hermes - 💡(How to fix) Fix Feature Request: Per-platform model configuration [2 comments, 2 participants]

TL;DR

To achieve per-platform model configuration, update the GatewayConfig to include model and provider fields for each platform.

Guidance

• Modify the GatewayConfig class in gateway/config.py to include model and provider fields for each platform, allowing for per-platform configuration.
• Update the hermes_cli/model_switch.py (switch_model() function) to handle the new per-platform model configuration.
• Integrate the per-platform model configuration with the existing session override system in gateway/run.py.
• Verify the changes by testing the model configuration for each platform, ensuring that the desired models are used.

Example

class PlatformConfig:
    def __init__(self, model, provider):
        self.model = model
        self.provider = provider

class GatewayConfig:
    def __init__(self):
        self.platforms = {
            'feishu': PlatformConfig('mimo-v2.5', 'Token-plan-cn.xiaomimimo.com'),
            'cli': PlatformConfig('MiniMax-M2.7', 'minimax-cn')
        }

Notes

The proposed solution assumes that the existing session override system can be extended to accommodate per-platform model configuration. Further testing and validation may be required to ensure seamless integration.

Recommendation

Apply the workaround by updating the GatewayConfig to include per-platform model configuration, as this allows for more flexibility and customization in model selection for each platform.