#retrieval
17 briefs tagged #retrieval.
Paper studies memory transfer across language models
The work tests frozen learned memories moved between model backbones using target-side reader adaptation.
Sentence Transformers adds multi-vector retrieval support
Hugging Face details ColBERT-style late interaction in Sentence Transformers v6.0.
Researchers push structured memory for AI agents
Six new papers propose schema, graph and skill-based systems for more reliable long-term agent memory.
Researchers target memory failures in long-term LLM agents
New papers propose graph, segment and interference methods for maintaining agent memory.
Researchers target reliability gaps in search agents
New papers propose evidence grounding, memory, training data and workflow methods for LLM agents.
New papers target long-document VQA bottlenecks
InSight-doc, DocAtlas and DocMemo propose adaptive ways to find evidence across visually rich documents.
Researchers propose ReMEMBER for dialogue summaries
The framework targets missing evidence in streaming dialogue summarization under fixed memory budgets.
Researchers target visual evidence gaps in VLMs
New arXiv papers propose evidence selection, retrieval and token-pruning methods for multimodal QA.
Researchers target credit assignment for search agents
RICE-PO and ABSeeker propose finer-grained training signals for retrieval and long-horizon search agents.
LoT improves KB-VQA accuracy without model training
The framework highlights image regions and text evidence before answer generation.
Researchers adapt Nemotron retrieval stack for Modern Greek
The work adds Greek retrieval training, reranking, grounded generation and a HERA benchmark.
Researchers target credit assignment for search agents
New papers propose denser training signals for multi-step search, retrieval and reasoning agents.
Benchmarks test whether LLM agents can improve skills
New papers evaluate skill learning, retrieval and memory in self-evolving LLM agents.
Researchers flag ALiBi numerical failure in attention
The paper says floating-point underflow can make some ALiBi attention heads partially blind.
Legal IR papers target leakage in case-law retrieval
New work proposes temporally fenced retrieval and a CJEU dataset for more realistic legal search evaluation.
Researchers release UEmbed multimodal embeddings
UEmbed generates sparse lexical and dense representations in one decoder-only forward pass.
New papers target long-horizon agent memory
Researchers propose harnesses and memory designs for agents that must work across extended tasks.