Weighted InfoNCE objectives realize specific target geometries in embedding space, with SupCon producing size-dependent inter-class similarities under imbalance while Soft SupCon and certain continuous variants preserve regular simplex or unique optima.
super hub Mixed citations
Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks
Mixed citation behavior. Most common role is method (44%).
280
Pith papers citing it
88
external citations · Pith
Method
44% of classified citations
abstract
BERT (Devlin et al., 2018) and RoBERTa (Liu et al., 2019) has set a new state-of-the-art performance on sentence-pair regression tasks like semantic textual similarity (STS). However, it requires that both sentences are fed into the network, which causes a massive computational overhead: Finding the most similar pair in a collection of 10,000 sentences requires about 50 million inference computations (~65 hours) with BERT. The construction of BERT makes it unsuitable for semantic similarity search as well as for unsupervised tasks like clustering. In this publication, we present Sentence-BERT (SBERT), a modification of the pretrained BERT network that use siamese and triplet network structures to derive semantically meaningful sentence embeddings that can be compared using cosine-similarity. This reduces the effort for finding the most similar pair from 65 hours with BERT / RoBERTa to about 5 seconds with SBERT, while maintaining the accuracy from BERT. We evaluate SBERT and SRoBERTa on common STS tasks and transfer learning tasks, where it outperforms other state-of-the-art sentence embeddings methods.
hub tools
citation-role summary
background 17
method 16
other 2
dataset 1
citation-polarity summary
claims ledger
- abstract BERT (Devlin et al., 2018) and RoBERTa (Liu et al., 2019) has set a new state-of-the-art performance on sentence-pair regression tasks like semantic textual similarity (STS). However, it requires that both sentences are fed into the network, which causes a massive computational overhead: Finding the most similar pair in a collection of 10,000 sentences requires about 50 million inference computations (~65 hours) with BERT. The construction of BERT makes it unsuitable for semantic similarity search as well as for unsupervised tasks like clustering. In this publication, we present Sentence-BER
authors
co-cited works
representative citing papers
Benign fine-tuning on audio data breaks safety alignment in Audio LLMs by raising jailbreak success rates up to 87%, with the dominant risk axis depending on model architecture and embedding proximity to harmful content.
VLRS-Bench is the first benchmark dedicated to complex vision-language reasoning in remote sensing, with 2000 QA pairs across 14 tasks in cognition, decision, and prediction dimensions.
DeepMath-103K is a new 103K-problem mathematical dataset with high difficulty, rigorous decontamination, and verifiable answers to support RL training of language-model reasoning.
Mind2Web is the first large-scale dataset of real-world web tasks for developing generalist language-guided agents that complete complex actions on diverse websites.
Pluralis v0.1 is a culture-first, multimodal, multilingual VLM safety benchmark spanning 6 APAC locales with 6,448 prompts and an agreement-gated LLM judge that disentangles safety from cultural appropriateness.
PromptGNN-sim uses GAT-based semantically aware neighborhood selection and structure-aware LLM prompts with bi-directional contrastive alignment to outperform prior GNN, LLM, and fusion methods on text-attributed graph datasets.
MKG-RAG-Bench is a cross-domain benchmark for retrieval in multimodal knowledge graph-augmented generation, constructed via LLM curation from two MKGs with aligned QA datasets.
Introduces CASU benchmark with four tasks to evaluate context-aware auditory scene understanding in LALMs via semi-synthetic audio compositions of speech, events, and environments.
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.
NeuroImprint attack assigns isolated memorization neurons to training samples in PEFT adapters, enabling closed-form reconstruction of 59-79% of samples across BERT, GPT-2, Qwen2, and Llama3.2 on multiple datasets.
NEURRATOR bridges neural spike trains to frozen CLIP patch embeddings via a learned encoder, then uses a multimodal LM and sparse autoencoder to produce validated natural-language narrations of viewed scenes from Neuropixel recordings.
LEDGER provides a corpus of 4,999 annual reports with 31 labeled KPIs and three benchmarks for page-level retrieval, needle-in-haystack lookup, and full KPI extraction from long documents.
RECON integrates LLM reasoning with backward static analysis to extract interpretable execution constraints from Android bytecode, reporting 5.8X speedup and 100% success versus symbolic execution plus 84% success on malware samples.
Re-ranking retrieval candidates via a cross-encoder trained on continuous perturbation-based attribution scores improves citation faithfulness and gold-answer alignment in legal QA over semantic similarity.
GFFMERGE formulates GNN force field merging as a convex embedding-alignment problem with an analytical solution, recovering near joint-training performance on MD17, MD22, LiPS20 and other benchmarks while delivering 5-27x speedups.
Introduces (ε,q,t,A)-behavioral indistinguishability and shows via Qwen/Llama experiments that LoRA distillation boosts semantic similarity but leaves detectable behavioral differences under adversarial evaluation.
