Model type: Finedefics is an open-source MLLM that enhances the model's FGVR capability by incorporating informative attribute descriptions of objects into the training phase. It is an auto-regressive language model, based on the transformer architecture. Base MLLM: HuggingFaceM4/idefics2-8b

Paper or resources for more information: OpenReview: https://openreview.net/forum?id=p3NKpom1VL Arxiv: https://arxiv.org/abs/2501.15140

Idefics2 is licensed under the Apache 2.0 license, and we release the Finedefics checkpoints under the same license.

Where to send questions or comments about the model: https://github.com/PKU-ICST-MIPL/Finedefics_ICLR2025/issues

Primary intended uses: The primary use of Finedefics is research on Fine-grained MLLM.

Primary intended users: The primary intended users of the model are researchers and hobbyists in computer vision, natural language processing, machine learning, and artificial intelligence.

A collection of 6 fine-grained visual recognition datasets, including Stanford Dog-120, Bird-200, FGVC-Aircraft, Flower-102, Oxford-IIIT Pet-37, and Stanford Car-196.