CROCScore

This is the CROCScore model from our publication CROC: Evaluating and Training T2I Metrics with Pseudo- and Human-Labeled Contrastive Robustness Checks.

Usage

For usage details, please view our repository at https://github.com/Gringham/CROC/tree/main.

Citation

@misc{leiter2025crocevaluatingtrainingt2i, title={CROC: Evaluating and Training T2I Metrics with Pseudo- and Human-Labeled Contrastive Robustness Checks}, author={Christoph Leiter and Yuki M. Asano and Margret Keuper and Steffen Eger}, year={2025}, eprint={2505.11314}, archivePrefix={arXiv}, primaryClass={cs.CV}, url={https://arxiv.org/abs/2505.11314}, }

Details

CROCScore is based on Phi4-Multimodal-Instruct (https://huggingface.co/microsoft/Phi-4-multimodal-instruct) that is released under MIT liecense (https://choosealicense.com/licenses/mit/).

Downloads last month
28
Safetensors
Model size
5B params
Tensor type
F32
·
BF16
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Paper for nllg/crocscore