Proves that for bounded n, there exist MV embeddings with m vectors each whose Chamfer matrix requires single-vector dimension D = (ε² m)^Ω(1/ε) to approximate within ε, separating MV from SV expressiveness.
hub Mixed citations
MTEB: Massive Text Embedding Benchmark
Mixed citation behavior. Most common role is background (67%).
57
Pith papers citing it
66
external citations · Pith
Background
67% of classified citations
abstract
Text embeddings are commonly evaluated on a small set of datasets from a single task not covering their possible applications to other tasks. It is unclear whether state-of-the-art embeddings on semantic textual similarity (STS) can be equally well applied to other tasks like clustering or reranking. This makes progress in the field difficult to track, as various models are constantly being proposed without proper evaluation. To solve this problem, we introduce the Massive Text Embedding Benchmark (MTEB). MTEB spans 8 embedding tasks covering a total of 58 datasets and 112 languages. Through the benchmarking of 33 models on MTEB, we establish the most comprehensive benchmark of text embeddings to date. We find that no particular text embedding method dominates across all tasks. This suggests that the field has yet to converge on a universal text embedding method and scale it up sufficiently to provide state-of-the-art results on all embedding tasks. MTEB comes with open-source code and a public leaderboard at https://github.com/embeddings-benchmark/mteb.
hub tools
citation-role summary
background 8
dataset 4
citation-polarity summary
representative citing papers
Tabular foundation models excel on tiny- to medium-sized IID data but are outperformed by traditional tree-based and deep learning models on non-IID, large, and high-dimensional datasets, based on evaluations across 11 models and 142 datasets in the new BeyondArena benchmark.
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.
An empirical comparison of thirteen control-plane placements in agent memory pipelines identifies three regimes with complementary forgetting recovery on a new 385-case adversarial benchmark, with mutation-time placement achieving 91.7-93.2% overall.
A 527-item GDPR-aligned privacy preference item bank was developed by extracting 669 statements from 99 GDPR articles and validating them through multi-round expert consensus and semantic clustering.
Chronicle is the first model jointly pretrained from scratch on text and time series in a unified transformer that matches a comparable language model on NLU tasks and sets new bars for time series classification and multimodal forecasting.
AcquisitionSynthesis uses acquisition functions as rewards to train generators that produce higher-quality synthetic data, delivering 2-7% gains on math, medical QA, and coding tasks with improved robustness to forgetting.
Large-scale analysis of unfiltered search queries shows geospatial intent at 18% of total, dominated by transactional categories outside traditional GIS scope.
CRAFT is a supervised LLM framework using retrieval-augmented generation, self-refinement, fine-tuning, and preference optimization to create fluent adversarial content that boosts target ranks in neural ranking models, outperforming baselines on MS MARCO and TREC benchmarks with cross-architecture
mEOL creates aligned embeddings for text, images, and SVGs using instruction-guided MLLM one-word summaries and semantic SVG rewriting, outperforming baselines on a new text-to-SVG retrieval benchmark.
DualGuard uses adaptive dual-stream watermark signals to detect and trace both paraphrase and spoofing attacks in LLM outputs while preserving text quality.
UWE is a task-agnostic bi-encoder that uses many-to-many InfoNCE and token-level soft late interaction to achieve zero-shot ranking across unseen work-related target spaces while using far fewer parameters than Qwen3-8B and improving MAP by 4.4 points.
Multi-Level Optimal Transport (MOT) jointly infers soft layer couplings and neuron transport plans to produce global alignment scores and structured hierarchical correspondences between networks of varying depths.
C-Pack releases a new Chinese embedding benchmark, large training dataset, and optimized models that outperform priors by up to 10% on C-MTEB while also delivering English SOTA results.
A scale-free gain rule applied to variational ELBO paths recovers true factor dimensionality in partially exploratory factor analysis where raw information criteria over-factor.
Graph-PRefLexOR fine-tunes graph-native models with GRPO to organize reasoning into phases, yielding 40-65% gains in traceable hypothesis generation and 2-3x semantic diversity on 100 materials science questions.
A per-component SimHash fingerprint supplies structural identity for AI agent skills, recovering family membership under paraphrase and refactoring with AUC 0.974 while localizing changes.
Large-scale HPC evaluation of Qdrant, Milvus, and Weaviate reveals that workload patterns limit scaling and extra cores can reduce throughput, exposing a cloud-to-HPC design mismatch.
