# DIP: Efficient Large Multimodal Model Training with Dynamic Interleaved Pipeline

arXiv: https://arxiv.org/abs/2504.14145 · HTML: https://mlsanity.com/paper/2504.14145/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Placed in MLLM training and visual instruction tuning, one of 16 directions in Multimodal LLMs / vision-language models.
- Builds on 29 of its 54 references that are in the map.
- First posted on arXiv 19 Apr 2025.

## Placed in

- Multimodal LLMs / vision-language models / [MLLM training and visual instruction tuning](https://mlsanity.com/mllm/mllm-training-and-visual-instruction-tuning/)

## Builds on (most cited first, 29 of 29)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [A Simple Framework for Contrastive Learning of Visual Representations](https://mlsanity.com/paper/2002.05709/) (2020)
- [Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolution](https://mlsanity.com/paper/2409.12191/) (2024)
- [LAION-5B: An open large-scale dataset for training next generation image-text models](https://mlsanity.com/paper/2210.08402/) (2022)
- [Learn to Explain: Multimodal Reasoning via Thought Chains for Science Question Answering](https://mlsanity.com/paper/2209.09513/) (2022)
- [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://mlsanity.com/paper/1909.08053/) (2019)
- [Scaling Vision Transformers to 22 Billion Parameters](https://mlsanity.com/paper/2302.05442/) (2023)
- [PyTorch FSDP: Experiences on Scaling Fully Sharded Data Parallel](https://mlsanity.com/paper/2304.11277/) (2023)
- [Movie Gen: A Cast of Media Foundation Models](https://mlsanity.com/paper/2410.13720/) (2024)
- [ShareGPT4Video: Improving Video Understanding and Generation with Better Captions](https://mlsanity.com/paper/2406.04325/) (2024)
- [OBELICS: An Open Web-Scale Filtered Dataset of Interleaved Image-Text Documents](https://mlsanity.com/paper/2306.16527/) (2023)
- [Hunyuan-DiT: A Powerful Multi-Resolution Diffusion Transformer with Fine-Grained Chinese Understanding](https://mlsanity.com/paper/2405.08748/) (2024)
- [Multimodal C4: An Open, Billion-scale Corpus of Images Interleaved with Text](https://mlsanity.com/paper/2304.06939/) (2023)
- [Step-Audio: Unified Understanding and Generation in Intelligent Speech Interaction](https://mlsanity.com/paper/2502.11946/) (2025)
- [Chimera: Efficiently Training Large-Scale Neural Networks with Bidirectional Pipelines](https://mlsanity.com/paper/2107.06925/) (2021)
- [M-Longdoc: A Benchmark For Multimodal Super-Long Document Understanding And A Retrieval-Aware Tuning Framework](https://mlsanity.com/paper/2411.06176/) (2024)
- [Hanayo: Harnessing Wave-like Pipeline Parallelism for Enhanced Large Model Training Efficiency](https://mlsanity.com/paper/2308.15762/) (2023)
- [ReCycle: Resilient Training of Large DNNs using Pipeline Adaptation](https://mlsanity.com/paper/2405.14009/) (2024)
- [Optimus: Accelerating Large-Scale Multi-Modal LLM Training by Bubble Exploitation](https://mlsanity.com/paper/2408.03505/) (2024)
- [MMTrail: A Multimodal Trailer Video Dataset with Language and Music Descriptions](https://mlsanity.com/paper/2407.20962/) (2024)
- [DynaPipe: Optimizing Multi-task Training through Dynamic Pipelines](https://mlsanity.com/paper/2311.10418/) (2023)
- [GraphPipe: Improving Performance and Scalability of DNN Training with Graph Pipeline Parallelism](https://mlsanity.com/paper/2406.17145/) (2024)
- [Spindle: Efficient Distributed Training of Multi-Task Large Models via Wavefront Scheduling](https://mlsanity.com/paper/2409.03365/) (2024)
- [DiffusionPipe: Training Large Diffusion Models with Efficient Pipelines](https://mlsanity.com/paper/2405.01248/) (2024)
- [SPPO:Efficient Long-sequence LLM Training via Adaptive Sequence Pipeline Parallel Offloading](https://mlsanity.com/paper/2503.10377/) (2025)
- [OrchMLLM: Orchestrate Multimodal Data with Batch Post-Balancing to Accelerate Multimodal Large Language Model Training](https://mlsanity.com/paper/2503.23830/) (2025)

## Built on by (most cited first, 3 of 3)

- [LongCat-Flash-Omni Technical Report](https://mlsanity.com/paper/2511.00279/) (2025)
- [OctoPipe: Reducing Pipeline Bubbles for Heterogeneous Models via Co-Optimizing Partitioning, Placement, and Scheduling](https://mlsanity.com/paper/2509.23722/) (2025)
- [FlowTrain: Flow-Based Decoupled Training for Industrial-Grade Vision-Language Models](https://mlsanity.com/paper/2606.23087/) (2026)

## Abstract (arXiv)

Large multimodal models (LMMs) have demonstrated excellent capabilities in both understanding and generation tasks with various modalities. While these models can accept flexible combinations of input data, their training efficiency suffers from two major issues: pipeline stage imbalance caused by heterogeneous model architectures, and training data dynamicity stemming from the diversity of multimodal data. In this paper, we present DIP, a dynamic and modality-aware pipeline scheduling framework designed for LMM training. DIP tackles the challenge of dynamic imbalance via two key techniques: (1) separating computations of different modalities into dedicated pipeline segments to balance workloads within a continuous set of stages; (2) dynamically splitting input data into finer-grained, modality-specific sub-microbatches to balance workloads across these segments. By asynchronously generating pipeline schedules on idle CPU resources during training, DIP dynamically tailors stage executions to each input batch without stalling the training process. We validate DIP on a diverse set of five LMMs, ranging from 12B to 94B parameters and including vision-language and diffusion models. Experimental results show that our system achieves up to 97.3% higher throughput compared to state-of-the-art systems, demonstrating strong adaptability to fluctuating multimodal training workloads.