GUOGUO-lab comments

Repositories
Issues
Comments

Results 2 comments of


                                            GUOGUO-lab

在网页版的demo中推理仅仅只能进行语言模型的对话，没有多模态的推理

> 是的, 网页端暂时不支持多模态模型的推理, 可以使用`swift infer`进行命令行的推理此前的1.5.4版本可以加载qwen-vl-chat，图片的推理也可以用“\xxx.jpg\“来进行输入，结果到了1.7.0这个也不支持了，请问这个能恢复吗？

Question on multiple image inputs at inference time

Maybe you can edit the gradio part to create more image components, and add them as "\path_to.jpg\" to your text input (I use the demo from swift, maybe diff from...