Meta-study of MTEB rankings introduces dataset-composition and ranking-scheme robustness indicators and finds only a small subset of models stay consistently strong across tasks, languages, and evaluation variations.
A sliced IGW distance is introduced with closed-form 1D expressions, rotational invariance, and studied structural and computational properties for efficient data alignment.
DKPS-based methods predict new model benchmark scores using cached responses, matching baseline mean absolute error with substantially fewer queries and an offline query selection approach.
Four axioms (Causality, Minimality, Separability, Stability) are formalized for latent thought representations; audits of open LLMs on 23 tasks show none satisfy all four and representations add little beyond input embeddings.
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.
JU'A is a new heterogeneous benchmark for Brazilian legal IR that distinguishes retrieval methods and shows domain-adapted models excel on aligned subsets while BM25 stays competitive elsewhere.
citing papers explorer
-
Multi-Vector Embeddings are Provably More Expressive than Single Vector Embeddings
Proves that for bounded n, there exist MV embeddings with m vectors each whose Chamfer matrix requires single-vector dimension D = (ε² m)^Ω(1/ε) to approximate within ε, separating MV from SV expressiveness.
-
Beyond IID: How General Are Tabular Foundation Models, Really?
Tabular foundation models excel on tiny- to medium-sized IID data but are outperformed by traditional tree-based and deep learning models on non-IID, large, and high-dimensional datasets, based on evaluations across 11 models and 142 datasets in the new BeyondArena benchmark.
-
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.
-
Control-Plane Placement Shapes Forgetting: An Architectural Study of Agent Memory Across Thirteen System Configurations
An empirical comparison of thirteen control-plane placements in agent memory pipelines identifies three regimes with complementary forgetting recovery on a new 385-case adversarial benchmark, with mutation-time placement achieving 91.7-93.2% overall.
-
Modernizing User Privacy Preference Measurement through GPPI: A GDPR-aligned Privacy Preference Item Bank
A 527-item GDPR-aligned privacy preference item bank was developed by extracting 669 statements from 99 GDPR articles and validating them through multi-round expert consensus and semantic clustering.
-
Chronicle: A Multimodal Foundation Model for Joint Language and Time Series Understanding
Chronicle is the first model jointly pretrained from scratch on text and time series in a unified transformer that matches a comparable language model on NLU tasks and sets new bars for time series classification and multimodal forecasting.
-
AcquisitionSynthesis: Targeted Data Generation using Acquisition Functions
AcquisitionSynthesis uses acquisition functions as rewards to train generators that produce higher-quality synthetic data, delivering 2-7% gains on math, medical QA, and coding tasks with improved robustness to forgetting.
-
Much of Geospatial Web Search Is Beyond Traditional GIS
Large-scale analysis of unfiltered search queries shows geospatial intent at 18% of total, dominated by transactional categories outside traditional GIS scope.
-
Led to Mislead: Adversarial Content Injection for Attacks on Neural Ranking Models
CRAFT is a supervised LLM framework using retrieval-augmented generation, self-refinement, fine-tuning, and preference optimization to create fluent adversarial content that boosts target ranks in neural ranking models, outperforming baselines on MS MARCO and TREC benchmarks with cross-architecture
-
mEOL: Training-Free Instruction-Guided Multimodal Embedder for Vector Graphics and Image Retrieval
mEOL creates aligned embeddings for text, images, and SVGs using instruction-guided MLLM one-word summaries and semantic SVG rewriting, outperforming baselines on a new text-to-SVG retrieval benchmark.
-
DualGuard: Dual-stream Large Language Model Watermarking Defense against Paraphrase and Spoofing Attack
DualGuard uses adaptive dual-stream watermark signals to detect and trace both paraphrase and spoofing attacks in LLM outputs while preserving text quality.
-
Unified Work Embeddings: Contrastive Learning of a Bidirectional Multi-task Ranker
UWE is a task-agnostic bi-encoder that uses many-to-many InfoNCE and token-level soft late interaction to achieve zero-shot ranking across unseen work-related target spaces while using far fewer parameters than Qwen3-8B and improving MAP by 4.4 points.
-
Representational Alignment Across Model Layers and Brain Regions with Multi-Level Optimal Transport
Multi-Level Optimal Transport (MOT) jointly infers soft layer couplings and neuron transport plans to produce global alignment scores and structured hierarchical correspondences between networks of varying depths.
-
C-Pack: Packed Resources For General Chinese Embeddings
C-Pack releases a new Chinese embedding benchmark, large training dataset, and optimized models that outperform priors by up to 10% on C-MTEB while also delivering English SOTA results.
