Automatically detects the model type and applies the appropriate chat template.
For Gemma models, uses the Gemma-specific format. For other models,
falls back to the standard apply_chat_template function.
Usage
smart_chat_template(model, messages, template = NULL, add_assistant = TRUE)
Arguments
- model
A model object created with model_load
- messages
A list of message objects
- template
Custom template (passed to apply_chat_template if not Gemma)
- add_assistant
Whether to add assistant turn prefix
Value
Formatted chat template string