Research Papers 论文研究 2d ago Updated 1d ago 更新于 1天前 44

Accelerating Visual On-Policy Distillation with Batched Speculative Jacobi Rollouts 使用批处理投机雅可比轨迹加速视觉在线策略蒸馏

Visual on-policy distillation (OPD) trains compact visual autoregressive models using student-generated trajectories, but token-by-token autoregressive decoding creates substantial computational overhead per training step Speculative Jacobi Decoding (SJD) enables parallel multi-token processing without an auxiliary draft model, but was previously limited to single-sequence inference HB-SJD introduces a batched SJD rollout backend that allows each image to advance independently based on its own d 视觉在线策略蒸馏(OPD)通过学生模型在线轨迹提升紧凑视觉自回归模型训练,但逐token解码计算成本高昂 推测Jacobi解码(SJD)可并行处理多token且无需辅助草稿模型,但原方法仅支持单序列推理 HB-SJD将SJD扩展为批处理后端,允许不同图像按自身进度独立推进,同时保持批处理验证 实验表明HB-SJD在LlamaGen上显著降低展开和端到端训练时间,同时保持蒸馏学生模型的生成质量

55
Hot 热度
72
Quality 质量
65
Impact 影响力

Analysis 深度分析

TL;DR

  • Visual on-policy distillation (OPD) trains compact visual autoregressive models using student-generated trajectories, but token-by-token autoregressive decoding creates substantial computational overhead per training step
  • Speculative Jacobi Decoding (SJD) enables parallel multi-token processing without an auxiliary draft model, but was previously limited to single-sequence inference
  • HB-SJD introduces a batched SJD rollout backend that allows each image to advance independently based on its own decoding progress while still leveraging batched model forwards for verification
  • The method dynamically switches between Full and Compact execution modes as images complete, reducing costs in later rollout rounds
  • Experiments with LlamaGen demonstrate substantial reductions in both rollout and end-to-end training time while preserving generation quality, with no changes required to the teacher, distillation objective, or optimization procedure

Why It Matters

This work directly addresses one of the most significant bottlenecks in training visual autoregressive models: the computational cost of on-policy rollouts. By enabling parallel token generation within a batching framework, HB-SJD makes visual OPD significantly more practical for real-world training pipelines, potentially accelerating the development cycle for compact image generation models.

Technical Details

  • Problem addressed: Visual OPD generates training trajectories token-by-token via autoregressive decoding, which is computationally expensive and limits training throughput
  • Core innovation: HB-SJD extends Speculative Jacobi Decoding to a batched setting, allowing independent per-image advancement while maintaining batched verification through model forwards
  • Execution modes: The system switches between Full and Compact execution as images finish decoding, optimizing resource allocation across rollout rounds
  • Compatibility: HB-SJD operates as a drop-in replacement for the student rollout backend only, leaving the teacher model, distillation objective, and optimization procedure completely unchanged
  • Validation: Evaluated on LlamaGen, showing significant speedups in rollout and end-to-end training time with preserved generation quality

Industry Insight

  • The approach demonstrates that parallel decoding techniques originally designed for text can be effectively adapted to visual generation, opening pathways for similar acceleration strategies across other modalities
  • The modular design—replacing only the rollout backend—suggests that existing visual OPD pipelines can be upgraded with minimal engineering effort, lowering the barrier to adoption
  • As compact visual autoregressive models become more competitive with diffusion-based approaches, efficient training methods like HB-SJD will be critical for making on-policy distillation viable at scale

TL;DR

  • 视觉在线策略蒸馏(OPD)通过学生模型在线轨迹提升紧凑视觉自回归模型训练,但逐token解码计算成本高昂
  • 推测Jacobi解码(SJD)可并行处理多token且无需辅助草稿模型,但原方法仅支持单序列推理
  • HB-SJD将SJD扩展为批处理后端,允许不同图像按自身进度独立推进,同时保持批处理验证
  • 实验表明HB-SJD在LlamaGen上显著降低展开和端到端训练时间,同时保持蒸馏学生模型的生成质量

为什么值得看

本文针对视觉自回归模型训练中的计算瓶颈提出了高效的批处理解码方案,对视觉生成领域研究者具有重要参考价值。HB-SJD在不改变蒸馏框架的前提下实现了训练加速,为紧凑视觉模型的工业化部署提供了可行路径。

技术解析

  • 视觉在线策略蒸馏(OPD)通过从当前学生模型生成的轨迹中学习,提升紧凑视觉自回归模型的训练效果,但在线展开仍采用逐token自回归解码,导致每个训练步骤成本高昂
  • 推测Jacobi解码(SJD)可并行处理多个token且无需辅助草稿模型,但原始方法仅针对单序列推理设计,无法直接应用于批量训练场景
  • HB-SJD作为批处理SJD展开后端,允许每张图像根据自身的解码进度独立推进,同时不同序列位置的图像仍能在批处理模型前向传播中进行验证
  • 当图像完成生成时,HB-SJD在Full和Compact执行模式之间切换,以降低后续展开轮次的计算开销
  • HB-SJD仅替换学生模型的展开后端,保持教师模型、蒸馏目标和优化流程不变,兼容现有OPD框架
  • 在LlamaGen上的实验表明,HB-SJD显著减少了展开和端到端训练时间,同时保持了蒸馏学生模型的生成质量

行业启示

  • 视觉自回归模型的训练效率优化是当前的研究热点,批处理并行解码技术有望成为标准实践
  • 在线策略蒸馏与高效解码技术的结合,为紧凑视觉生成模型的训练提供了新的思路
  • 无需辅助草稿模型的并行解码方法降低了部署复杂度,对实际应用场景更具吸引力

Disclaimer: The above content is generated by AI and is for reference only. 免责声明:以上内容由 AI 生成,仅供参考。

Research 科学研究 Training 训练 Inference 推理 Multimodal 多模态