Hugging Face's logo

openbmb
/
MiniCPM-Llama3-V-2_5

Image-Text-to-Text
Transformers
Safetensors
multilingual
minicpmv
feature-extraction
minicpm-v
vision
ocr
custom_code
conversational
Model card Files Community
75
MiniCPM-Llama3-V-2_5 / README.md
finalf0's picture
finalf0
Create README.md
b988d87 verified
preview code
|
raw
history blame
16.4 kB
metadata
pipeline_tag: visual-question-answering
language:
  - en
  - zh
datasets:
  - HaoyeZhang/RLHF-V-Dataset
  - Yirany/UniMM-Chat

GitHub | Demo

MiniCPM-Llama3-V 2.5

MiniCPM-Llama3-V 2.5 is the latest model in the MiniCPM-V series. The model is built on SigLip-400M and Llama3-8B-Instruct with a total of 8B parameters. It exhibits a significant performance improvement over MiniCPM-V 2.0. Notable features of MiniCPM-Llama3-V 2.5 include:

  • πŸ”₯ Leading Performance. MiniCPM-Llama3-V 2.5 has achieved an average score of 65.0 on OpenCompass, a comprehensive evaluation over 11 popular benchmarks. It surpasses widely used proprietary models like GPT-4V-1106, Gemini Pro and Claude 3 with 8B parameters, greatly outperforming other multimodal large models built on Llama 3.

  • πŸ’ͺ Strong OCR Capabilities. MiniCPM-Llama3-V 2.5 can process images with any aspect ratio up to 1.8 million pixels, achieving an 700+ score on OCRBench, surpassing proprietary models such as GPT-4o, GPT-4V-0409, QWEN-VL-Max and Gemini Pro. Based on recent user feedback, MiniCPM-Llama3-V 2.5 has now enhanced full-text OCR extraction, table-to-markdown conversion, and other high-utility capabilities, and has further strengthened its instruction-following and complex reasoning abilities, enhancing multimodal interaction experiences.

  • πŸ† Trustworthy Behavior. Leveraging the latest RLAIF-V method (the newest technology in the RLHF-V [CVPR'24] series), MiniCPM-Llama3-V 2.5 exhibits trustworthy multimodal behavior. It achieves 10.3% hallucination rate on Object HalBench, lower than GPT-4V-1106 (13.6%), achieving the best level within the open-source community.

  • 🌏 Multilingual Support. Thanks to Llama 3’s robust multilingual capabilities and VisCPM's cross-lingual generalization technology, MiniCPM-Llama3-V 2.5 extends its foundational bilingual (Chinese-English) multimodal capabilities to support 30+ languages including German, French, Spanish, Italian, Russian etc. We achieve this extension through only minimal instruction-tuning with translated multimodal data. All Supported Languages.

  • πŸš€ Efficient Deployment. MiniCPM-Llama3-V 2.5 systematically employs model quantization, CPU optimizations, NPU optimizations and compilation optimizations as acceleration techniques, achieving high-efficiency deployment on edge devices. For mobile phones with Qualcomm chips, we have integrated the NPU acceleration framework QNN into llama.cpp for the first time. After systematic optimization, MiniCPM-Llama3-V 2.5 has realized a 150-fold acceleration in multimodal large model edge-side image encoding and a 3-fold increase in language decoding speed.

Evaluation

Click to view results on TextVQA, DocVQA, OCRBench, OpenCompass, MME, MMBench, MMMU, MathVista, LLaVA Bench, RealWorld QA, Object HalBench.
Results on OCR-specific and General Multimodal Benchmarks
Model Size OCRBench TextVQA val DocVQA test Open-Compass MME MMB dev (en) MMB dev (zh) MMMU val Math-Vista LLaVA Bench RealWorld QA Object HalBench
Proprietary
Gemini Pro - 680 74.6 88.1 63.8 2148.9 75.2 74.0 48.9 45.8 79.9 60.4 -
GPT-4V (2023.11.06) - 645 78.0 88.4 63.2 1771.5 75.1 75.0 53.8 47.8 93.1 63.0 86.4
Open-source
DeepSeek-VL-1.3B 1.7B 413 58.4* 37.9* 46.0 1531.6 64.0 61.2 33.8 29.4 51.1 49.7 -
Mini-Gemini 2.2B - 56.2 34.2* - 1653.0 59.8 - 31.7 - - - -
Yi-VL-6B 6.7B 290 45.5* 17.1* 49.3 1915.1 68.6 68.3 40.3 28.8 51.9 53.5 -
Qwen-VL-Chat 9.6B 488 61.5 62.6 52.1 1860.0 60.6 56.7 37.0 33.8 67.7 49.3 56.2 / 80.0
DeepSeek-VL-7B 7.3B 435 64.7* 47.0* 55.6 1765.4 74.1 72.8 38.3 36.8 77.8 54.2 91.5 / 95.3
Yi-VL-34B 34B 290 43.4* 16.9* 52.6 2050.2 71.1 71.4 45.1 30.7 62.3 54.8 79.3 / 86.0
CogVLM-Chat 17.4B 590 70.4 33.3* 52.5 1736.6 63.7 53.8 37.3 34.7 73.9 60.3 73.6 / 87.4
TextMonkey 9.7B 558 64.3 66.7 - - - - - - - - -
IDEFICS2-8B 8.0B - - - 57.2 1847.6 75.7 68.6 45.2 52.2 49.1 60.7 -
Bunny-LLama-3-8B 8.4B - - - 54.3 1920.3 77.0 73.9 41.3 31.5 61.2 58.8 -
XTuner-Llama-3-8B-v1.1 8.4B - - - 53.3 1818.0 71.7 63.2 39.2 40.0 69.2 - -
LLaVA-NeXT Llama-3-8B 8.4B - - 78.2 - 1971.5 72.1 - 41.7 37.5 80.1 60.0 -
MiniCPM-V 1.0 2.8B 366 60.6 38.2 47.6 1650.2 67.9 65.3 38.3 28.9 51.3 51.2 78.4 / 88.5
MiniCPM-V 2.0 2.8B 605 74.1 71.9 55.0 1808.6 69.6 68.1 38.2 38.7 69.2 55.8 85.5 / 92.2
MiniCPM-Llama3-V 2.5 8.5B 725 76.6 84.8 65.0 2024.6 76.7 73.4 45.8 54.3 86.7 63.5 89.7 / 95.0
* We evaluate the officially released checkpoint by ourselves.

