Per-Field Selective Risk Control for Document Extraction
Bhaskar Gurram · hf · 2026-08-18
This paper discusses per-field selective risk control for document extraction tasks. It proposes a validity ladder involving fit/val splits and Mondrian PAC certificates, revealing that support-bin provenance outperforms learned fusion under specific model conditions.
More from Research
- Research: Models Can Detect Deception and Learn Who to Ignore — logangraham · 2026-08-19
- OpenAI Frontier Red Team Probes Socioeconomic Risks of Trillions of Agents — logangraham · 2026-08-19
- Super-Scaled Motion Tracking Model Enables Natural Humanoid Movement — zhengyiluo · 2026-08-19
- Open Source Project: Probing LLM Latent Structure via Anthropological Methods — No-Reserve2026 · 2026-08-19
- Nature Comms study: 9 of 15 human-AI groups preserved the optimal strategy across generations — iyadrahwan · 2026-08-18
- New Benchmark Live: Independent Measure of AI Usage Based on 24k Real Conversations — sanmikoyejo · 2026-08-18