LLM-based dense retrievers generalize better when instruction-tuned but pay a specialization tax when optimized for reasoning; they resist typos and corpus poisoning better than encoder-only baselines yet remain vulnerable to semantic perturbations, with larger models and certain embedding geometry,
Revisiting Representation Degeneration Problem in Language Modeling,
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
citation-role summary
background 1
citation-polarity summary
years
2026 2roles
background 1polarities
background 1representative citing papers
A multi-agent system with hybrid RAG and two new enforcement mechanisms shows strong results on semantic extraction phases of IT-Grundschutz but weak results on logical reasoning phases when evaluated against a BSI case study.
citing papers explorer
-
On the Robustness of LLM-Based Dense Retrievers: A Systematic Analysis of Generalizability and Stability
LLM-based dense retrievers generalize better when instruction-tuned but pay a specialization tax when optimized for reasoning; they resist typos and corpus poisoning better than encoder-only baselines yet remain vulnerable to semantic perturbations, with larger models and certain embedding geometry,
-
Probabilistic Agents in Deterministic Audits: Evaluating Multi-Agent Systems for Automated Audits Based on the German IT-Grundschutz
A multi-agent system with hybrid RAG and two new enforcement mechanisms shows strong results on semantic extraction phases of IT-Grundschutz but weak results on logical reasoning phases when evaluated against a BSI case study.