HAKARI-Bench reconstructs 35 benchmarks into 551 tasks across 43 languages, reproducing full MTEB, MMTEB, and BEIR rankings with Spearman correlation above 0.97 while supporting efficiency variant comparisons.
hub Mixed citations
Nomic Embed: Training a Reproducible Long Context Text Embedder
Mixed citation behavior. Most common role is baseline (38%).
37
Pith papers citing it
25
external citations · Pith
Baseline
38% of classified citations
abstract
This technical report describes the training of nomic-embed-text-v1, the first fully reproducible, open-source, open-weights, open-data, 8192 context length English text embedding model that outperforms both OpenAI Ada-002 and OpenAI text-embedding-3-small on the short-context MTEB benchmark and the long context LoCo benchmark. We release the training code and model weights under an Apache 2.0 license. In contrast with other open-source models, we release the full curated training data and code that allows for full replication of nomic-embed-text-v1. You can find code and data to replicate the model at https://github.com/nomic-ai/contrastors.
hub tools
citation-role summary
baseline 3
method 3
background 1
other 1
citation-polarity summary
representative citing papers
Proves Voronoi complexity equals sign-rank for top-1 retrieval, introduces CUS diagnostic predicting retrieval failure at AUC >0.8 without labels, and AT-DW-InfoNCE objective with derived alpha^*=2.0 that improves Recall@100 on synthetic data.
OpenIIR provides a shared core and pluggable interface for running reproducible multi-agent simulations of information retrieval using LLM personas in four defined study archetypes.
XGRAG uses graph perturbations to quantify component contributions in GraphRAG and achieves 14.81% better explanation quality than text-based baselines on QA datasets, with correlations to graph centrality.
A new auditing framework measures how much of the public's submitted viewpoints is lost in AI-generated consultation summaries, finding that official summaries represent the population worse than a random set of participants and that critical voices are most likely excluded.
MARVEL reaches 37.9 nDCG@10 on the MM-BRIGHT benchmark by combining LLM query expansion, a reasoning-enhanced dense retriever, and GPT-4o CoT reranking, beating prior multimodal encoders by 10.3 points.
Presents Invoice Haystack benchmark for homogeneous document retrieval and VL-RAG hybrid framework achieving 60% Recall@1 and up to 13.5 point gains over prior methods.
PACMS uses submodular context selection as a pluggable engine to choose relevant context from conversation, memory, and tool outputs instead of recency truncation.
HistoRAG embeds historiographical principles into RAG via temporal windowing, decoupled retrieval, and contestable LLM relevance judgments, evaluated on 102k Der Spiegel articles from 1950-1979.
Constrained Dominant Sets on query-augmented graphs select complementary evidence for long multimodal document QA, claiming new SOTA of 66.99 on VisDoMBench and gains of 37.1 and 4.8 points over baselines.
Covariance structure and coordinate heterogeneity in InfoNCE embeddings control binary quantization fidelity, with off-diagonals contributing 30-50% of signal and heterogeneity determining rotation benefit and bit utility under a Gaussian model.
LRD framework with Frenet, NRS, and GFMI metrics shows layer-wise structure in 31 models provides usable signal for model selection and pruning on MTEB tasks.
TrajPrism introduces a multi-task benchmark with 300K real-world urban trajectories and 2.1M language-grounded task instances across three cities, plus proof-of-concept models showing large gaps versus geometry-only baselines.
Discriminative factorization distinguishes high-quality query sets for black-box model classification, with chance-level error decaying exponentially in query budget and parameters predicting empirical decay rates on auditing tasks.
MLAIRE is a protocol that evaluates multilingual retrievers on both semantic accuracy and query-language preference using parallel passages and new metrics like LPR and Lang-nDCG, showing that standard metrics hide distinct behavioral differences among retrievers.
Machine translation preserves embedding similarity structure for ten languages but distorts it for four in the Manifesto Corpus, via a new non-inferiority testing framework.
LLMs corrupt an average of 25% of document content during long delegated editing workflows across 52 domains, even frontier models, and agentic tools do not mitigate the issue.
ASTRA combines an eight-axis conceptual framework with text embeddings and unsupervised clustering to map and group 78 art-technology institutions into coherent thematic clusters.
SitEmb-v1.5 uses a new training paradigm to produce context-situated embeddings for short chunks, outperforming larger models by over 10% on a curated book-plot retrieval benchmark.
MINT defines multi-vector search index tuning and provides algorithms that achieve 2.1X to 8.3X latency speedup over baselines under storage and recall constraints.
ArchRAG proposes attributed-community hierarchical indexing and LLM clustering to improve accuracy and lower token usage in graph-based retrieval-augmented generation.
Mid-scale MLLMs reach 62-88% object-level exact-match accuracy in zero-shot localized concept naming via closed-set prompting and an embedding-based Open-CoNa strategy across datasets.
Multi-Prefix Embedding extracts per-chunk embeddings from a single forward pass over EOS-separated document chunks and matches via MaxSim while training only on document-level labels.
