Fully Open Framework for Democratized Multimodal Training

🤗 Models and Datasets | 🖥️ Demo | 📄 Tech Report

---

Contents

• Introduction
• Models
• Datasets
• Results
• Quick Start with HuggingFace
• Evaluation
• Quick Start For Training
• Fully Reproducing Guide
• Citation
• Acknowledgement

Introduction

LLaVA-OneVision1.5 introduces a novel family of fully open-source Large Multimodal Models (LMMs) that achieves state-of-the-art performance with substantially lower cost through training on native resolution images.

• Superior Performance A family of fully open-source large multimodal models demonstrating

  • Superior performance across multiple multimodal benchmarks
  • outperforming Qwen2.5-VL in most evaluation tasks.

• High-Quality Data at Scale Meticulously curated pre-training and SFT data with rigorous filtering and quality control, achieving superior data efficiency with only 64B tokens.

  • Concept-balanced, highly diverse, high-quality caption data
  • Comprehensive instruction fine-tuning data covering a wide range of tasks

• Ultra-Efficient Training Framework Complete end-to-end training framework designed for maximum efficiency:

  • $16000 total budget for full model training on A100 GPUs ($0.6 per GPU/Hour)
  • 45% HFU efficiency in 8k context length
  • Built on MegatronLM with support for MoE, FP8, and long sequence parallelization
  • Optimized codebase for cost-effective scaling

• Fully Open Framework for community access and reproducibility:

  • High-quality pre-training & SFT data
  • Complete training framework & code
  • Training recipes & configurations
  • Comprehensive training logs & metrics

Models

Model	HF Link	Training Log
LLaVA-OV-1.5-4B-Instruct	🤗 HF / 4B-Instruct	📈 Tensorboard
LLaVA-OV-1.5-8B-Instruct	🤗 HF / 8B-Instruct	📈 Tensorboard

Datasets

(a) The vocabulary coverage proportion in the LLaVA-OneVision-1.5 Mid-Training dataset before and after concept balancing. (b) Distribution of data sources within the LLaVA-OneVision-1.5 Mid-Training dataset. (c) Distribution of data sources within the LLaVA-OneVision-1.5 Insturct dataset.

Description	Link	Status
LLaVA-OV-1.5-Mid-Training-85M	🤗HF / Mid-Training 85M	Uploading…
LLaVA-OV-1.5-Instruct	🤗HF / Insturct-Data	Uploading…

Evaluation Results

All evaluations were conducted using lmms_eval.

Quick Start with HuggingFace

from transformers import AutoTokenizer, AutoProcessor, AutoModelForCausalLM
from qwen_vl_utils import process_vision_info
model_path = "lmms-lab/LLaVA-One-Vision-1.5-8B-Instruct"

# default: Load the model on the available device(s)
model = AutoModelForCausalLM.from_pretrained(
    model_path, torch_dtype="auto", device_map="auto", trust_remote_code=True
)

# default processer
processor = AutoProcessor.from_pretrained(model_path, trust_remote_code=True)

messages = [
    {
        "role": "user",
        "content": [
            {
                "type": "image",
                "image": "https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen-VL/assets/demo.jpeg",
            },
            {"type": "text", "text": "Describe this image."},
        ],
    }
]

# Preparation for inference
text = processor.apply_chat_template(
    messages, tokenize=False, add_generation_prompt=True
)
image_inputs, video_inputs = process_vision_info(messages)
inputs = processor(
    text=[text],
    images=image_inputs,
    videos=video_inputs,
    padding=True,
    return_tensors="pt",
)
inputs = inputs.to("cuda")

# Inference: Generation of the output
generated_ids = model.generate(**inputs, max_new_tokens=1024)
generated_ids_trimmed = [
    out_ids[len(in_ids) :] for in_ids, out_ids in zip(inputs.input_ids, generated_ids)
]
output_text = processor.batch_decode(
    generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False
)
print(output_text)

Evaluation

# pip install git+https://github.com/EvolvingLMMs-Lab/lmms-eval.git  

accelerate launch --num_processes=8 --main_process_port 12399 -m lmms_eval \
    --model=llava_onevision1_5 \
    --model_args=pretrained=lmms-lab/LLaVA-OneVision-1.5-8B-Instruct,attn_implementation=flash_attention_2,max_pixels=3240000 \
    --tasks=mmmu_val,mmmu_pro_standard,mmbench_en_test,mmerealworld,mmerealworld_cn,ai2d,ai2d_no_mask,vstar_bench,chartqa,charxiv,docvqa_test,mathvista_testmini,mmstar,scienceqa \
    --batch_size=1

Quick Start Guide

1.🐳 Docker (Recommended)

We strongly recommend using the docker environment for a seamless experience. The following instructions are tailored for the A100 80GB GPU environment.

