-lab tested — QuanquanGu · 2026-11-17
- Singularity-Bench: Opus 5.5 with Claude Code beats human baselines on building AI systems — armand_ruiz · 2026-11-17
- Open Vision Banana: fully open recipe turns image generators into multi-task perception systems — NielsRogge · 2026-11-17
- WebArena-Pro debuts as largest browser-agent benchmark, top agents score 62.3% — xhluca · 2026-11-17
- NeurIPS 2026 paper: RL-trained skills transfer poorly to composed reasoning tasks — StanfordAILab · 2026-11-17
- OpenAI warns CoT monitorability will degrade as models scale, researchers push back — gleech · 2026-11-17