Miracle

Hi, thanks for your interest. As I answered in this [issue](https://github.com/THUDM/CogCoM/issues/3), you may follow the requirements.txt to ensure the proper versions of `pydantic` and `gradio` (can use --no-deps to ignore...

Hi，感谢你对我们工作的关注和提问。因为我们默认CogCoM是一个针对图像的多模态模型，所以在训练时没有考虑输入不包含图像的情况。然而，你可以通过将`chat.py`中198～228行的代码替换为如下代码，使得模型兼容兼容不包含图像输入的情况： ``` # if image_position < 5: # no image # inputs = text_processor.tokenizer([prompt], return_tensors="pt").to(model.parameters().__next__().device)['input_ids'][0] # # pre_image = 0 # else: new_prompt = prompt[image_position:] if image_position >= 5 else...

> I am not able to find the model on Hugging Face. Can you help me out how to find the model in Hugging Face? Also Gradio is not working...

> > > I am not able to find the model on Hugging Face. Can you help me out how to find the model in Hugging Face? Also Gradio is...

> I tried to run the gradio demo as follows: > > `python web_demo.py --from_pretrained cogcom-base-17b --local_tokenizer lmsys/vicuna-7b-v1.5 --bf16` > > resulting in this error: > > ``` > drawn_imgs...

> Hi, > > Interesting work, but I have a question regarding the grounding process used for the CoM samples. > > In the paper, you mention using a grounding...

@Wanglvcheng Hi Lvcheng, thank you for your interest in our work. The LINE function was added based on our manually annotated data, CoM-Math, that we released later, after the first...