MindVL: Towards Efficient and Effective Training of Multimodal Large Language Models on Ascend NPUs

• URL: http://arxiv.org/abs/2509.11662v3
• Date: Tue, 30 Sep 2025 02:27:18 GMT
• Title: MindVL: Towards Efficient and Effective Training of Multimodal Large Language Models on Ascend NPUs
• Authors: Feilong Chen, Yijiang Liu, Yi Huang, Hao Wang, Miren Tian, Ya-Qi Yu, Minghui Liao, Jihao Wu,
• Abstract summary: MindVL is a multimodal large language model trained end-to-end on Ascend NPUs.<n>We introduce MindSpeed-MLLM, a highly efficient training framework that supports stable and high-performance training.<n>We find that averaging weights from checkpoints trained with different sequence lengths is particularly effective.
• Score: 20.842336447426682
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: We propose MindVL, a multimodal large language model (MLLMs) trained on Ascend NPUs. The training of state-of-the-art MLLMs is often confined to a limited set of hardware platforms and relies heavily on massive, undisclosed data recipes, which hinders reproducibility and open research. To change the common perception that Ascend hardware is unsuitable for efficient full-stage MLLM training, we introduce MindSpeed-MLLM, a highly efficient training framework that supports stable and high-performance training of large-scale Dense and Mixture-of-Experts (MoE) models on Ascend hardware. Based on this, we provide a systematic and open description of the data production methods and mixing strategies for all training stages. Furthermore, we present MindVL, a data-efficient multimodal large language model trained end-to-end on Ascend NPUs. In addition, we find that averaging weights from checkpoints trained with different sequence lengths is particularly effective and yields further gains when combined with test-time resolution search. Our experiments demonstrate superior data efficiency: MindVL-8B matches the performance of Qwen2.5VL-7B using only 10\% of its training data, while our MoE model, MindVL-671B-A37B, matches Qwen2.5VL-72B using only 3\% of the Qwen2.5VL training data, and achieves comparable performance with other leading multimodal MoE models. Our work provides the community with a valuable hardware alternative, open data recipes, and effective performance-enhancing techniques.

Related papers

• MiniCPM-V 4.5: Cooking Efficient MLLMs via Architecture, Data, and Training Recipe [68.04078852416248]
  MiniCPM-V 4.5 is an 8B parameter model designed for high efficiency and strong performance.<n>We introduce three core improvements in model architecture, data strategy and training method.<n>MiniCPM-V 4.5 achieves state-of-the-art performance among models under 30B size.
  arXiv  Detail & Related papers  (2025-09-16T19:41:48Z)
• Kwai Keye-VL Technical Report [80.53170317017147]
  We introduce textbfKwai Keye-VL, a multimodal foundation model for short-video understanding.<n>The development of Keye-VL rests on two core pillars: a massive, high-quality dataset with a strong emphasis on video, and an innovative training recipe.<n>To validate our approach, we conduct extensive evaluations, showing that Keye-VL achieves state-of-the-art results on public video benchmarks and remains highly competitive on general image-based tasks.
  arXiv  Detail & Related papers  (2025-07-02T17:57:28Z)
• Pangu Ultra MoE: How to Train Your Big MoE on Ascend NPUs [111.69640966866059]
  Sparse large language models (LLMs) with Mixture of Experts (MoE) and close to a trillion parameters are dominating the realm of most capable language models.<n>In this paper, we aim to uncover a recipe to harness such scale on Ascend NPUs.<n>The key goals are better usage of the computing resources under the dynamic sparse model structures and materializing the expected performance gain on the actual hardware.
  arXiv  Detail & Related papers  (2025-05-07T15:46:36Z)
• InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Models [139.19991097260115]
  We introduce InternVL3, a significant advancement in the InternVL series featuring a native multimodal pre-training paradigm.<n>In particular, InternVL3-78B achieves a score of 72.2 on the MMMU benchmark, setting a new state-of-the-art among open-source MLLMs.<n>In pursuit of open-science principles, we will publicly release both the training data and model weights to foster further research and development in next-generation MLLMs.
  arXiv  Detail & Related papers  (2025-04-14T17:59:25Z)
• Expanding Performance Boundaries of Open-Source Multimodal Models with Model, Data, and Test-Time Scaling [191.7830199016589]
  We introduce InternVL 2.5, an advanced multimodal large language model (MLLM) series that builds upon InternVL 2.0.<n>InternVL 2.5 exhibits competitive performance, rivaling leading commercial models such as GPT-4o and Claude-3.5-Sonnet.<n>We hope this model contributes to the open-source community by setting new standards for developing and applying multimodal AI systems.
  arXiv  Detail & Related papers  (2024-12-06T18:57:08Z)
• NVLM: Open Frontier-Class Multimodal LLMs [64.00053046838225]
  We introduce NVLM 1.0, a family of frontier-class multimodal large language models (LLMs) that achieve state-of-the-art results on vision-language tasks. We propose a novel architecture that enhances both training efficiency and multimodal reasoning capabilities. We develop production-grade multimodality for the NVLM-1.0 models, enabling them to excel in vision-language tasks.
  arXiv  Detail & Related papers  (2024-09-17T17:59:06Z)
• NV-Embed: Improved Techniques for Training LLMs as Generalist Embedding Models [38.41524186248607]
  We introduce NV-Embed, incorporating architectural designs, training procedures, and curated datasets.<n>For model architecture, we propose a latent attention layer to obtain pooled embeddings.<n>For training algorithm, we introduce a two-stage contrastive instruction-tuning method.
  arXiv  Detail & Related papers  (2024-05-27T17:59:45Z)
• Efficient Multimodal Learning from Data-centric Perspective [21.35857180519653]
  We introduce Bunny, a family of lightweight MLLMs with flexible vision and language backbones for efficient multimodal learning. Experiments show that our Bunny-4B/8B outperforms the state-of-the-art large MLLMs on multiple benchmarks.
  arXiv  Detail & Related papers  (2024-02-18T10:09:10Z)
• MoE-LLaVA: Mixture of Experts for Large Vision-Language Models [49.32669226551026]
  We propose a simple yet effective training strategy MoE-Tuning for LVLMs.<n>MoE-LLaVA, a MoE-based sparse LVLM architecture, uniquely activates only the top-k experts through routers.<n>Experiments show the significant performance of MoE-LLaVA in a variety of visual understanding and object hallucination benchmarks.
  arXiv  Detail & Related papers  (2024-01-29T08:13:40Z)
• An Empirical Study of Training End-to-End Vision-and-Language Transformers [50.23532518166621]
  We present METER(textbfMultimodal textbfEnd-to-end textbfTransformtextbfER), through which we investigate how to design and pre-train a fully transformer-based VL model. Specifically, we dissect the model designs along multiple dimensions: vision encoders (e.g., CLIP-ViT, Swin transformer), text encoders (e.g., RoBERTa, DeBERTa), multimodal fusion (e.g., merged attention vs. co-
  arXiv  Detail & Related papers  (2021-11-03T17:55:36Z)

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.