Close Menu
techwavedispatch.com

    Subscribe to Updates

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

    What's Hot

    Setting Up An Apple HomeView Is Ridiculously Easy Thanks To Your iPhone

    November 21, 2026

    When Do Rankings Beat Signs? Gradient Direction Recovery under Even Nuisance

    November 21, 2026

    What’s The Difference Between Apple’s And Amazon’s Smart Displays?

    November 21, 2026
    Facebook X (Twitter) Instagram
    Trending
    • Setting Up An Apple HomeView Is Ridiculously Easy Thanks To Your iPhone
    • When Do Rankings Beat Signs? Gradient Direction Recovery under Even Nuisance
    • What’s The Difference Between Apple’s And Amazon’s Smart Displays?
    • [2607.12304] What Does a Temporal Benchmark Score Measure? Decomposing Channel Use in Video VLM Evaluation
    • The FBI and USDA were reportedly hacked
    • Where You Can Find Which GPU You Have In Your Windows PC
    • The Delayed Societal Cost of Unfairness
    • This Retro Classic Is The Most Re-Released Game Of All Time — Here’s Why
    Facebook X (Twitter) Instagram
    techwavedispatch.com
    Subscribe
    Saturday, November 21
    • Home
    • AI Tools & Software
    • Machine Learning
    • Generative AI
    • Robotics & Automation
    • Future Tech & Hardware
    • AI in Business
    • Ethics & Regulations
    • More
      • Gadgets & Smart Devices
      • Cybersecurity & AI
      • Tech Breakthroughs
    techwavedispatch.com
    Home»Machine Learning»On-Policy Distillation with Transferable Relative Proxy Update for Large Language Models
    Machine Learning

    On-Policy Distillation with Transferable Relative Proxy Update for Large Language Models

    adminBy adminNovember 18, 2026No Comments2 Mins Read
    Facebook Twitter Pinterest LinkedIn Tumblr Email
    On-Policy Distillation with Transferable Relative Proxy Update for Large Language Models
    Share
    Facebook Twitter LinkedIn Pinterest Email

    [Submitted on 13 Jul 2026 (v1), last revised 16 Nov 2026 (this version, v3)]

    View a PDF of the paper titled Proxy OPD: On-Policy Distillation with Transferable Relative Proxy Update for Large Language Models, by Daocheng Fu and 9 other authors

    View PDF
    HTML (experimental)

    Abstract:Reward-based reinforcement learning (RL) improves large language models but incurs substantial costs when repeated across targets. On-policy distillation (OPD) reuses experts by matching their distributions on target-generated trajectories, yet differing base policies can confound reward-induced updates with pre-existing policy differences, potentially distorting transferred update directions. We propose Proxy OPD (P-OPD), a post-training method that transfers reward-induced relative policy updates using the target’s frozen base policy as a reference. A proxy undergoes reward-guided exploration, yielding token-level log-probability changes between its initial and optimized policies. These signals guide learning on target-generated trajectories, with calibration controlling signal absorption, enabling reuse of reward-induced improvements across targets. Experiments on Qwen3-family targets in mathematical reasoning, code generation, and instruction following show that P-OPD improves targets and outperforms conventional OPD in most settings. For 1.7B-to-8B transfer, average Math Mean@16 and Code Mean@8 scores exceed OPD by 8.3 and 3.4 points, respectively. Mechanism analyses show more faithful reward-induced supervision and controllable adaptation. Cross-scale and cross-generation experiments confirm signal transferability and reusability, while multi-domain distillation reveals synergistic gains and cross-domain interference. In the measured 4B-to-8B setting, target adaptation after proxy exploration takes approximately 15\% of direct target RL time; repeated signal reuse amortizes the one-time exploration cost across compatible models. These findings support relative policy updates as effective, transferable, and composable supervision for an “explore once, transfer multiple times” post-training paradigm.

    Submission history

    From: Daocheng Fu [view email]
    [v1]
    Mon, 13 Jul 2026 12:56:21 UTC (3,679 KB)
    [v2]
    Mon, 10 Aug 2026 09:36:14 UTC (3,871 KB)
    [v3]
    Mon, 16 Nov 2026 12:58:54 UTC (4,552 KB)

    Distillation Language Large Models OnPolicy proxy Relative Transferable Update
    Share. Facebook Twitter Pinterest LinkedIn Tumblr Email
    admin
    • Website

    Related Posts

    When Do Rankings Beat Signs? Gradient Direction Recovery under Even Nuisance

    November 21, 2026

    [2607.12304] What Does a Temporal Benchmark Score Measure? Decomposing Channel Use in Video VLM Evaluation

    November 21, 2026

    The Delayed Societal Cost of Unfairness

    November 21, 2026
    Leave A Reply Cancel Reply

    Top Posts

    NVIDIA Is Investigating Reports That CEO Jensen Huang Was The Target Of Chinese Spy Efforts

    November 8, 202619 Views

    ADIndex: Layer-Adaptive Sparse Attention Indexing via Group-Level Distillation

    November 11, 20265 Views

    Anthropic releases new Mythos and Fable models amid RSI concerns

    November 9, 20265 Views

    If AGI is so close, why is everyone leaving? The neolabs and their bets

    November 6, 20265 Views
    Don't Miss
    Gadgets & Smart Devices

    Setting Up An Apple HomeView Is Ridiculously Easy Thanks To Your iPhone

    By adminNovember 21, 20260

    You can also easily add other users to your HomeView display. …

    When Do Rankings Beat Signs? Gradient Direction Recovery under Even Nuisance

    November 21, 2026

    What’s The Difference Between Apple’s And Amazon’s Smart Displays?

    November 21, 2026

    [2607.12304] What Does a Temporal Benchmark Score Measure? Decomposing Channel Use in Video VLM Evaluation

    November 21, 2026
    About Us

    Welcome to TechWave Dispatch

    At TechWave Dispatch, we bring you the latest news, insights, and developments from the ever-evolving world of technology. Our mission is to keep readers informed about the innovations shaping our digital future.

    From artificial intelligence and machine learning to cybersecurity, robotics, smart devices, and groundbreaking technologies, we cover the topics that matter in today's fast-moving tech landscape.

    Our Picks

    Setting Up An Apple HomeView Is Ridiculously Easy Thanks To Your iPhone

    November 21, 2026

    When Do Rankings Beat Signs? Gradient Direction Recovery under Even Nuisance

    November 21, 2026

    What’s The Difference Between Apple’s And Amazon’s Smart Displays?

    November 21, 2026
    Most Popular

    After blocking Muse, Amazon now lets users text Alexa to help them shop

    November 6, 20260 Views

    A controlled world for robot learning and evaluation

    November 6, 20260 Views

    Egyptian and Nigerian English join the Open ASR Leaderboard

    November 6, 20260 Views
    Facebook X (Twitter) Instagram Pinterest
    • Terms and Conditions
    • Privacy Policy
    • Contact Us
    • About Us
    © 2026 TechWave Dispatch. All Rights Reserved.

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