·
DataBubble
  • Home
  • Models
  • News
  • Compare
  • Boards
  • Pricing
  • About
  • Newsletter
  • Methodology
  • Contact
Latest
MMGist: A Comprehensive Multimodal Benchmark for 20272h◆Visualizing "We the People": Bridging the Perception Gap through Pluralistic Data Storytelling2h◆Small edits, large models: How Wikipedia advocacy shapes LLM values2h◆Noise-Aware Boundary-Enhanced Generative Learning for Ultrasound Speckle Reduction2h◆Wan-Streamer v0.1: End-to-end Real-time Interactive Foundation Models2h◆MiniOpt: Reasoning to Model and Solve General Optimization Problems with Limited Resources2h◆HierBias: Context-Conditioned Hierarchical Media Bias Detection with Multi-Task Type Classification2h◆Where Larger Models Excel: The Primacy of Constraint-Guided Reasoning2h◆Dynamic-dLLM: Dynamic Cache-Budget and Adaptive Parallel Decoding for Training-Free Acceleration of Diffusion LLM2h◆Phonetic and semantic analyses of spoken corpora of Beijing and Taiwan Mandarin indicate that the neutral tone is a lexical tone2h◆ProfileFoundry: A Synthetic Person-Object Substrate for Privacy, Memory, and Tool-Use Evaluation in LLM Agent2h◆AnySimLite: A Lightweight Few-Shot Similarity Encoder for On-Device Speech-Adjacent Classification2h◆Extracting Problem and Method Sentence from Scientific Papers: A Context-enhanced Transformer Using Formulaic Expression Desensitization2h◆Utilizing Cognitive Signals Generated during Human Reading to Enhance Keyphrase Extraction from Microblogs2h◆Comparing BERT Sentence-Pair Classification and Few-Shot LLM Prompting for Detecting Threat and Solution Framing in German Climate News2h◆Nemotron-TwoTower: Diffusion Language Modeling with Pretrained Autoregressive Context2h◆Assessing Post-Reform Changes in Risk Disclosure Quality with a Multidimensional Text Analysis Approach2h◆Erase-then-Delta Attention: Decoupling Erase and Write Addresses in Delta-Rule Linear Attention2h◆Zero-shot Tweet-Level Stance Detection Enhanced by External Knowledge and Reflective Chain-of-Thought Reasoning2h◆Closing the Quality Gap in Low-Resource Text-to-Speech: LoRA Fine-Tuning of VoxCPM2 for Khmer and Korean2h◆MMGist: A Comprehensive Multimodal Benchmark for 20272h◆Visualizing "We the People": Bridging the Perception Gap through Pluralistic Data Storytelling2h◆Small edits, large models: How Wikipedia advocacy shapes LLM values2h◆Noise-Aware Boundary-Enhanced Generative Learning for Ultrasound Speckle Reduction2h◆Wan-Streamer v0.1: End-to-end Real-time Interactive Foundation Models2h◆MiniOpt: Reasoning to Model and Solve General Optimization Problems with Limited Resources2h◆HierBias: Context-Conditioned Hierarchical Media Bias Detection with Multi-Task Type Classification2h◆Where Larger Models Excel: The Primacy of Constraint-Guided Reasoning2h◆Dynamic-dLLM: Dynamic Cache-Budget and Adaptive Parallel Decoding for Training-Free Acceleration of Diffusion LLM2h◆Phonetic and semantic analyses of spoken corpora of Beijing and Taiwan Mandarin indicate that the neutral tone is a lexical tone2h◆ProfileFoundry: A Synthetic Person-Object Substrate for Privacy, Memory, and Tool-Use Evaluation in LLM Agent2h◆AnySimLite: A Lightweight Few-Shot Similarity Encoder for On-Device Speech-Adjacent Classification2h◆Extracting Problem and Method Sentence from Scientific Papers: A Context-enhanced Transformer Using Formulaic Expression Desensitization2h◆Utilizing Cognitive Signals Generated during Human Reading to Enhance Keyphrase Extraction from Microblogs2h◆Comparing BERT Sentence-Pair Classification and Few-Shot LLM Prompting for Detecting Threat and Solution Framing in German Climate News2h◆Nemotron-TwoTower: Diffusion Language Modeling with Pretrained Autoregressive Context2h◆Assessing Post-Reform Changes in Risk Disclosure Quality with a Multidimensional Text Analysis Approach2h◆Erase-then-Delta Attention: Decoupling Erase and Write Addresses in Delta-Rule Linear Attention2h◆Zero-shot Tweet-Level Stance Detection Enhanced by External Knowledge and Reflective Chain-of-Thought Reasoning2h◆Closing the Quality Gap in Low-Resource Text-to-Speech: LoRA Fine-Tuning of VoxCPM2 for Khmer and Korean2h◆
News/TOPS: First-Principles Visual Token Pruning via Constructing Token Optimal Preservation Sets for Efficient MLLM Inference
arxiv
PublishedJune 26, 2026 at 4:00 AM
▲bullish

TOPS: First-Principles Visual Token Pruning via Constructing Token Optimal Preservation Sets for Efficient MLLM Inference

Source
arxiv.orgfull article ↗
Read on arxiv→
Publisher summary· verbatim

arXiv:2606.27161v1 Announce Type: new Abstract: Multimodal large language models (MLLMs) have achieved strong multimodal reasoning capabilities, but their efficiency is limited by the large number of visual tokens, which introduces substantial computational overhead. Visual token pruning offers a na

Stay posted· Newsletter

A 5-min weekly brief — top movers, price watch, story of the week.

// no spam · unsubscribe one-click · free forever

Discussion
Mentioned models
01
  • 01
    LLaVA-NeXT
Source
↗
arxiv
Read original ↗All from arxiv →
Tags
03
#multimodal#pruning#efficiency

No replies yet. Be first.

Mentioned models
01
  • 01
    LLaVA-NeXT
Source
↗
arxiv
Read original ↗All from arxiv →
Tags
03
#multimodal#pruning#efficiency

Related coverage

More from ARXIV
arxivMMGist: A Comprehensive Multimodal Benchmark for 20272harxivVisualizing "We the People": Bridging the Perception Gap through Pluralistic Data Storytelling2harxivSmall edits, large models: How Wikipedia advocacy shapes LLM values2harxivNoise-Aware Boundary-Enhanced Generative Learning for Ultrasound Speckle Reduction2h
The Bubble Brief
WEEKLY

Read multimodal insights every Tuesday — top movers, new releases, story of the week.

// no spam · unsubscribe one-click · free forever

Originally published on arxiv ↗
HomeModelsNews