2 papers
cs.CV2026
Capability-Routed Visual Retrieval and Evidence Threading for Long-Context Document Question Answering
Amirul Rahman, Aisha Karim, Kenji Nakamura +1
Annual reports, diligence packs, and infographic dashboards bury numbers in page images: axes, cell grids, and footnotes that OCR pipelines flatten and that page-level visual retri…
cs.CL2026
Avoiding Overthinking and Underthinking: Curriculum-Aware Budget Scheduling for LLMs
Amirul Rahman, Aisha Karim, Kenji Nakamura +1
Scaling test-time compute via extended reasoning has become a key paradigm for improving the capabilities of large language models (LLMs). However, existing approaches optimize rea…