SoM-LLaVA Model Card

LLaVA-v1.5 mixed trained with SoM style data (QA+listing).

The model can understand tag-style visual prompts on the image (e.g., what is the object tagged with id 9?), also gained improved performance on MLLM benchmarks (POPE, MME, SEED, MM-Vet, LLav-wild), even when the input testing images has no tags.

For more information about SoM-LLaVA, check our github page and paper!

Getting Started

If you would like to load our model in huggingface, here is an example script:

from PIL import Image
import requests
from transformers import AutoProcessor, LlavaForConditionalGeneration

model_path = "zzxslp/som-llava-v1.5-13b-hf"

model = LlavaForConditionalGeneration.from_pretrained(model_path)
processor = AutoProcessor.from_pretrained(model_path)

prompt = "USER: <image>\nWhat's the content of the image? ASSISTANT:"
url = "https://www.ilankelman.org/stopsigns/australia.jpg"
image = Image.open(requests.get(url, stream=True).raw)

inputs = processor(text=prompt, images=image, return_tensors="pt")

# Generate
generate_ids = model.generate(**inputs, max_new_tokens=20)
output = processor.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0]
print (output)

Our original model weights: [SoM-LLaVA-v1.5-13B], to be used in official LLaVA repo

Citation

If you find our data or model useful for your research and applications, please cite our paper:

@article{yan2024list,
  title={List Items One by One: A New Data Source and Learning Paradigm for Multimodal LLMs},
  author={Yan, An and Yang, Zhengyuan and Wu, Junda and Zhu, Wanrong and Yang, Jianwei and Li, Linjie and Lin, Kevin and Wang, Jianfeng and McAuley, Julian and Gao, Jianfeng and others},
  journal={arXiv preprint arXiv:2404.16375},
  year={2024}
}