[Question] 能否把图片上传做到每个模型？包括自定义模式？

[deepphp]

🥰 需求描述 | Feature Description

能否把图片上传功能安装到所有模型，包括自定义模型？有的中转把名字设置了，咱们的图片上传只有在选定gpt-4-vision-preview。。就很尴尬

🧐 解决方案 | Proposed Solution

能否把图片上传功能安装到所有模型，包括自定义模型？有的中转把名字设置了，咱们的图片上传只有在选定gpt-4-vision-preview。。就很尴尬

📝 补充信息 | Additional Information

No response

[lobehubbot]

👀 @deepphp

Thank you for raising an issue. We will investigate into the matter and get back to you as soon as possible.
Please make sure you have given us as much context as possible.
非常感谢您提交 issue。我们会尽快调查此事，并尽快回复您。 请确保您已经提供了尽可能多的背景信息。

[lobehubbot]

Bot detected the issue body's language is not English, translate it automatically. 👯👭🏻🧑‍🤝‍🧑👫🧑🏿‍🤝‍🧑🏻👩🏾‍🤝‍👨🏿👬🏿

---

🥰 Requirement Description | Feature Description

Can the image upload feature be installed on all models, including custom models? Some transfers have the name set, and our image upload can only be done when gpt-4-vision-preview is selected. . It's embarrassing

🧐 Solution | Proposed Solution

Can the image upload feature be installed on all models, including custom models? Some transfers have the name set, and our image upload can only be done when gpt-4-vision-preview is selected. . It's embarrassing

📝 Additional Information |

No response

[canisminor1990]

因为目前只有 gpt-4-vision-preview 支持了视觉识别

[lobehubbot]

Bot detected the issue body's language is not English, translate it automatically. 👯👭🏻🧑‍🤝‍🧑👫🧑🏿‍🤝‍🧑🏻👩🏾‍🤝‍👨🏿👬🏿

---

Because currently only gpt-4-vision-preview supports visual recognition

[arvinxx]

@canisminor1990 可能后面还真得这么做。 gpt-4-vision-preview 是不支持 function call 的：#427 (reply in thread)

也就说如果在用 vision model 的同时又需要使用插件的话，现在的实现方案没法支持。用户可能要多个模型间互相切换，感觉灵活性就不太够了。

所以可能更优的解法模式是将 vision 做成一个内置能力，而且这个能力可能也是可插拔替换的，比如可以换成其他视觉识别的模型。

[lobehubbot]

Bot detected the issue body's language is not English, translate it automatically. 👯👭🏻🧑‍🤝‍🧑👫🧑🏿‍🤝‍🧑🏻👩🏾‍🤝‍👨🏿👬🏿

---

@canisminor1990 Maybe I will have to do this later. gpt-4-vision-preview does not support function call: #427 (reply in thread)

In other words, if you need to use a plug-in when using the vision model, the current implementation solution cannot help but support it. Users may have to switch between multiple models to achieve this, which feels like the flexibility is not enough.

Therefore, a better solution may be to make vision a built-in capability, and this capability may also be pluggable and replaceable, for example, it can be replaced with other visual recognition models.

[canisminor1990]

@canisminor1990 可能后面还真得这么做。 gpt-4-vision-preview 是不支持 function call 的：#427 (reply in thread)

也就说如果在用 vision model 的同时又需要使用插件的话，现在的实现方案没法支持。用户可能要多个模型间互相切换，感觉灵活性就不太够了。

所以可能更优的解法模式是将 vision 做成一个内置能力，而且这个能力可能也是可插拔替换的，比如可以换成其他视觉识别的模型。

把图片当成fc走呢

[lobehubbot]

Bot detected the issue body's language is not English, translate it automatically. 👯👭🏻🧑‍🤝‍🧑👫🧑🏿‍🤝‍🧑🏻👩🏾‍🤝‍👨🏿👬🏿

---

@canisminor1990 Maybe I will have to do this later. gpt-4-vision-preview does not support function call: #427 (reply in thread)

In other words, if you need to use a plug-in while using the vision model, the current implementation solution cannot support it. Users may need to switch between multiple models, which feels like the flexibility is not enough.

So perhaps a better solution is to make vision a built-in capability, and this capability may also be pluggable and replaceable, for example, it can be replaced with other visual recognition models.

Think of the picture as fc

[arvinxx]

@canisminor1990 可能后面还真得这么做。 gpt-4-vision-preview 是不支持 function call 的：#427 (reply in thread)
也就说如果在用 vision model 的同时又需要使用插件的话，现在的实现方案没法支持。用户可能要多个模型间互相切换，感觉灵活性就不太够了。
所以可能更优的解法模式是将 vision 做成一个内置能力，而且这个能力可能也是可插拔替换的，比如可以换成其他视觉识别的模型。

把图片当成fc走呢

图片咋当成 fc 走

[lobehubbot]

Bot detected the issue body's language is not English, translate it automatically. 👯👭🏻🧑‍🤝‍🧑👫🧑🏿‍🤝‍🧑🏻👩🏾‍🤝‍👨🏿👬🏿

---

@canisminor1990 Maybe I will have to do this later. gpt-4-vision-preview does not support function call: #427 (reply in thread)
In other words, if you need to use a plug-in while using the vision model, the current implementation plan cannot support it. Users may need to switch between multiple models, which feels like the flexibility is not enough.
So perhaps a better solution is to make vision a built-in capability, and this capability may also be pluggable and replaceable, for example, it can be replaced with other visual recognition models.

Treat the picture as fc

How to take the picture as FC

[lobehubbot]

✅ @deepphp

This issue is closed, If you have any questions, you can comment and reply.
此问题已经关闭。如果您有任何问题，可以留言并回复。