SwanBench-Speech is a new benchmark that decomposes long-form speech quality into seven disentangled metrics across 17 scenarios to evaluate generation models.
Spectral Retrieval uses multi-scale sinc convolutions on token embeddings to interpolate between per-token MaxSim and mean-pooling, achieving large gains on synthetic and LIMIT-small benchmarks for localized retrieval.
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.
CAMERA is an ego-decoupled mixture-of-experts model with context-informed gating and one-class objectives for unsupervised fraud detection in text-attributed graphs facing semantic camouflage.
Preference fine-tuning outperforms prompting for personalisation but amplifies sycophancy and relationship-seeking, while simulated users recover aggregate rankings yet show far lower self-consistency and different topic and position biases than real humans.
WinDeskGround is a parametrically generated benchmark of 1,356 instruction-target pairs that reveals accuracy declines in state-of-the-art MLLMs under partial occlusion in multi-window GUI settings.
DiagnosticIQ benchmark shows frontier LLMs perform similarly on standard rule-to-action tasks but lose substantial accuracy under distractor expansion and condition inversion, pointing to calibration as the key deployment issue.
citing papers explorer
-
A Unified Geometric Framework for Weighted Contrastive Learning
Weighted InfoNCE objectives realize specific target geometries in embedding space, with SupCon producing size-dependent inter-class similarities under imbalance while Soft SupCon and certain continuous variants preserve regular simplex or unique optima.
-
Benign Fine-Tuning Breaks Safety Alignment in Audio LLMs
Benign fine-tuning on audio data breaks safety alignment in Audio LLMs by raising jailbreak success rates up to 87%, with the dominant risk axis depending on model architecture and embedding proximity to harmful content.
-
VLRS-Bench: A Vision-Language Reasoning Benchmark for Remote Sensing
VLRS-Bench is the first benchmark dedicated to complex vision-language reasoning in remote sensing, with 2000 QA pairs across 14 tasks in cognition, decision, and prediction dimensions.
-
DeepMath-103K: A Large-Scale, Challenging, Decontaminated, and Verifiable Mathematical Dataset for Advancing Reasoning
DeepMath-103K is a new 103K-problem mathematical dataset with high difficulty, rigorous decontamination, and verifiable answers to support RL training of language-model reasoning.
-
Mind2Web: Towards a Generalist Agent for the Web
Mind2Web is the first large-scale dataset of real-world web tasks for developing generalist language-guided agents that complete complex actions on diverse websites.
-
Pluralis v0.1: Towards a Multicultural, Multimodal, Multilingual Benchmark for AI Risk and Reliability
Pluralis v0.1 is a culture-first, multimodal, multilingual VLM safety benchmark spanning 6 APAC locales with 6,448 prompts and an agreement-gated LLM judge that disentangles safety from cultural appropriateness.
-
PromptGNN-sim: Deep Fusion and Alignment of GNN and LLMs for Text-Attributed Graph Learning
PromptGNN-sim uses GAT-based semantically aware neighborhood selection and structure-aware LLM prompts with bi-directional contrastive alignment to outperform prior GNN, LLM, and fusion methods on text-attributed graph datasets.
-
MKG-RAG-Bench: Benchmarking Retrieval in Multimodal Knowledge Graph-Augmented Generation
MKG-RAG-Bench is a cross-domain benchmark for retrieval in multimodal knowledge graph-augmented generation, constructed via LLM curation from two MKGs with aligned QA datasets.
-
From Sounds to Scenes: A Benchmark for Evaluating Context-Aware Auditory Scene Understanding in Large Audio Language Models
Introduces CASU benchmark with four tasks to evaluate context-aware auditory scene understanding in LALMs via semi-synthetic audio compositions of speech, events, and environments.
-
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.
-
From Efficiency to Leakage -- Privacy Backdoor in Federated Language Model Fine-Tuning
NeuroImprint attack assigns isolated memorization neurons to training samples in PEFT adapters, enabling closed-form reconstruction of 59-79% of samples across BERT, GPT-2, Qwen2, and Llama3.2 on multiple datasets.
-
Can neurons speak? Semantic narration of vision at single-cell resolution
NEURRATOR bridges neural spike trains to frozen CLIP patch embeddings via a learned encoder, then uses a multimodal LM and sparse autoencoder to produce validated natural-language narrations of viewed scenes from Neuropixel recordings.
-
LEDGER: A Long-Context Benchmark of Corporate Annual Reports for Grounded Financial Retrieval and Extraction
LEDGER provides a corpus of 4,999 annual reports with 31 labeled KPIs and three benchmarks for page-level retrieval, needle-in-haystack lookup, and full KPI extraction from long documents.