-
Recovering Latent Structures after Variational Bayesian Variable Selection: Fit Assessment and Factor-Number Selection in Partially Exploratory Factor Analysis
A scale-free gain rule applied to variational ELBO paths recovers true factor dimensionality in partially exploratory factor analysis where raw information criteria over-factor.
-
Graph-Native Reinforcement Learning Enables Traceable Scientific Hypothesis Generation through Conceptual Recombination
Graph-PRefLexOR fine-tunes graph-native models with GRPO to organize reasoning into phases, yielding 40-65% gains in traceable hypothesis generation and 2-3x semantic diversity on 100 materials science questions.
-
The Decomposition Is the Fingerprint: Per-Component Identity for Agent Skills
A per-component SimHash fingerprint supplies structural identity for AI agent skills, recovering family membership under paraphrase and refactoring with AUC 0.974 while localizing changes.
-
When More Cores Hurts: The Vector Database Scaling Paradox in HPC
Large-scale HPC evaluation of Qdrant, Milvus, and Weaviate reveals that workload patterns limit scaling and extra cores can reduce throughput, exposing a cloud-to-HPC design mismatch.
-
On the Robustness of Multilingual Text Embedding Rankings Across Learning Tasks, Languages, and Benchmark Datasets
Meta-study of MTEB rankings introduces dataset-composition and ranking-scheme robustness indicators and finds only a small subset of models stay consistently strong across tasks, languages, and evaluation variations.
-
Sliced Inner Product Gromov-Wasserstein Distances
A sliced IGW distance is introduced with closed-form 1D expressions, rotational invariance, and studied structural and computational properties for efficient data alignment.
-
Query-efficient model evaluation using cached responses
DKPS-based methods predict new model benchmark scores using cached responses, matching baseline mean absolute error with substantially fewer queries and an offline query selection approach.
-
Formalizing Latent Thoughts: Four Axioms of Thought Representation in LLMs
Four axioms (Causality, Minimality, Separability, Stability) are formalized for latent thought representations; audits of open LLMs on 23 tasks show none satisfy all four and representations add little beyond input embeddings.
-
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.
-
JU\'A -- A Benchmark for Information Retrieval in Brazilian Legal Text Collections
JU'A is a new heterogeneous benchmark for Brazilian legal IR that distinguishes retrieval methods and shows domain-adapted models excel on aligned subsets while BM25 stays competitive elsewhere.
-
Semantic Data Processing with Holistic Data Understanding
HoldUp uses LLM-guided clustering to provide holistic dataset context for semantic operators, yielding up to 33% higher classification accuracy and 30% higher scoring accuracy than row-by-row LLM processing across 15 datasets.
-
Mitigating Membership Inference in Intermediate Representations with Differentially Private Training
LM-DP-SGD estimates layer-specific MIA risks from shadow models and reweights gradients to give stronger protection to vulnerable layers, improving the privacy-utility trade-off over uniform DP-SGD.
-
SHINE: A Scalable In-Context Hypernetwork for Mapping Context to LoRA in a Single Pass
SHINE generates LoRA adapters from a document in one forward pass, letting a frozen LLM answer questions about the document without the document in its context.
-
Efficient-DLM: From Autoregressive to Diffusion Language Models, and Beyond in Speed
Efficient-DLM converts AR models to dLMs via block-wise causal attention and position-dependent masking, yielding higher accuracy and 2.7-4.5x throughput than Dream 7B and Qwen3 4B.
-
LLM-MemCluster: Empowering Large Language Models with Dynamic Memory for Text Clustering
LLM-MemCluster gives LLMs stateful memory and prompts that let them decide cluster count and iteratively refine groupings, outperforming baselines on benchmarks in a tuning-free end-to-end setup.
-
Detecting LLM-Generated Spam Reviews by Integrating Language Model Embeddings and Graph Neural Network
Introduces FraudSquad, a hybrid model using language model embeddings and a gated graph transformer that outperforms baselines on newly created LLM-generated spam review datasets.
-
EmbeddingGemma: Powerful and Lightweight Text Representations
A 300M-parameter open embedding model sets new SOTA on MTEB for its size class and matches models twice as large while staying effective when compressed.
-
NV-Embed: Improved Techniques for Training LLMs as Generalist Embedding Models
NV-Embed achieves first place on the MTEB leaderboard across 56 tasks by combining a latent attention layer, causal-mask removal, two-stage contrastive training, and data curation for LLM-based embedding models.