# Clone repository
git clone https://github.com/EvolvingLMMs-Lab/LLaVA-OneVision-1.5.git
cd LLaVA-OneVision-1.5

docker build -t llava_megatron:25.04 .

# Run container with -w to set working directory directly to the mounted volume
docker run -it --gpus all \
    --ipc host --net host --privileged --cap-add IPC_LOCK \
    --ulimit memlock=-1 --ulimit stack=67108864 --rm \
    -v $(pwd):/workspace/LLaVA-OneVision-1.5 \
    -w /workspace/LLaVA-OneVision-1.5 \
    --name "llava_megatron_container" \
    llava_megatron:25.04 /bin/bash

2. Checkpoint and Format Conversion

You have two options to get started with LLaVA-OneVision-1.5-stage-0:

Option 1: Download pre-trained model from HuggingFace

Download our LLaVA-OneVision-1.5-4B-stage0 model directly from HuggingFace.

Option 2: Merge initial weights yourself

Alternatively, you can merge the initial weights from the original ViT and LLM:

python ds/merge_model.py \
--vit_path DeepGlint-AI/rice-vit-large-patch14-560 \
--llm_path Qwen/Qwen3-4B-Instruct-2507 \
--output LLaVA-OneVision-1.5-4B-stage0

Note: When merging weights, the adapter component will be initialized with default values.

Convert the model from HuggingFace format to Megatron format:

AIAK_TRAINING_PATH=/workspace/LLaVA-OneVision-1.5 bash examples/llava_ov_1_5/convert/convert_4b_hf_to_mcore.sh \
LLaVA-OneVision-1.5-4B-stage0 \
LLaVA-OneVision-1.5-4B-stage0_mcore_tp1_pp1 \
1 1

3. Stage 1 Alignment-Training

Download LLaVA from LLaVA-558K-Webdataset.

# ============================================================
# Required environment variables:
#   AIAK_TRAINING_PATH  Root directory of the AIAK-Training-LLM project
#   DATA_PATH           Directory with WebDataset shards (.tar) for pretraining
#   TOKENIZER_PATH      Hugging Face tokenizer directory
#   CHECKPOINT_PATH     Megatron-formatted checkpoint directory (e.g., mcore TP1/PP1)
#   SAVE_CKPT_PATH      Output directory for saving training checkpoints
AIAK_TRAINING_PATH=/workspace/LLaVA-OneVision-1.5 \
DATA_PATH=LLaVA-558K-Webdataset \
TOKENIZER_PATH=LLaVA-OneVision-1.5-4B-stage0 \
CHECKPOINT_PATH=LLaVA-OneVision-1.5-4B-stage0_mcore_tp1_pp1 \
bash examples/llava_ov_1_5/quick_start/stage_1_alignment_llava_ov_4b.sh

4. Stage 1.5 Mid-Training

Download our lightweight packed subset from LLaVA-OneVision-1.5-Mid-Training-Quick-Start-3M-Webdataset.

# ============================================================
# Convert model to release format
bash examples/llava_ov_1_5/convert/convert_4b_mcore_to_release.sh \
stage_1_alignment_llava_ov_4b/iter_0002500/ \
stage_1_alignment_llava_ov_4b_release 1 1
# ============================================================
# Launch
AIAK_TRAINING_PATH=/workspace/LLaVA-OneVision-1.5 \
DATA_PATH=LLaVA-OneVision-1.5-Mid-Training-Quick-Start-3M-Webdataset \
TOKENIZER_PATH=LLaVA-OneVision-1.5-4B-stage0 \
CHECKPOINT_PATH=stage_1_alignment_llava_ov_4b_release \
bash examples/llava_ov_1_5/quick_start/stage_1.5_mid_training_llava_ov_4b.sh

5. Stage 2 Instruct-Training

Download LLaVA-NeXT-780k-webdataset at LLaVA-NeXT-780K Dataset.

# ============================================================
# Convert model to release format
bash examples/llava_ov_1_5/convert/convert_4b_mcore_to_release.sh \
stage_1.5_mid_training_llava_ov_4b/iter_0020000/ \
stage_1.5_mid_training_llava_ov_4b_release 1 1
# ============================================================
# # Launch
AIAK_TRAINING_PATH=/workspace/LLaVA-OneVision-1.5 \
DATA_PATH=LLaVA-NeXT-780k-Webdataset \
TOKENIZER_PATH=LLaVA-OneVision-1.5-4B-stage0 \
CHECKPOINT_PATH=stage_1.5_mid_training_llava_ov_4b_release \
bash examples/llava_ov_1_5/quick_start/stage_2_instruct_llava_ov_4b.sh

