--- license: mit metrics: - accuracy - mean_iou --- # Model Card for InternVL This repository contains the PyTorch version of the InternVL model weights. ## What is InternVL? \[[Paper](https://arxiv.org/abs/2312.14238)\] \[[GitHub](https://github.com/OpenGVLab/InternVL)\] \[[Chat Demo](https://internvl.opengvlab.com/)\] InternVL scales up the ViT to _**6B parameters**_ and aligns it with LLM. It is _**the largest open-source vision/vision-language foundation model (14B)**_ to date, achieving _**32 state-of-the-art**_ performances on a wide range of tasks such as visual perception, cross-modal retrieval, multimodal dialogue, etc. ![image/png](https://cdn-uploads.huggingface.co/production/uploads/64119264f0f81eb569e0d569/k5UATwX5W2b5KJBN5C58x.png) ## Pretrained Weights | model name | type | download | size | | ----------------------- | ------- | ---------------------------------------------------------------------------------------------- | :-----: | | InternViT-6B-224px | pytorch | 🤗 [HF link](https://huggingface.co/OpenGVLab/InternVL/blob/main/intern_vit_6b_224px.pth) | 12 GB | | InternVL-C-13B-224px | pytorch | 🤗 [HF link](https://huggingface.co/OpenGVLab/InternVL/blob/main/internvl_c_13b_224px.pth) | 25.4 GB | ## Linear-Probe Image Classification (ImageNet Series) | model name | IN-1K | IN-ReaL | IN-V2 | IN-A | IN-R | IN-Sketch | download | | ------------------ | :---: | :-----: | :---: | :--: | :--: | :-------: | :---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: | | InternViT-6B-224px | 88.2 | 90.4 | 79.9 | 77.5 | 89.8 | 69.1 | [ckpt](https://huggingface.co/OpenGVLab/InternVL/resolve/main/intern_vit_6b_224px_head.pth) \| [log](https://github.com/OpenGVLab/InternVL/blob/main/classification/work_dirs/intern_vit_6b_1k_224/log_rank0.txt) | ## Semantic Segmentation (ADE20K) | type | backbone | head | mIoU | config | download | | --------------- | --------------------- | :-----: | :--: | :--------------------------------------------------------------------------------------------------------: | :-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: | | few-shot (1/16) | InternViT-6B | Linear | 46.5 | [config](https://github.com/OpenGVLab/InternVL/blob/main/segmentation//configs/intern_vit_6b/few_shot/linear_intern_vit_6b_504_5k_ade20k_bs16_lr4e-5_1of16.py) | [ckpt](https://huggingface.co/OpenGVLab/InternVL/resolve/main/linear_intern_vit_6b_504_5k_ade20k_bs16_lr4e-5_1of16.pth) \| [log](https://huggingface.co/OpenGVLab/InternVL/raw/main/linear_intern_vit_6b_504_5k_ade20k_bs16_lr4e-5_1of16.log) | | few-shot (1/8) | InternViT-6B | Linear | 50.0 | [config](https://github.com/OpenGVLab/InternVL/blob/main/segmentation//configs/intern_vit_6b/few_shot/linear_intern_vit_6b_504_10k_ade20k_bs16_lr4e-5_1of8.py) | [ckpt](https://huggingface.co/OpenGVLab/InternVL/resolve/main/linear_intern_vit_6b_504_10k_ade20k_bs16_lr4e-5_1of8.pth) \| [log](https://huggingface.co/OpenGVLab/InternVL/raw/main/linear_intern_vit_6b_504_10k_ade20k_bs16_lr4e-5_1of8.log) | | few-shot (1/4) | InternViT-6B | Linear | 53.3 | [config](https://github.com/OpenGVLab/InternVL/blob/main/segmentation//configs/intern_vit_6b/few_shot/linear_intern_vit_6b_504_20k_ade20k_bs16_lr4e-5_1of4.py) | [ckpt](https://huggingface.co/OpenGVLab/InternVL/resolve/main/linear_intern_vit_6b_504_20k_ade20k_bs16_lr4e-5_1of4.pth) \| [log](https://huggingface.co/OpenGVLab/InternVL/raw/main/linear_intern_vit_6b_504_20k_ade20k_bs16_lr4e-5_1of4.log) | | few-shot (1/2) | InternViT-6B | Linear | 55.8 | [config](https://github.com/OpenGVLab/InternVL/blob/main/segmentation//configs/intern_vit_6b/few_shot/linear_intern_vit_6b_504_40k_ade20k_bs16_lr4e-5_1of2.py) | [ckpt](https://huggingface.co/OpenGVLab/InternVL/resolve/main/linear_intern_vit_6b_504_40k_ade20k_bs16_lr4e-5_1of2.pth) \| [log](https://huggingface.co/OpenGVLab/InternVL/raw/main/linear_intern_vit_6b_504_40k_ade20k_bs16_lr4e-5_1of2.log) | | few-shot (1/1) | InternViT-6B | Linear | 57.2 | [config](https://github.com/OpenGVLab/InternVL/blob/main/segmentation//configs/intern_vit_6b/few_shot/linear_intern_vit_6b_504_80k_ade20k_bs16_lr4e-5_1of1.py) | [ckpt](https://huggingface.co/OpenGVLab/InternVL/resolve/main/linear_intern_vit_6b_504_80k_ade20k_bs16_lr4e-5_1of1.pth) \| [log](https://huggingface.co/OpenGVLab/InternVL/raw/main/linear_intern_vit_6b_504_80k_ade20k_bs16_lr4e-5_1of1.log) | | linear probing | InternViT-6B (frozen) | Linear | 47.2 | [config](https://github.com/OpenGVLab/InternVL/blob/main/segmentation//configs/intern_vit_6b/linear_probing/linear_intern_vit_6b_504_80k_ade20k_bs16_lr4e-5_frozen.py) | [ckpt](https://huggingface.co/OpenGVLab/InternVL/resolve/main/linear_intern_vit_6b_504_80k_ade20k_bs16_lr4e-5_frozen.pth) \| [log](https://huggingface.co/OpenGVLab/InternVL/raw/main/linear_intern_vit_6b_504_80k_ade20k_bs16_lr4e-5_frozen.log) | | head tuning | InternViT-6B (frozen) | UperNet | 54.9 | [config](https://github.com/OpenGVLab/InternVL/blob/main/segmentation//configs/intern_vit_6b/head_tuning/upernet_intern_vit_6b_504_80k_ade20k_bs16_lr4e-5_frozen.py) | [ckpt](https://huggingface.co/OpenGVLab/InternVL/resolve/main/upernet_intern_vit_6b_504_80k_ade20k_bs16_lr4e-5_frozen.pth) \| [log](https://huggingface.co/OpenGVLab/InternVL/raw/main/upernet_intern_vit_6b_504_80k_ade20k_bs16_lr4e-5_frozen.log) | | full tuning | InternViT-6B | UperNet | 58.9 | [config](https://github.com/OpenGVLab/InternVL/blob/main/segmentation//configs/intern_vit_6b/full_tuning/upernet_intern_vit_6b_504_80k_ade20k_bs16_lr4e-5.py) | [ckpt](https://huggingface.co/OpenGVLab/InternVL/resolve/main/upernet_intern_vit_6b_504_80k_ade20k_bs16_lr4e-5.pth) \| [log](https://huggingface.co/OpenGVLab/InternVL/raw/main/upernet_intern_vit_6b_504_80k_ade20k_bs16_lr4e-5.log) | ## License This project is released under the MIT license. Parts of this project contain code and models from other sources, which are subject to their respective licenses. ## Citation If you find this project useful in your research, please consider cite: ```BibTeX @article{chen2023internvl, title={InternVL: Scaling up Vision Foundation Models and Aligning for Generic Visual-Linguistic Tasks}, author={Chen, Zhe and Wu, Jiannan and Wang, Wenhai and Su, Weijie and Chen, Guo and Xing, Sen and Zhong, Muyan and Zhang, Qinglong and Zhu, Xizhou and Lu, Lewei and Li, Bin and Luo, Ping and Lu, Tong and Qiao, Yu and Dai, Jifeng}, journal={arXiv preprint arXiv:2312.14238}, year={2023} } ``` ## Acknowledgement InternVL is built with reference to the code of the following projects: [OpenAI CLIP](https://github.com/openai/CLIP), [Open CLIP](https://github.com/mlfoundations/open_clip), [CLIP Benchmark](https://github.com/LAION-AI/CLIP_benchmark), [EVA](https://github.com/baaivision/EVA/tree/master), [InternImage](https://github.com/OpenGVLab/InternImage), [ViT-Adapter](https://github.com/czczup/ViT-Adapter), [MMSegmentation](https://github.com/open-mmlab/mmsegmentation), [Transformers](https://github.com/huggingface/transformers), [DINOv2](https://github.com/facebookresearch/dinov2), [BLIP-2](https://github.com/salesforce/LAVIS/tree/main/projects/blip2), [Qwen-VL](https://github.com/QwenLM/Qwen-VL/tree/master/eval_mm), and [LLaVA-1.5](https://github.com/haotian-liu/LLaVA). Thanks for their awesome work!