Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[Bug] 自定义模型支持图片输入,但选择自定义模型时默认无图片输入按钮 #4663

Open
3 tasks
desktop4025 opened this issue May 11, 2024 · 21 comments
Labels
bug Something isn't working

Comments

@desktop4025
Copy link

Bug Description

自定义模型支持图片输入,但选择自定义模型时默认无图片输入按钮

Steps to Reproduce

Expected Behavior

希望把图片上传按钮加入

Screenshots

No response

Deployment Method

  • Docker
  • Vercel
  • Server

Desktop OS

No response

Desktop Browser

No response

Desktop Browser Version

No response

Smartphone Device

No response

Smartphone OS

No response

Smartphone Browser

No response

Smartphone Browser Version

No response

Additional Logs

No response

@desktop4025 desktop4025 added the bug Something isn't working label May 11, 2024
@Issues-translate-bot
Copy link

Bot detected the issue body's language is not English, translate it automatically.


Title: [Bug] Custom models support image input, but when selecting a custom model, there is no image input button by default.

@QAbot-zh
Copy link

你把模型名称修改为带vision的就有了

@Issues-translate-bot
Copy link

Bot detected the issue body's language is not English, translate it automatically.


You can change the model name to include vision.

@desktop4025
Copy link
Author

你把模型名称修改为带vision的就有了

可是我是用的中转站如果在结尾加上vison就识别不到这个模型了欸

@Issues-translate-bot
Copy link

Bot detected the issue body's language is not English, translate it automatically.


You can change the model name to include vision.

But the transfer station I am using will not be able to recognize the model if vison is added at the end.

@QAbot-zh
Copy link

你把模型名称修改为带vision的就有了

可是我是用的中转站如果在结尾加上vison就识别不到这个模型了欸

自己的中转站还是上游的,如果是自己部署的,类似oneapi、newapi那种,可以自己重定向一下模型名称

@Issues-translate-bot
Copy link

Bot detected the issue body's language is not English, translate it automatically.


You can change the model name to include vision.

But the transfer station I am using will not be able to recognize the model if vison is added at the end.

Your own transfer station is still upstream. If you deploy it yourself, like oneapi or newapi, you can redirect the model name yourself.

@desktop4025
Copy link
Author

你把模型名称修改为带vision的就有了

可是我是用的中转站如果在结尾加上vison就识别不到这个模型了欸

自己的中转站还是上游的,如果是自己部署的,类似oneapi、newapi那种,可以自己重定向一下模型名称

上游的,用的gpt4的中转站,商家自己定义了一个叫gpt4-all的模型可以接受图片输入与4turbo一致

@Issues-translate-bot
Copy link

Bot detected the issue body's language is not English, translate it automatically.


You can change the model name to include vision.

But the transfer station I am using will not be able to recognize the model if vison is added at the end.

Your own transfer station is still upstream. If you deploy it yourself, like oneapi or newapi, you can redirect the model name yourself.

Upstream, the transfer station using gpt4 is used. The merchant defines a model called gpt4-all, which can accept image input consistent with 4turbo.

@QAbot-zh
Copy link

那就只好改代码了,修改isVisionModel那个判断函数

@Issues-translate-bot
Copy link

Bot detected the issue body's language is not English, translate it automatically.


Then we have to change the code and modify the isVisionModel judgment function

@desktop4025
Copy link
Author

那就只好改代码了,修改isVisionModel那个判断函数

好吧,或者可以提个enhancement?可以选择是否是visionmodel

@Issues-translate-bot
Copy link

Bot detected the issue body's language is not English, translate it automatically.


Then we have to change the code and modify the isVisionModel judgment function

Okay, maybe you can give me an enhancement? You can choose whether it is visionmodel

@Dean-YZG
Copy link
Contributor

那就只好改代码了,修改isVisionModel那个判断函数

好吧,或者可以提个enhancement?可以选择是否是visionmodel

是的,已经有计划会在最新版本支持用户自由度更高的模型配置,目前,您可以在您的代码中,先修改 isVisionModel ,将需要支持图片上传能力的模型加入进去

@Issues-translate-bot
Copy link

Bot detected the issue body's language is not English, translate it automatically.


Then we have to change the code and modify the isVisionModel judgment function

Okay, maybe you can give me an enhancement? You can choose whether it is visionmodel

Yes, there are plans to support model configuration with higher user freedom in the latest version. Currently, you can modify isVisionModel in your code first and add models that need to support image upload capabilities.

@robotJie
Copy link

@Dean-YZG 如果新版没那么快的话,考虑到LLM目前发展的速度,是否可以先将isVisionModel always return true?
不然每次最新的模型出来,都得hard code模型名,再发版,把事情弄复杂了好像。(毕竟各家LLM名字没有一套约束,一些判断逻辑总是滞后的)

@Issues-translate-bot
Copy link

Bot detected the issue body's language is not English, translate it automatically.


@Dean-YZG If the new version is not that fast, considering the current development speed of LLM, can isVisionModel always return true first?
Otherwise, every time the latest model comes out, we have to hard code the model name and then release it, which seems to complicate things. (After all, each LLM name does not have a set of constraints, and some judgment logic is always lagging behind)

@JsonSong89
Copy link

我觉得最简单的办法是直接再增加一个配置项
直接由用户指定哪些模型支持图片,一劳永逸

@Issues-translate-bot
Copy link

Bot detected the issue body's language is not English, translate it automatically.


I think the easiest way is to directly add another configuration item
Directly let the user specify which models support images, once and for all

@LovelyGuYiMeng
Copy link

马上就11月了,这个还是没更吗,其实很简单,给custommodel的env里面加个是否支持图片的参数就行了

@Issues-translate-bot
Copy link

Bot detected the issue body's language is not English, translate it automatically.


It’s almost November. Is this still not updated? It’s actually very simple. Just add a parameter to the custommodel’s env to determine whether it supports images.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
bug Something isn't working
Projects
None yet
Development

No branches or pull requests

7 participants