Retrieval from Within: An Intrinsic Capability of Attention-Based Models

arXiv:2605.05806 · cs.LG · Submitted 2026-05-07 · Read on arXiv

cs.LG

Submitted: 2026-05-07

Updated: 2026-09-30

Code: https://github.com/rapidsai/cuvs

Terminology

• RAG — 2× in this paper · explained in 3 episodes
• Benchmark — 1× in this paper · explained in 2 episodes
• Retrieval-Augmented Generation (RAG) — 1× in this paper · explained in 16 episodes

Sources

• Longformer: The Long-Document Transformer
• Rethinking Attention with Performers
• ColPali: Efficient Document Retrieval with Vision Language Models
• Mamba: Linear-Time Sequence Modeling with Selective State Spaces
• CLaRa: Bridging Retrieval and Generation with Continuous Latent Reasoning
• Search-o1: Agentic Search-Enhanced Large Reasoning Models
• Jamba: A Hybrid Transformer-Mamba Language Model
• NoLiMa: Long-Context Evaluation Beyond Literal Matching
• Hyena Hierarchy: Towards Larger Convolutional Language Models
• Agentic Retrieval-Augmented Generation: A Survey on Agentic RAG
• InstructRetro: Instruction Tuning post Retrieval-Augmented Pretraining
• HELMET: How to Evaluate Long-Context Language Models Effectively and Thoroughly
• Big Bird: Transformers for Longer Sequences
• T5Gemma 2: Seeing, Reading, and Understanding Longer
• Qwen3 Embedding: Advancing Text Embedding and Reranking Through Foundation Models

Related papers

• Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
• AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
• Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
• Convergence issues in Relational Concept Analysis based on AOC-posets
• Beliefs Beyond Posteriors: The Predictive Cost of ELBO Binding
• Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks