search

X-PLUG / MobileAgent

Mobile-Agent: The Powerful Mobile Device Operation Assistant Family
https://arxiv.org/abs/2406.01014
MIT License
2.96k stars 276 forks source link

Qwen API #32

Open shp216 opened 4 months ago

shp216 commented 4 months ago

Hello. Thank you for your great research. I want to use mobileagentv2, and I'm wondering if the qwen API is essential for using it. Since i'm an international, it seems impossible to obtain the qwen API. Could you let me know if this means I cannot use this model?

I know that version 1 could be used with just the GPT API, but I want to know if version 2 cannot be used without the qwen API.

junyangwang0410 commented 4 months ago

Hello. Thank you for your great research. I want to use mobileagentv2, and I'm wondering if the qwen API is essential for using it. Since i'm an international, it seems impossible to obtain the qwen API. Could you let me know if this means I cannot use this model?

I know that version 1 could be used with just the GPT API, but I want to know if version 2 cannot be used without the qwen API.

Hello.

Qwen API is not necessary because you can choose to deploy through local environment. Maybe the following can help you: https://github.com/X-PLUG/MobileAgent/tree/main/Mobile-Agent-v2#choose-the-appropriate-execution-method-for-your-needs

Choose the caption model : If you choose the "local" method, you need to choose between "qwen-vl-chat" and "qwen-vl-chat-int4", where the "qwen-vl-chat" requires more GPU memory but offers better performance compared to "qwen-vl-chat-int4". At the same time, "qwen_api" can be vacant.

shp216 commented 4 months ago

Could you please let me know the minimum GPU requirements for running the “Local” method? < If your device is not enough to run a 7B LLM, choose the "api" method. We use parallel calls to ensure efficiency. >

junyangwang0410 commented 4 months ago

Could you please let me know the minimum GPU requirements for running the “Local” method? < If your device is not enough to run a 7B LLM, choose the "api" method. We use parallel calls to ensure efficiency. >

12GB if you use qwen-int4