WebKnoGraph is a framework that represents websites as graphs, scores internal links via GraphSAGE, and compares automatic versus expert-assisted link selection on authority and coherence metrics in empirical and synthetic host graphs.
citing papers explorer
-
HAKARI-Bench: A Lightweight Benchmark for Comparing Retrieval Architectures and Efficiency Settings under Unified Conditions
HAKARI-Bench reconstructs 35 benchmarks into 551 tasks across 43 languages, reproducing full MTEB, MMTEB, and BEIR rankings with Spearman correlation above 0.97 while supporting efficiency variant comparisons.
-
The Voronoi Bottleneck: Capacity-Aware Dense Retrieval for Product Search
Proves Voronoi complexity equals sign-rank for top-1 retrieval, introduces CUS diagnostic predicting retrieval failure at AUC >0.8 without labels, and AT-DW-InfoNCE objective with derived alpha^*=2.0 that improves Recall@100 on synthetic data.
-
OpenIIR: An Open Simulation Platform for Information Retrieval Research
OpenIIR provides a shared core and pluggable interface for running reproducible multi-agent simulations of information retrieval using LLM personas in four defined study archetypes.
-
XGRAG: A Graph-Native Framework for Explaining KG-based Retrieval-Augmented Generation
XGRAG uses graph perturbations to quantify component contributions in GraphRAG and achieves 14.81% better explanation quality than text-based baselines on QA datasets, with correlations to graph centrality.
-
Participatory provenance as representational auditing for AI-mediated public consultation
A new auditing framework measures how much of the public's submitted viewpoints is lost in AI-generated consultation summaries, finding that official summaries represent the population worse than a random set of participants and that critical voices are most likely excluded.
-
MARVEL: Multimodal Adaptive Reasoning-intensiVe Expand-rerank and retrievaL
MARVEL reaches 37.9 nDCG@10 on the MM-BRIGHT benchmark by combining LLM query expansion, a reasoning-enhanced dense retriever, and GPT-4o CoT reranking, beating prior multimodal encoders by 10.3 points.
-
Invoice Haystack: Benchmarking Document Retrieval and Visual Question Answering Under Strong Visual Homogeneity
Presents Invoice Haystack benchmark for homogeneous document retrieval and VL-RAG hybrid framework achieving 60% Recall@1 and up to 13.5 point gains over prior methods.
-
PACMS: Submodular Context Selection as a Pluggable Engine for LLM Agents
PACMS uses submodular context selection as a pluggable engine to choose relevant context from conversation, memory, and tool outputs instead of recency truncation.
-
HistoRAG: Embedding Historical Methodology in Retrieval-Augmented Generation Through Critical Technical Practice
HistoRAG embeds historiographical principles into RAG via temporal windowing, decoupled retrieval, and contestable LLM relevance judgments, evaluated on 102k Der Spiegel articles from 1950-1979.
-
Constrained Dominant Sets for Multimodal Document Question Answering
Constrained Dominant Sets on query-augmented graphs select complementary evidence for long multimodal document QA, claiming new SOTA of 66.99 on VisDoMBench and gains of 37.1 and 4.8 points over baselines.
-
Covariance Structure and Coordinate Heterogeneity Govern Binary Quantization of Contrastive Embeddings
Covariance structure and coordinate heterogeneity in InfoNCE embeddings control binary quantization fidelity, with off-diagonals contributing 30-50% of signal and heterogeneity determining rotation benefit and bit utility under a Gaussian model.
-
Layer-wise Representation Dynamics: An Empirical Investigation Across Embedders and Base LLMs
LRD framework with Frenet, NRS, and GFMI metrics shows layer-wise structure in 31 models provides usable signal for model selection and pruning on MTEB tasks.
-
TrajPrism: A Multi-Task Benchmark for Language-Grounded Urban Trajectory Understanding
TrajPrism introduces a multi-task benchmark with 300K real-world urban trajectories and 2.1M language-grounded task instances across three cities, plus proof-of-concept models showing large gaps versus geometry-only baselines.
-
Black-box model classification under the discriminative factorization
Discriminative factorization distinguishes high-quality query sets for black-box model classification, with chance-level error decaying exponentially in query budget and parameters predicting empirical decay rates on auditing tasks.
-
MLAIRE: Multilingual Language-Aware Information Retrieval Evaluation Protocal
MLAIRE is a protocol that evaluates multilingual retrievers on both semantic accuracy and query-language preference using parallel passages and new metrics like LPR and Lang-nDCG, showing that standard metrics hide distinct behavioral differences among retrievers.
-
Is Textual Similarity Invariant under Machine Translation? Evidence Based on the Political Manifesto Corpus
Machine translation preserves embedding similarity structure for ten languages but distorts it for four in the Manifesto Corpus, via a new non-inferiority testing framework.
-
LLMs Corrupt Your Documents When You Delegate
LLMs corrupt an average of 25% of document content during long delegated editing workflows across 52 domains, even frontier models, and agentic tools do not mitigate the issue.
-
ASTRA: Mapping Art-Technology Institutions via Conceptual Axes, Text Embeddings, and Unsupervised Clustering
ASTRA combines an eight-axis conceptual framework with text embeddings and unsupervised clustering to map and group 78 art-technology institutions into coherent thematic clusters.