-
RECON: An LLM-Enhanced Backward Constraint Analysis Framework
RECON integrates LLM reasoning with backward static analysis to extract interpretable execution constraints from Android bytecode, reporting 5.8X speedup and 100% success versus symbolic execution plus 84% success on malware samples.
-
Re-Ranking Through an Attribution Lens for Citation Quality in Legal QA
Re-ranking retrieval candidates via a cross-encoder trained on continuous perturbation-based attribution scores improves citation faithfulness and gold-answer alignment in legal QA over semantic similarity.
-
GFFMERGE: Efficient Merging of Graph Neural Force Fields and Beyond
GFFMERGE formulates GNN force field merging as a convex embedding-alignment problem with an analytical solution, recovering near joint-training performance on MD17, MD22, LiPS20 and other benchmarks while delivering 5-27x speedups.
-
Bounded Behavioral Indistinguishability for Black-Box LLM Distillation
Introduces (ε,q,t,A)-behavioral indistinguishability and shows via Qwen/Llama experiments that LoRA distillation boosts semantic similarity but leaves detectable behavioral differences under adversarial evaluation.
-
Comprehensive Benchmarking of Long-Form Speech Generation in Diverse Scenarios
SwanBench-Speech is a new benchmark that decomposes long-form speech quality into seven disentangled metrics across 17 scenarios to evaluate generation models.
-
Spectral Retrieval: Multi-Scale Sinc Convolution over Token Embeddings for Localized Retrieval in LLM Multi-Agent Systems
Spectral Retrieval uses multi-scale sinc convolutions on token embeddings to interpolate between per-token MaxSim and mean-pooling, achieving large gains on synthetic and LIMIT-small benchmarks for localized retrieval.
-
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.
-
CAMERA: Adapting to Semantic Camouflage in Unsupervised Text-Attributed Graph Fraud Detection
CAMERA is an ego-decoupled mixture-of-experts model with context-informed gating and one-class objectives for unsupervised fraud detection in text-attributed graphs facing semantic camouflage.
-
PRISM-X: Experiments on Personalised Fine-Tuning with Human and Simulated Users
Preference fine-tuning outperforms prompting for personalisation but amplifies sycophancy and relationship-seeking, while simulated users recover aggregate rankings yet show far lower self-consistency and different topic and position biases than real humans.
-
WinDeskGround: A Benchmark for Robust GUI Grounding in Complex Multi-Window Desktop Environments
WinDeskGround is a parametrically generated benchmark of 1,356 instruction-target pairs that reveals accuracy declines in state-of-the-art MLLMs under partial occlusion in multi-window GUI settings.
-
DiagnosticIQ: A Benchmark for LLM-Based Industrial Maintenance Action Recommendation from Symbolic Rules
DiagnosticIQ benchmark shows frontier LLMs perform similarly on standard rule-to-action tasks but lose substantial accuracy under distractor expansion and condition inversion, pointing to calibration as the key deployment issue.
-
PersonaTeaming: Supporting Persona-Driven Red-Teaming for Generative AI
Persona-driven workflow and interface improve automated and human-AI red-teaming of generative AI by incorporating diverse perspectives into adversarial prompt creation.
-
Privacy Without Losing Place: A Paradigm for Private Retrieval in Spatial RAGs
PAS encodes locations via relative anchors and bins to deliver roughly 370-400m adversarial error in spatial RAG while retaining over half the baseline retrieval performance and keeping generation quality robust.
-
TabEmbed: Benchmarking and Learning Generalist Embeddings for Tabular Understanding
TabEmbed is the first generalist embedding model for tabular data that unifies classification and retrieval in one space via contrastive learning and outperforms text embedding models on the new TabBench benchmark.
-
Automated Large-scale CVRP Solver Design via LLM-assisted Flexible MCTS
LaF-MCTS uses LLM-assisted flexible MCTS with a three-tier hierarchy, semantic pruning, and branch regrowth to automatically compose decomposition-enhanced CVRP solvers that outperform state-of-the-art methods on CVRPLib benchmarks.
-
ContextualJailbreak: Evolutionary Red-Teaming via Simulated Conversational Priming
ContextualJailbreak uses evolutionary search over simulated primed dialogues with novel mutations to reach 90-100% attack success on open LLMs and transfers to some closed frontier models at 15-90% rates.
-
RepoDoc: A Knowledge Graph-Based Framework to Automatic Documentation Generation and Incremental Updates
RepoDoc uses a repository knowledge graph with module clustering and semantic impact propagation to generate more complete documentation 3x faster with 85% fewer tokens and handle incremental updates 73% faster than prior LLM-based tools.
-
Beyond Accuracy: Benchmarking Cross-Task Consistency in Unified Multimodal Models
XTC-Bench reveals that strong performance on generation or understanding tasks in unified multimodal models does not guarantee cross-task semantic consistency, which instead depends on how tightly coupled the learning objectives are across modalities.
