Close Menu
    Main Menu
    • Home
    • News
    • Tech
    • Robotics
    • ML & Research
    • AI
    • Digital Transformation
    • AI Ethics & Regulation
    • Thought Leadership in AI

    Subscribe to Updates

    Get the latest creative news from FooBar about art, design and business.

    What's Hot

    Chinese language Menace Group ‘Jewelbug’ Quietly Infiltrated Russian IT Community for Months

    October 15, 2025

    Anthropic is freely giving its highly effective Claude Haiku 4.5 AI at no cost to tackle OpenAI

    October 15, 2025

    How To Navigate Ambiguity With Himanshu Palsule, The CEO of Cornerstone

    October 15, 2025
    Facebook X (Twitter) Instagram
    UK Tech InsiderUK Tech Insider
    Facebook X (Twitter) Instagram
    UK Tech InsiderUK Tech Insider
    Home»Machine Learning & Research»StreamBridge: Turning Your Offline Video Giant Language Mannequin right into a Proactive Streaming Assistant
    Machine Learning & Research

    StreamBridge: Turning Your Offline Video Giant Language Mannequin right into a Proactive Streaming Assistant

    Oliver ChambersBy Oliver ChambersMay 12, 2025No Comments1 Min Read
    Facebook Twitter Pinterest Telegram LinkedIn Tumblr Email Reddit
    StreamBridge: Turning Your Offline Video Giant Language Mannequin right into a Proactive Streaming Assistant
    Share
    Facebook Twitter LinkedIn Pinterest Email Copy Link


    We current StreamBridge, a easy but efficient framework that seamlessly transforms offline Video-LLMs into streaming-capable fashions. It addresses two basic challenges in adapting present fashions into on-line eventualities: (1) restricted functionality for multi-turn real-time understanding, and (2) lack of proactive response mechanisms. Particularly, StreamBridge incorporates (1) a reminiscence buffer mixed with a round-decayed compression technique, supporting long-context multi-turn interactions, and (2) a decoupled, light-weight activation mannequin that may be effortlessly built-in into present Video-LLMs, enabling steady proactive responses. To additional assist StreamBridge, we assemble Stream-IT, a large-scale dataset tailor-made for streaming video understanding, that includes interleaved video-text sequences and numerous instruction codecs. Intensive experiments present that StreamBridge considerably improves the streaming understanding capabilities of offline Video-LLMs throughout varied duties, outperforming even proprietary fashions corresponding to GPT-4o and Gemini 1.5 Professional. Concurrently, it achieves aggressive or superior efficiency on commonplace video understanding benchmarks.

    † Fudan College
    ‡‡ Work performed throughout Apple internship

    Determine 1: Illustration of streaming eventualities. Prime: Multi-turn interactions. Backside: Proactive responses.
    Share. Facebook Twitter Pinterest LinkedIn Tumblr Email
    Oliver Chambers
    • Website

    Related Posts

    FS-DFM: Quick and Correct Lengthy Textual content Era with Few-Step Diffusion Language Fashions

    October 15, 2025

    Construct a tool administration agent with Amazon Bedrock AgentCore

    October 15, 2025

    Information Analytics Automation Scripts with SQL Saved Procedures

    October 15, 2025
    Top Posts

    Evaluating the Finest AI Video Mills for Social Media

    April 18, 2025

    Utilizing AI To Repair The Innovation Drawback: The Three Step Resolution

    April 18, 2025

    Midjourney V7: Quicker, smarter, extra reasonable

    April 18, 2025

    Meta resumes AI coaching utilizing EU person knowledge

    April 18, 2025
    Don't Miss

    Chinese language Menace Group ‘Jewelbug’ Quietly Infiltrated Russian IT Community for Months

    By Declan MurphyOctober 15, 2025

    A menace actor with ties to China has been attributed to a five-month-long intrusion concentrating…

    Anthropic is freely giving its highly effective Claude Haiku 4.5 AI at no cost to tackle OpenAI

    October 15, 2025

    How To Navigate Ambiguity With Himanshu Palsule, The CEO of Cornerstone

    October 15, 2025

    FS-DFM: Quick and Correct Lengthy Textual content Era with Few-Step Diffusion Language Fashions

    October 15, 2025
    Stay In Touch
    • Facebook
    • Twitter
    • Pinterest
    • Instagram
    • YouTube
    • Vimeo

    Subscribe to Updates

    Get the latest creative news from SmartMag about art & design.

    UK Tech Insider
    Facebook X (Twitter) Instagram
    • About Us
    • Contact Us
    • Privacy Policy
    • Terms Of Service
    • Our Authors
    © 2025 UK Tech Insider. All rights reserved by UK Tech Insider.

    Type above and press Enter to search. Press Esc to cancel.