Multimodal Large Language Models (MLLMs) have achieved remarkable advances by integrating text, image, and audio understanding within a unified architecture. However, existing distributed training frameworks remain fundamentally data-blind: they parallelize computation without accounting for variations in input data characteristics. This data unawareness leads to severe computation skew across stages and microbatches, where heterogeneous multimodal inputs incur different processing costs. Consequently, GPU resources are unevenly utilized, synchronization delays accumulate, and overall training efficiency degrades. To address this limitation, we present DFLOP, a data-driven framework for multimodal LLM training pipeline optimization. DFLOP continuously profiles runtime behavior to capture data-induced computation variance and employs predictive scheduling to balance workloads across stages and microbatches. By coupling data characteristics with execution planning, DFLOP substantially improves GPU utilization and throughput. Extensive experiments on large-scale multimodal benchmarks show that DFLOP achieves up to 3.6x faster training compared to state-of-the-art distributed training frameworks.

翻译：多模态大语言模型（Multimodal Large Language Models, MLLMs）通过将文本、图像和音频理解集成于统一架构中取得了显著进展。然而，现有分布式训练框架本质上存在数据盲区：它们并行化计算时未考虑输入数据特性的差异。这种数据感知缺失导致跨阶段和微批次（microbatches）出现严重的计算倾斜——异构多模态输入会产生不同的处理开销。GPU资源因此利用不均，同步延迟持续累积，整体训练效率下降。为解决此问题，我们提出DFLOP——一个面向多模态大语言模型训练流水线优化的数据驱动框架。DFLOP持续分析运行时行为以捕捉数据引发的计算差异，并采用预测性调度平衡跨阶段和微批次的工作负载。通过将数据特性与执行规划耦合，DFLOP显著提升了GPU利用率和吞吐量。在大规模多模态基准上的大量实验表明，相较于现有最优分布式训练框架，DFLOP实现了最高3.6倍的训练加速。