-
Similar Users-Augmented Interest Network
SUIN improves CTR prediction by augmenting target user sequences with similar users' behaviors via embedding-based retrieval, user-specific position encoding, and user-aware target attention.
-
Prompt-Unknown Promotion Attacks against LLM-based Sequential Recommender Systems
PUDA enables effective promotion of unpopular target items in black-box LLM sequential recommenders by using evolutionary LLM refinement to infer hidden prompts, training a surrogate model, and combining adversarial text revision with surrogate-generated poisoning sequences.
-
R2Code: A Self-Reflective LLM Framework for Requirements-to-Code Traceability
R2Code improves requirement-to-code traceability with a bidirectional alignment network, self-reflective consistency verification, and dynamic context-adaptive retrieval, yielding 7.4% average F1 gain and up to 41.7% lower token use on five datasets.
-
Multilingual and Domain-Agnostic Tip-of-the-Tongue Query Generation for Simulated Evaluation
An LLM simulation framework generates multilingual tip-of-the-tongue queries, validated by rank correlation with real queries, producing the first large-scale ToT benchmarks for four languages.
-
Semantic Recall for Vector Search
Semantic Recall is a new evaluation metric for approximate nearest neighbor search that focuses only on semantically relevant results, with Tolerant Recall as a proxy when relevance labels are unavailable.
-
HumanScore: Benchmarking Human Motions in Generated Videos
HumanScore defines six metrics for kinematic plausibility, temporal stability, and biomechanical consistency to benchmark human motions in videos from thirteen state-of-the-art generation models, revealing gaps between visual appeal and physical fidelity.
-
LLM-Viterbi: Semantic-Aware Decoding for Convolutional Codes
An LLM-enhanced Viterbi decoder achieves roughly 1.5 dB extra coding gain in block error rate and over 50% better semantic similarity than conventional Viterbi for constraint-length-3 convolutional codes on AWGN channels.
-
DocQAC: Adaptive Trie-Guided Decoding for Effective In-Document Query Auto-Completion
Adaptive trie-guided decoding with document context and tunable penalties improves in-document query auto-completion, outperforming baselines and larger models like LLaMA-3 on seen queries.
-
Bayesian Active Learning with Gaussian Processes Guided by LLM Relevance Scoring for Dense Passage Retrieval
BAGEL is a Bayesian active learning framework that uses Gaussian Processes to propagate LLM relevance signals across embedding space and guide global exploration, outperforming standard LLM reranking under identical budgets on four retrieval benchmarks.
-
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.
-
Efficient Personalization of Generative User Interfaces
A dataset revealing high inter-designer disagreement on UI preferences motivates a sample-efficient method that personalizes generative interfaces by embedding new users in the space of prior designers, outperforming baselines in both modeling and user preference.
-
Skill-Conditioned Visual Geolocation for Vision-Language Models
GeoSkill lets vision-language models improve geolocation accuracy and reasoning by maintaining an evolving Skill-Graph that grows through autonomous analysis of successful and failed rollouts on web-scale image data.
-
Harnessing Hyperbolic Geometry for Harmful Prompt Detection and Sanitization
HyPE detects harmful prompts as outliers in hyperbolic space and HyPS sanitizes them using explainable attribution, outperforming prior defenses in accuracy and robustness across datasets and adversarial scenarios.
-
WorkRB: A Community-Driven Evaluation Framework for AI in the Work Domain
WorkRB is the first open community-driven benchmark for AI in the work domain, organizing 13 tasks from 7 groups with dynamic multilingual ontology loading and modular design for proprietary task integration.
-
SensorPersona: An LLM-Empowered System for Continual Persona Extraction from Longitudinal Mobile Sensor Streams
SensorPersona uses LLMs for hierarchical reasoning on longitudinal mobile sensor streams to continually extract stable personas, showing up to 31.4% higher recall and 85.7% win rate over baselines on a 20-user dataset.
-
How AI Coding Agents Modify Code: A Large-Scale Study of GitHub Pull Requests
AI coding agents produce pull requests with substantially more commits and slightly higher description-to-diff similarity than human developers, based on analysis of 29,095 merged PRs.
-
ProAgent: Harnessing On-Demand Sensory Contexts for Proactive LLM Agent Systems in the Wild
ProAgent uses on-demand tiered perception and context-aware LLM reasoning to deliver proactive assistance on AR glasses, achieving up to 27.7% higher prediction accuracy and 20.5% lower false detections than baselines.
-
ConsentDiff at Scale: Longitudinal Audits of Web Privacy Policy Changes and UI Frictions
ConsentDiff enables longitudinal tracking of privacy policy churn and consent UI patterns, finding ongoing changes, shifts away from high-friction banners, and higher policy-UI alignment when rejection options are visible.
-
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.