6. Convert mcore to huggingface

AIAK_TRAINING_PATH=/workspace/LLaVA-OneVision-1.5 \
bash examples/llava_ov_1_5/convert/convert_4b_mcore_to_hf.sh \
stage_2_instruct_llava_ov_4b/iter_0003500 \
LLaVA-OneVision-1.5-4B-3M-Mid-Training-780K-Instruct \
1 1
# Copy non-model files (e.g., tokenizer config) to the new directory
find LLaVA-OneVision-1.5-4B-stage0/ -type f -not -iname '*safetensors*' -exec cp {}  LLaVA-OneVision-1.5-4B-3M-Mid-Training-780K-Instruct/ ';'

7. Evaluation

# pip install git+https://github.com/EvolvingLMMs-Lab/lmms-eval.git
CUDA_VISIBLE_DEVICES=4,5,6,7 accelerate launch \
--num_processes=4 --main_process_port 12399 -m lmms_eval --model=llava_onevision1_5 --batch_size=1 --tasks=mme \
--model_args=pretrained=/workspace/LLaVA-OneVision-1.5/LLaVA-OneVision-1.5-4B-3M-Mid-Training-780K-Instruct,max_pixels=3240000

Fully Reproducing Guide

Tip

More detailed reproduction steps for the complete process will be provided after the dataset upload is completed.

Mid-Training

To improve model training efficiency, we implement offline sample packing:

1. Download the Mid-Training-85M Dataset
2. Pack the data into webdataset format, refer to Offline Padding-Free Data Packing

Instruct

1. Download the LLaVA-OneVision-1.5-Insturct-Data
2. Convert the data into webdataset format, refer to Conversion for Mixed Instruction Data

Roadmaps

Q4 2025 Key Deliverables:

1. Ultra-efficient MoE Training
2. Full Video Input LLM

Contributors

Thanks so much to all of our amazing contributors!

jiankangdeng

anxiangsir

yiyexy

wideyard

chengzheng345

mathCrazyy

yunglechao

RobitYadda

Citation

If you find LLaVA-OneVision-1.5 useful in your research, please consider to cite the following related papers:

@inproceedings{LLaVA-OneVision-1.5,
  title={LLaVA-OneVision-1.5: Fully Open Framework for Democratized Multimodal Training},
  author={An, Xiang and Xie, Yin and Yang, Kaicheng and Zhang, Wenkang and Zhao, Xiuwei and Cheng, Zheng and Wang, Yirui and Xu, Songcen and Chen, Changrui and Wu, Chunsheng and Tan, Huajie and Li, Chunyuan and Yang, Jing and Yu, Jie and Wang, Xiyao and Qin, Bin and Wang, Yumeng and Yan, Zizhen and Feng, Ziyong and Liu, Ziwei and Li, Bo and Deng, Jiankang},
  booktitle={arxiv},  
  year={2025}
 }

@inproceedings{xie2025region,
  title={Region-based Cluster Discrimination for Visual Representation Learning},
  author={Xie, Yin and Yang, Kaicheng and An, Xiang and Wu, Kun and Zhao, Yongle and Deng, Weimo and Ran, Zimin and Wang, Yumeng and Feng, Ziyong and Miles, Roy and Elezi, Ismail and Deng, Jiankang},
  booktitle={ICCV},
  year={2025}
}

@article{lillava,
  title={LLaVA-OneVision: Easy Visual Task Transfer},
  author={Li, Bo and Zhang, Yuanhan and Guo, Dong and Zhang, Renrui and Li, Feng and Zhang, Hao and Zhang, Kaichen and Zhang, Peiyuan and Li, Yanwei and Liu, Ziwei and Li, Chunyuan},
  journal={Transactions on Machine Learning Research}
  year={2024}
}

Acknowledgement

We extend our sincere gratitude to AIAK team of the Baige AI computing platform from Baidu AI Cloud for providing the exceptional training framework. The outstanding capabilities of AIAK-Training-LLM and AIAK-Megatron have significantly accelerated our training process with remarkable efficiency. These cutting-edge frameworks have been instrumental in achieving our research goals. To get full AIAK support, you can contact Baidu Cloud.

We also thank the maintainers and contributors of the following open-source projects, whose work greatly inspired and supported our research:

• LLaVA: Large Language-and-Vision Assistant — LLaVA
• LLaVA-NeXT: Next-generation multi-modal assistant — LLaVA-NeXT
• lmms-eval: A standardized evaluation framework for Large Multimodal Models — lmms-eval
• Megatron-LM: Efficient, scalable training for large language models — Megatron-LM
• Qwen2.5-VL: Strong vision-language foundation model — Qwen2.5-VL
• Qwen3: Next-generation Qwen LLM — Qwen
• MetaCLIP: Scalable contrastive pretraining — MetaCLIP