STAIR paper: ToC-based generative retrieval with SearchTome benchmark released on arXiv
alex_verem · x · 2026-11-03
The arXiv paper (2609.03874) behind IBM's STAIR retriever proposes leveraging global structure like a table of contents, letting an LLM store the corpus in its parameters and locate answers by section. Ablations against a fine-tuned Differentiable Search Index show hallucination below 0.05% with generalization in low-sample settings.
The team also released the SearchTome benchmark (18 books, 6 domains). STAIR hits 82.6% Recall@1 vs 76.9% for fine-tuned DSI, 59.5% for BM25, and 68.7% for DPR.
Related event: IBM's STAIR Retriever Uses Document Structure to Slash Hallucinations(3 posts)→
More from Research
- Paperclip builds biomedical datasets in one shot, cutting months-long curation to about a minute — james_y_zou · 2026-11-04
- Embodied Agent Arena: Closed-Weight VLMs Lead All Five Robot Task Domains, Open-Weight Models Lag — Haojian Huang · 2026-11-04
- FPT University Paper Shows RAG Improves Stock Market Information Retrieval Accuracy and Latency — Huan Phong Thai · 2026-11-04
- Alibaba's adaptive GUI agent training cuts environment interactions by 56-61% — Yuxiang Chai · 2026-11-04
- Large-Scale Study: AI Reviewers Often Ignore How Much You Improved Your Paper — umd-zhou-lab · 2026-11-04
- Kling Team's report: video generation bottleneck shifts from rendering to semantic control — KlingTeam · 2026-11-04