Beir
PulseAugur coverage of Beir — every cluster mentioning Beir across labs, papers, and developer communities, ranked by signal.
12 day(s) with sentiment data
-
Decision models show promise as calibrated relevance scorers in IR research
A new research paper explores the effectiveness of decision models, specifically the Jev model, as calibrated relevance scorers in information retrieval. The study reproduced previous findings on FLAN-UL2 and BEIR colle…
-
Yuanbao-Reranker models achieve state-of-the-art in text reranking
Researchers have developed Yuanbao-Reranker, a new family of dense rerankers built on the Qwen3 backbone, available in 0.6B, 4B, and 8B parameter scales. These models are designed for high-precision text reranking in se…
-
New MV-IVF index boosts multi-vector retrieval efficiency
Researchers have developed MV-IVF, a new inverted-file index designed for multi-vector retrieval. This system utilizes a novel clustering algorithm, MV-Lloyd, to efficiently index sets of embeddings, which represent que…
-
DecisionRank model achieves SOTA retrieval with faster, distilled decision models
Researchers have developed DecisionRank, a novel decision model that significantly improves retrieval performance by using on-policy distillation to transfer knowledge from larger generative language models. This method…
-
New LLM-IRA method improves retrieval ranking aggregation
Researchers have developed LLM-IRA, a novel method for aggregating rankings from multiple retrieval systems, particularly useful for RAG applications. This approach utilizes large language models for pairwise judgments …
-
New method enhances AI retrieval embedding compression
Researchers have developed a new method called Relevance-Guided Residual Calibration (RRC) to improve the spectral compression of dense retrieval embeddings. This technique uses relevance judgments to refine spectral we…
-
LLMs applied to information retrieval and recommendation systems · 2 sources tracked
Two new research papers explore the application of large language models (LLMs) in information retrieval and recommendation systems. The first paper, GistRank, proposes a method for efficient listwise ranking by compres…
-
System One models not better rerankers than fine-tuned alternatives, study finds
A new research paper published on arXiv suggests that "System One" models, also known as decision models, are not superior rerankers compared to models specifically fine-tuned for reranking tasks. The study, which teste…
-
Research paper reveals 'fusion tax' in hybrid AI retrieval methods
A new research paper titled "The Fusion Tax" analyzes the drawbacks of hybrid retrieval methods, which combine sparse and dense retrieval techniques. The study found that while hybrid retrieval often shows aggregate gai…
-
FLINT architecture boosts retrieval model efficiency and scalability
Researchers have introduced FLINT (Fixed-budget Late INTeraction), a novel architecture designed to improve the scalability of multi-vector retrieval models. FLINT employs attention-based pooling to represent queries an…
-
New UIBK-Embedding model unifies retrieval methods, rivals large models
Researchers have developed UIBK-Embedding-v1, a new 179M-parameter English encoder designed to unify dense and late-interaction retrieval methods for more effective text retrieval. This model produces both representatio…
-
New FormBench benchmark reveals poor retrieval performance from commercial AI APIs
A new benchmark called FormBench has been introduced to evaluate the fine-grained retrieval capabilities of models in formulation patents. The benchmark includes over 55,000 queries and two corpus variants, designed to …
-
New GRaSP system optimizes geo-distributed RAG latency and cost
Researchers have developed GRaSP, a novel re-scoring layer for retrieval-augmented generation (RAG) systems designed to optimize performance in geo-distributed infrastructures. This system addresses challenges like wide…
-
New research explores LLM integration with sparse retrieval systems
Two new research papers explore methods for improving large language model (LLM) integration with sparse retrieval systems. The first paper, "Aligning LLMs with Sparse Retrieval via Adaptive Sparse Supervision," propose…
-
UltraBERT advances encoder performance on classification and retrieval tasks
Researchers have developed UltraBERT, a novel encoder model that significantly improves performance on both classification and retrieval tasks. Through an investigation into BERT's pre-training recipe, they discovered t…
-
New method recovers retrieval quality lost to document vector compression
Researchers have developed a method to improve the quality of compressed document vectors in late-interaction models. By learning a global linear correction, the method can recover retrieval quality lost due to aggressi…
-
New SAFE-QE method tackles LLM hallucinated terms in query expansion
Researchers have developed a new method called SAFE-QE to address the issue of hallucinated terms generated by large language models (LLMs) during query expansion. This training-free, retriever-agnostic approach identif…
-
RAGFlip paper measures negative flips in retriever upgrades
A new research paper, RAGFlip, introduces a method for evaluating retriever upgrades by focusing on query-level negative flips. These flips occur when a new retriever fails to find a relevant passage that a previous ret…
-
New method optimizes LLM-based retrieval models for efficiency
Researchers have developed a new method to optimize the sparsity of neural sparse retrieval models, which utilize Large Language Models (LLMs) for semantic term expansion. This approach combines adaptive strategies like…
-
New methods boost visual document reranking efficiency and accuracy
Researchers have developed two novel methods, RenderRank and RidgeRank, for efficient visual document reranking. RenderRank utilizes compressed visual tokens derived from document images to learn query-dependent relevan…