-
SitEmb-v1.5: Improved Context-Aware Dense Retrieval for Semantic Association and Long Story Comprehension
SitEmb-v1.5 uses a new training paradigm to produce context-situated embeddings for short chunks, outperforming larger models by over 10% on a curated book-plot retrieval benchmark.
-
MINT: Multi-Vector Search Index Tuning
MINT defines multi-vector search index tuning and provides algorithms that achieve 2.1X to 8.3X latency speedup over baselines under storage and recall constraints.
-
ArchRAG: Attributed Community-based Hierarchical Retrieval-Augmented Generation
ArchRAG proposes attributed-community hierarchical indexing and LLM clustering to improve accuracy and lower token usage in graph-based retrieval-augmented generation.
-
Low-cost concept-based localized explanations: How far can we get with training-free approaches?
Mid-scale MLLMs reach 62-88% object-level exact-match accuracy in zero-shot localized concept naming via closed-set prompting and an embedding-based Open-CoNa strategy across datasets.
-
Improving Long-Context Retrieval with Multi-Prefix Embedding
Multi-Prefix Embedding extracts per-chunk embeddings from a single forward pass over EOS-separated document chunks and matches via MaxSim while training only on document-level labels.
-
WebKnoGraph: GNN-Powered Internal Linking
WebKnoGraph is a framework that represents websites as graphs, scores internal links via GraphSAGE, and compares automatic versus expert-assisted link selection on authority and coherence metrics in empirical and synthetic host graphs.
-
S-SPPO: Semantic-Calibrated Self-Play Preference Optimization
S-SPPO stabilizes SPPO via semantic calibration in supervision and representation spaces, reporting 52.19% win rate on AlpacaEval 2.0 with Llama-3-8B.
-
Jailbreak susceptibility prediction and mitigation via the behavioral geometry of models
Behavioral geometry of model populations enables high-accuracy jailbreak susceptibility prediction and defense transfer with 98% fewer evaluations.
-
Benchmarking Patent Embeddings: A Multi-Task Evaluation of 22 Models Across Retrieval, Classification, and Clustering
Multi-task evaluation of 22 patent embedding models finds task-specific fine-tuning benefits and significant cross-landscape retrieval degradation that cannot be fixed by hybrid fusion.
-
GraphRAG on Consumer Hardware: Benchmarking Local LLMs for Healthcare EHR Schema Retrieval
GraphRAG with 7-8B local LLMs on 8GB VRAM hardware builds knowledge graphs from EHR docs and answers queries, with Llama 3.1 creating the largest graph, Qwen 2.5 scoring highest on quality, and models below ~7B failing to complete the pipeline.
-
Control Charts for Multi-agent Systems
Adaptive control charts can monitor learning multi-agent systems but are vulnerable to gradual adversarial defection, revealing a fundamental tradeoff between allowing agents to learn and maintaining security against adversaries.
-
Health System Scale Semantic Search Across Unstructured Clinical Notes
Production semantic search over 166M clinical notes achieves sub-second latency at ~$4k/month and recovers 98% of genetic-disease patients versus ≤75% by ICD-10 codes.
-
Towards Platonic Representation for Table Reasoning: A Foundation for Permutation-Invariant Retrieval
Table representations must be permutation-invariant to preserve semantic structure, and a new header-aligned encoder moves toward this ideal while exposing fragility in existing LLM table embeddings.
-
BRIDGE: Multimodal-to-Text Retrieval via Reinforcement-Learned Query Alignment
BRIDGE reaches 29.7 nDCG@10 on MM-BRIGHT by RL-aligning multimodal queries to text and using a reasoning retriever, beating multimodal encoders and, when combined with Nomic-Vision, exceeding the best text-only retriever at 33.3.
-
Gaussian mixture models as a proxy for interacting language models
Interacting Gaussian mixture models with RAG-style updates are shown to mimic aspects of interacting LLMs and are used to prove lower bounds on polarization probability in the resulting Markov chain.
-
Smarter, Better, Faster, Longer: A Modern Bidirectional Encoder for Fast, Memory Efficient, and Long Context Finetuning and Inference
ModernBERT is a new bidirectional encoder model achieving SOTA performance on diverse classification and retrieval benchmarks while offering superior speed and memory efficiency for long-context inference.
-
A Multi-Agent Audit Framework for High-Stakes Reasoning: Evaluation and Interpretability in Clinical Mental Health Screening
Multi-agent audit framework reduces MAE for PHQ-8 depression prediction from 5.35 to 5.02 on DAIC-WOZ using open-source LLMs with open code.
-
ESOFinder: an LLM-powered tool to help users navigate ESO documentation
ESOFinder is an in-house RAG-based LLM chatbot that integrates over 100 ESO manuals and 3500 links to give context-aware answers on proposals, observations, data reduction, and archiving while running on open-source models for privacy.
- PDE-Agents: An LLM-Orchestrated Multi-Agent Framework for Automated Finite Element Simulations with Knowledge Graph-Augmented Reasoning