long-context-qa

Tag

Cards List
#long-context-qa

SeDeM: Selective Decompression of Hidden-State Memories for Long-Context Question Answering

arXiv cs.CL · 2026-08-04 Cached

SeDeM is a selective decompression framework that stores long-context hidden states in a compact memory bank and decompresses only query-relevant blocks for decoder conditioning, improving QA accuracy and efficiency over compression baselines.

0 favorites 0 likes
#long-context-qa

REFACT: Adaptive Fact Restatement for Compact and Faithful Chain-of-Thought Reasoning

arXiv cs.CL · 2026-07-24 Cached

ReFact proposes an adaptive fact-restatement citation framework that trains LLMs to decide when reasoning steps need contextual grounding, improving faithfulness and compactness in chain-of-thought reasoning while reducing token consumption.

0 favorites 0 likes
← Back to home

Submit Feedback