uta-smile/instructcir_llava_phi35_clip224_lp

GitHub Repository: https://github.com/zhwl2117/InstructCIR

NOTE: InstructCIR adopts the training codes from xtuner/llava-phi-3-mini-hf but uses ViT-L-224 and Phi-3.5-Mini for consistency with existing CIR models

@article{zhong2024compositional,
  title={Compositional Image Retrieval via Instruction-Aware Contrastive Learning},
  author={Zhong, Wenliang and An, Weizhi and Jiang, Feng and Ma, Hehuan and Guo, Yuzhi and Huang, Junzhou},
  journal={arXiv preprint arXiv:2412.05756},
  year={2024}
}