-
StarCoder 2 and The Stack v2: The Next Generation
StarCoder2-15B matches or beats CodeLlama-34B on code tasks despite being smaller, and StarCoder2-3B outperforms prior 15B models, with open weights and exact training data identifiers released.
-
Scaling Data-Constrained Language Models
Repeating training data up to 4 epochs yields negligible loss increase versus unique data for fixed compute, and a new scaling law accounts for the decaying value of repeated tokens and excess parameters.
-
REPLUG: Retrieval-Augmented Black-Box Language Models
REPLUG improves frozen black-box LMs by prepending LM-supervised retrieved documents, delivering 6.3% better language modeling on GPT-3 and 5.1% better five-shot MMLU on Codex.
-
BLOOM: A 176B-Parameter Open-Access Multilingual Language Model
BLOOM is a 176B-parameter open-access multilingual language model trained on the ROOTS corpus that achieves competitive performance on benchmarks, with improved results after multitask prompted finetuning.
-
When Does Complexity Conditioning Help a Frozen Sentence Embedding? A Controlled Study of Per-Sentence and Pair-Level Difficulty Adaptation
Controlled experiments on frozen embeddings show per-sentence difficulty adaptation fails to help while pair-level gating by a held-out signal gives modest gains on larger tasks.
-
Adapting Multilingual Embedding Models to Turkish via Cross-Lingual Tokenizer Surgery and Offline Distillation
A 200M-parameter Turkish sentence embedding model is adapted from a multilingual teacher via tokenizer pruning, mean-composition initialization, and offline cosine distillation, achieving 77.55% Pearson correlation on STSbTR and 7th place on TR-MTEB.
-
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.
-
CALMem : Application-Layer Dual Memory for Conversational AI
CALMem delivers virtually unbounded effective context for LLM conversations via an application-layer dual memory architecture with intra-session retrieval and token-adaptive injection.
-
Measuring Embedding Sensitivity to Authorial Style in French: Comparing Literary Texts with Language Model Rewritings
Embeddings reliably capture authorial stylistic features in French literary texts, and these signals persist after LLM rewriting while showing model-specific patterns.
-
How Does Chunking Affect Retrieval-Augmented Code Completion? A Controlled Empirical Study
Function-based chunking underperforms other strategies in RAG code completion by 3.57-5.64 points, with context length as the dominant factor.
-
MIPIC: Matryoshka Representation Learning via Self-Distilled Intra-Relational and Progressive Information Chaining
MIPIC trains Matryoshka representations using self-distilled intra-relational alignment and progressive information chaining, yielding competitive results on STS, NLI, and classification tasks especially at low dimensions.
-
MMEB-V3: Measuring the Performance Gaps of Omni-Modality Embedding Models
On 190 tasks and a 12-direction cross-modal diagnostic, seven embedding models frequently fail to honor explicit target-modality instructions: retrieval is biased toward the query modality and instruction-induced shifts are misaligned.
-
Towards Better Static Code Analysis Reports: Sentence Transformer-based Filtering of Non-Actionable Alerts
STAF applies sentence embeddings from transformers to classify SCA findings, reaching 89% F1 and beating prior filters by 11% within projects and 6% across projects.
-
Text-as-Signal: Quantitative Semantic Scoring with Embeddings, Logprobs, and Noise Reduction
A configurable pipeline turns text corpora into quantitative semantic signals via embeddings, logprobs, and UMAP-based noise reduction for document positioning and corpus profiling.
-
Robustness Risk of Conversational Retrieval: Identifying and Mitigating Noise Sensitivity in Qwen3-Embedding Model
Qwen3-embedding models show noise sensitivity in conversational retrieval where dialogue artifacts rank highly despite lacking semantic value, a problem reduced by query prompting and more severe than in prior Qwen versions or other baselines.
-
GovScape: A Public Multimodal Search System for 70 Million Pages of Government PDFs
GovScape delivers multimodal search over 10 million government PDFs using metadata, exact text, semantic embeddings, and visual page features at an estimated $1,500 preprocessing cost.
-
Search-R3: Unifying Reasoning and Embedding in Large Language Models
Search-R3 trains LLMs to output search embeddings as a direct product of step-by-step reasoning via supervised pre-training and a specialized RL environment that avoids full corpus re-encoding.
-
Verbalized Algorithms: Classical Algorithms are All You Need (Mostly)
Wrapping LLM calls as oracles in classical algorithms improves sorting and clustering accuracy for small models, but several advertised applications are missing from the experiments.