Beyond Single-Sample: Reliable Multi-Sample Distillation for Video Understanding

• URL: http://arxiv.org/abs/2603.11423v1
• Date: Thu, 12 Mar 2026 01:30:37 GMT
• Title: Beyond Single-Sample: Reliable Multi-Sample Distillation for Video Understanding
• Abstract summary: R-MSD (Reliable Multi-Sample Distillation) is a framework that explicitly models teacher sampling variance to enhance distillation stability.<n>By integrating quality-aware signal matching with an adversarial distillation objective, our approach effectively filters teacher noise while maximizing knowledge transfer.
• Score: 6.759788796474815
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Traditional black-box distillation for Large Vision-Language Models (LVLMs) typically relies on a single teacher response per input, which often yields high-variance responses and format inconsistencies in multimodal or temporal scenarios. To mitigate this unreliable supervision, we propose R-MSD (Reliable Multi-Sample Distillation), a framework that explicitly models teacher sampling variance to enhance distillation stability. Rather than relying on a single teacher response, our approach leverages a task-adaptive teacher pool to provide robust supervision tailored to both closed-ended and open-ended reasoning. By integrating quality-aware signal matching with an adversarial distillation objective, our approach effectively filters teacher noise while maximizing knowledge transfer. Extensive evaluations across comprehensive video understanding benchmarks demonstrate that R-MSD consistently outperforms single sample distillation methods. We additionally include an original SFT+RL 4B baseline under the same training budget, which shows only marginal gains, while our method achieves significant improvements. With a 4B student model, our approach delivers gains on VideoMME (+1.5%), Video-MMMU (+3.2%), and MathVerse (+3.6%).

Related papers

• Divergence Meets Consensus: A Multi-Source Negative Sampling Framework for Sequential Recommendation [12.425187572716574]
  "Teacher-Peer-Self" framework is inspired by Vygotsky's Zone of Proximal Development (ZPD) theory.<n>It incorporates multi-source scoring, divergence re-ranking, and consensus distillation.<n>It consistently outperforms state-of-the-art negative sampling methods.
  arXiv  Detail & Related papers  (2026-05-19T10:42:23Z)
• Uni-OPD: Unifying On-Policy Distillation with a Dual-Perspective Recipe [53.40076304466524]
  We propose Uni-OPD, a unified OPD framework that generalizes across Large Language Models (LLMs) and Multimodal Large Language Models (MLLMs)<n>Specifically, from the student's perspective, we adopt two data balancing strategies to promote exploration of informative student-generated states during training.<n>We develop an outcome-guided margin calibration mechanism to restore order consistency between correct and incorrect trajectories.
  arXiv  Detail & Related papers  (2026-05-05T12:15:21Z)
• Distribution-Aligned Sequence Distillation for Superior Long-CoT Reasoning [48.041170200238206]
  We introduce DASD-4B-Thinking, a lightweight yet highly capable, fully open-source reasoning model.<n>It achieves SOTA performance among open-source models of comparable scale across challenging benchmarks in mathematics, scientific reasoning, and code generation.
  arXiv  Detail & Related papers  (2026-01-14T02:43:17Z)
• AMoE: Agglomerative Mixture-of-Experts Vision Foundation Model [23.785186661138734]
  We study multi-teacher distillation for vision foundation models and identify key factors that enable training at lower computational cost.<n>We introduce Agglomerative Mixture-of-Experts Vision Foundation Models (AMoE), which distill knowledge from SigLIP2 and DINOv3 simultaneously into a Mixture-of-Experts student.<n>We show that our Asymmetric Relation-Knowledge Distillation loss preserves the geometric properties of each teacher while enabling effective knowledge transfer.
  arXiv  Detail & Related papers  (2025-12-23T08:37:11Z)
• MMT-ARD: Multimodal Multi-Teacher Adversarial Distillation for Robust Vision-Language Models [123.90007730845876]
  We propose MMT-ARD: a Multimodal Multi-Teacher Adversarial Distillation framework.<n>Our key innovation is a dual-teacher knowledge fusion architecture that collaboratively optimize clean feature preservation and robust feature enhancement.<n>Experiments on ImageNet and zero-shot benchmarks demonstrate that MMT-ARD improves robust accuracy by +4.32% and zero-shot accuracy by +3.5%.
  arXiv  Detail & Related papers  (2025-11-21T17:46:44Z)
• AdaSwitch: Adaptive Switching Generation for Knowledge Distillation [58.647880811071495]
  Small language models (SLMs) are crucial for applications with strict latency and computational constraints.<n>We propose AdaSwitch, a novel approach that combines on-policy and off-policy generation at the token level.<n>AdaSwitch consistently improves accuracy, offering a practical and effective method for distilling SLMs with acceptable additional overhead.
  arXiv  Detail & Related papers  (2025-10-09T06:38:37Z)
• Adversarial Distribution Matching for Diffusion Distillation Towards Efficient Image and Video Synthesis [65.77083310980896]
  We propose Adrial Distribution Matching (ADM) to align latent predictions between real and fake score estimators for score distillation.<n>Our proposed method achieves superior one-step performance on SDXL compared to DMD2 while consuming less GPU time.<n>Additional experiments that apply multi-step ADM distillation on SD3-Medium, SD3.5-Large, and CogVideoX set a new benchmark towards efficient image and video synthesis.
  arXiv  Detail & Related papers  (2025-07-24T16:45:05Z)
• TwT: Thinking without Tokens by Habitual Reasoning Distillation with Multi-Teachers' Guidance [32.6122298575412]
  We propose TwT, a method that reduces inference-time costs through habitual reasoning distillation with multi-teachers' guidance.<n>Our approach internalizes explicit reasoning into the model's habitual behavior through a Teacher-Guided compression strategy.<n> Experimental results demonstrate that TwT effectively reduces inference costs while preserving superior performance.
  arXiv  Detail & Related papers  (2025-03-31T15:16:31Z)
• Distill Not Only Data but Also Rewards: Can Smaller Language Models Surpass Larger Ones? [58.80794196076336]
  Distilling large language models (LLMs) typically involves transferring the teacher model's responses through supervised fine-tuning (SFT)<n>We propose a novel distillation pipeline that transfers both responses and rewards.<n>Our method generates pseudo-rewards through a self-supervised mechanism that leverages the inherent structure of both teacher and student responses.
  arXiv  Detail & Related papers  (2025-02-26T20:50:11Z)
• VideoAdviser: Video Knowledge Distillation for Multimodal Transfer Learning [6.379202839994046]
  Multimodal transfer learning aims to transform pretrained representations of diverse modalities into a common domain space for effective multimodal fusion. We propose VideoAdviser, a video knowledge distillation method to transfer multimodal knowledge of video-enhanced prompts from a multimodal fundamental model to a specific modal fundamental model. We evaluate our method in two challenging multimodal tasks: video-level sentiment analysis and audio-visual retrieval.
  arXiv  Detail & Related papers  (2023-09-27T08:44:04Z)
• Hybrid Distillation: Connecting Masked Autoencoders with Contrastive Learners [102.20090188997301]
  We explore how to obtain a model that combines Contrastive Learning (CL) and Masked Image Modeling (MIM) strengths. In order to better obtain both discrimination and diversity, we propose a simple but effective Hybrid Distillation strategy. Experiment results prove that Hybrid Distill can achieve superior performance on different benchmarks.
  arXiv  Detail & Related papers  (2023-06-28T02:19:35Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.