[Feature]: Enable Deployment of Different Models for Different Modules
hccngu opened this issue · comments
Is your feature request related to a problem? Please describe.
Currently, OS-Copilot does not support deploying different models for different modules.
Describe the solution you'd like
We hope OS-Copilot can support deploying different models for different modules, such as Planner, Generator, and Critic. Each module requires different capabilities; for example, some modules may be better suited to using the GPT API, while others might benefit from a fine-tuned LLAMA-3 model. This functionality should allow users to conveniently deploy different models for each module and switch between them with ease.
Describe alternatives you've considered
An alternative could be implementing a modular framework where each module's model can be manually configured and swapped, but this might not be as user-friendly or efficient as an integrated solution with one-click switching.
Additional context
None.