Examples

We deploy MiniCPM-V 2.0 on end devices. The demo video is the raw screen recording on a Xiaomi 14 Pro at double speed.

Demo

Click here to try out the Demo of MiniCPM-Llama3-V 2.5.

Deployment on Mobile Phone

MiniCPM-Llama3-V 2.5 can be deployed on mobile phones with Android operating systems. πŸš€ Try it out here.

Usage

Inference using Huggingface transformers on Nivdia GPUs. Requirements tested on python 3.10：

Pillow==10.1.0
torch==2.1.2
torchvision==0.16.2
transformers==4.40.0
sentencepiece==0.1.99

# test.py
import torch
from PIL import Image
from transformers import AutoModel, AutoTokenizer

model = AutoModel.from_pretrained('openbmb/MiniCPM-Llama3-V-2_5', trust_remote_code=True, torch_dtype=torch.float16)
model = model.to(device='cuda')

tokenizer = AutoTokenizer.from_pretrained('openbmb/MiniCPM-Llama3-V-2_5', trust_remote_code=True)
model.eval()

image = Image.open('xx.jpg').convert('RGB')
question = 'What is in the image?'
msgs = [{'role': 'user', 'content': question}]

res = model.chat(
    image=image,
    msgs=msgs,
    tokenizer=tokenizer,
    sampling=True,
    temperature=0.7
)
print(res)

Please look at GitHub for more detail about usage.

MiniCPM-V 2.0

Please see the info about MiniCPM-V 2.0 here.

License

Model License

Statement

  • As a LLM, MiniCPM-Llama3-V 2.5 generates contents by learning a large mount of texts, but it cannot comprehend, express personal opinions or make value judgement. Anything generated by MiniCPM-Llama3-V 2.5 does not represent the views and positions of the model developers
  • We will not be liable for any problems arising from the use of the MinCPM-V open Source model, including but not limited to data security issues, risk of public opinion, or any risks and problems arising from the misdirection, misuse, dissemination or misuse of the model.

Other Multimodal Projects from Our Team

VisCPM | RLHF-V | LLaVA-UHD

Citation

If you find our work helpful, please consider citing the following papers

@article{yu2023rlhf,
  title={Rlhf-v: Towards trustworthy mllms via behavior alignment from fine-grained correctional human feedback},
  author={Yu, Tianyu and Yao, Yuan and Zhang, Haoye and He, Taiwen and Han, Yifeng and Cui, Ganqu and Hu, Jinyi and Liu, Zhiyuan and Zheng, Hai-Tao and Sun, Maosong and others},
  journal={arXiv preprint arXiv:2312.00849},
  year={2023}
}
@article{viscpm,
    title={Large Multilingual Models Pivot Zero-Shot Multimodal Learning across Languages}, 
    author={Jinyi Hu and Yuan Yao and Chongyi Wang and Shan Wang and Yinxu Pan and Qianyu Chen and Tianyu Yu and Hanghao Wu and Yue Zhao and Haoye Zhang and Xu Han and Yankai Lin and Jiao Xue and Dahai Li and Zhiyuan Liu and Maosong Sun},
    journal={arXiv preprint arXiv:2308.12038},
    year={2023}
}
@article{xu2024llava-uhd,
  title={{LLaVA-UHD}: an LMM Perceiving Any Aspect Ratio and High-Resolution Images},
  author={Xu, Ruyi and Yao, Yuan and Guo, Zonghao and Cui, Junbo and Ni, Zanlin and Ge, Chunjiang and Chua, Tat-Seng and Liu, Zhiyuan and Huang, Gao},
  journal={arXiv preprint arXiv:2403.11703},
  year={2024}
}